23
Research Article Recent Changes in the Annual Mean Regional Hadley Circulation and Their Impacts on South America Ana Carolina Vasques Freitas and Tércio Ambrizzi Institute of Astronomy, Geophysics and Atmospheric Sciences, 05508-090 S˜ ao Paulo, SP, Brazil Correspondence should be addressed to Ana Carolina Vasques Freitas; [email protected] Received 12 February 2015; Revised 16 April 2015; Accepted 23 April 2015 Academic Editor: Anthony R. Lupo Copyright © 2015 A. C. V. Freitas and T. Ambrizzi. is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. is work employs the regional climate model RegCM4 and observational datasets to investigate the impacts of changes in the intensity and poleward edge of regional HC over South America (HCSA) on the patterns of wind, geopotential height, precipitation, and temperature during the period 1996–2011. Significant trends of HCSA intensification and poleward expansion are found during the period analyzed. To evaluate the effects of these changes over SA, two composites, representing the intensification and poleward expansion cases, are examined separately. Significant correlations are seen between the temperature, zonal wind, and the HCSA intensity over the northern, central, and southern regions of SA and South Atlantic. Results show that, in both composites, regions with anomalous easterly (westerly) winds coming from (towards) the Atlantic Ocean have negative (positive) correlations with the HCSA intensity and poleward edge. e model performance varies regionally and the southern SA region exhibits better agreement with the observations. e role of the sea surface temperatures in driving the changes in the HCSA is examined. Notable similarity is found in the results for the two cases analyzed, which could indicate that, in most cases, the changes in the intensity and poleward edge of the HCSA are occurring simultaneously. 1. Introduction e differential heating between the tropics-extratropics results in the formation of a meridional circulation, the Hadley circulation, with warmer air rising in the ascending branch over equatorial areas and colder air sinking over the subtropics in both the Southern and Northern Hemispheres [1]. is circulation is one of the fundamental regulators of the earth’s energy budget, transporting heat poleward and, thus, reducing the resulting equator-to-pole temperature gra- dient. ere has been a recent and growing interest in studying the Hadley Cell (HC) changes in response to global climate change. Most studies have focused on the changes in HC strength and width, with the objective of verifying how these modifications affect the regional and global climates. Observations show that the HC has widened by about 2 –5 since 1979 and this observed widening cannot be explained by natural variability [2]. is widening and the concomitant poleward displacement of the subtropical dry zones may be accompanied by large-scale drying near 30 N and 30 S. As well, idealized and comprehensive global circulation models have shown a widening of the HC in response to increases in greenhouse gases and changes in the thermal structure of the polar stratosphere [3, 4]. Quan et al. [5] found consistent evidence for an intensi- fication of the Northern Hemisphere winter HC since 1950, as an atmospheric response to the observed tropical ocean warming trend, together with the intensification in El Ni˜ no’s interannual fluctuations, including larger amplitude and increased frequency aſter 1976. However, a similar trend of an intensified southward overturning HC is not detected during the Southern Hemisphere winter. e authors stated that this can indicate that such an ocean-atmosphere feedback process has been much weaker or has not even existed over the South Pacific Ocean during the southern winter since 1950. Most of the studies cited above are focused on the inves- tigation of the changes in the zonal mean HC, whereas the modifications in the regional HCs are still lesser known. As land, sea, and topography are not evenly distributed, changes Hindawi Publishing Corporation Advances in Meteorology Volume 2015, Article ID 780205, 22 pages http://dx.doi.org/10.1155/2015/780205

Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Research ArticleRecent Changes in the Annual Mean Regional HadleyCirculation and Their Impacts on South America

Ana Carolina Vasques Freitas and Teacutercio Ambrizzi

Institute of Astronomy Geophysics and Atmospheric Sciences 05508-090 Sao Paulo SP Brazil

Correspondence should be addressed to Ana Carolina Vasques Freitas anavasquesgmailcom

Received 12 February 2015 Revised 16 April 2015 Accepted 23 April 2015

Academic Editor Anthony R Lupo

Copyright copy 2015 A C V Freitas and T Ambrizzi This is an open access article distributed under the Creative CommonsAttribution License which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited

This work employs the regional climate model RegCM4 and observational datasets to investigate the impacts of changes in theintensity and poleward edge of regional HC over SouthAmerica (HCSA) on the patterns of wind geopotential height precipitationand temperature during the period 1996ndash2011 Significant trends of HCSA intensification and poleward expansion are found duringthe period analyzed To evaluate the effects of these changes over SA two composites representing the intensification and polewardexpansion cases are examined separately Significant correlations are seen between the temperature zonal wind and the HCSAintensity over the northern central and southern regions of SA and South Atlantic Results show that in both composites regionswith anomalous easterly (westerly) winds coming from (towards) the Atlantic Ocean have negative (positive) correlations with theHCSA intensity and poleward edgeThemodel performance varies regionally and the southern SA region exhibits better agreementwith the observations The role of the sea surface temperatures in driving the changes in the HCSA is examined Notable similarityis found in the results for the two cases analyzed which could indicate that in most cases the changes in the intensity and polewardedge of the HCSA are occurring simultaneously

1 Introduction

The differential heating between the tropics-extratropicsresults in the formation of a meridional circulation theHadley circulation with warmer air rising in the ascendingbranch over equatorial areas and colder air sinking over thesubtropics in both the Southern and Northern Hemispheres[1] This circulation is one of the fundamental regulators ofthe earthrsquos energy budget transporting heat poleward andthus reducing the resulting equator-to-pole temperature gra-dient

There has been a recent and growing interest in studyingthe Hadley Cell (HC) changes in response to global climatechange Most studies have focused on the changes in HCstrength and width with the objective of verifying howthese modifications affect the regional and global climatesObservations show that the HC has widened by about 2∘ndash5∘since 1979 and this observed widening cannot be explainedby natural variability [2] This widening and the concomitantpoleward displacement of the subtropical dry zones may be

accompanied by large-scale drying near 30∘N and 30∘S Aswell idealized and comprehensive global circulation modelshave shown a widening of the HC in response to increases ingreenhouse gases and changes in the thermal structure of thepolar stratosphere [3 4]

Quan et al [5] found consistent evidence for an intensi-fication of the Northern Hemisphere winter HC since 1950as an atmospheric response to the observed tropical oceanwarming trend together with the intensification in El Ninorsquosinterannual fluctuations including larger amplitude andincreased frequency after 1976 However a similar trend of anintensified southward overturning HC is not detected duringthe Southern Hemisphere winterThe authors stated that thiscan indicate that such an ocean-atmosphere feedback processhas been much weaker or has not even existed over the SouthPacific Ocean during the southern winter since 1950

Most of the studies cited above are focused on the inves-tigation of the changes in the zonal mean HC whereas themodifications in the regional HCs are still lesser known Asland sea and topography are not evenly distributed changes

Hindawi Publishing CorporationAdvances in MeteorologyVolume 2015 Article ID 780205 22 pageshttpdxdoiorg1011552015780205

2 Advances in Meteorology

in theHC should be different in various regions [6] Althoughthe HC is referred to in some studies as a global symmetriczonally meridional circulation [7] it is possible to analyze itsimpact and variability in a regional level However it is goodto highlight that in this case the atmospheric circulationsmay not be closed cells in particular if we also consider therotational component of wind [8]

Some studies have demonstrated that alterations inregional HCs could have an important influence on theanomalies of the regional climates Ambrizzi et al [9]analyzed the influence of the El Nino-Southern Oscillation(ENSO) events on the regional Hadley and Walker cells andtheir respective impacts on the South American seasonalrainfall As not all ENSO events follow a canonical patternthe authors demonstrated that depending on the conditionsof the Atlantic Sea Surface Temperature (SST) anomalies theposition of the Intertropical Convergence Zone (ITCZ) canbe modified the ascending and descending branches of therelated regional HC may vary and the South American rainyseason may be strongly affected Chen et al [6] investigatedlong-term trends in the intensity and poleward edge of theregional HCs in six selected regions using six reanalysesOutgoing Longwave Radiation (OLR) and precipitationdatasets In the Southern Hemisphere (SH) the authorsfound a significant poleward expansion trend of the HClocalized over the South America with significant impact onprecipitation anomalies in this region The results found bythe authors also indicated that the poleward expansion ofthe zonal mean HC identified by several studies could beattributed mainly to the poleward expansion of the regionalHC over South America Zeng et al [10] demonstrated thatchanges in theHC intensity in the western and eastern Pacificare associated with anomalous East Asian winter monsoonThus changes in regional HCs are of great importance for theregional climate variability [6]

In the present study composite analyses are performed toexamine the impact of the changes in intensity and polewardedge of the regional HC over South America on the patternsof wind geopotential height precipitation and temperatureduring the 1996ndash2011 period This is addressed using mainlythe regional climate model version 4 (RegCM4) [11] and theEra-Interim reanalysis [12] of European Centre for MediumRange Forecasting (ECMWF)

The studies about regional HCs cited above used obser-vational datasets Thus as far as we know little attentionhas been paid to the investigation of the impacts of changesin regional HCs using regional climate modelling It isimportant to assess the performance of the regional climatemodels in simulating local anomalies associatedwith changesin regional HCs Local changes in temperature and precipita-tion could be quite different from the coarse-scale changesprojected by global models Thus this assessment may provevery useful in the search to understand the future of thesecirculations in a changing climate and the local anomaliesassociated

Details of the model and simulation are given in theSections 21 and 22 In Section 23 we describe our method-ology and show how we set up the composites that representchanges in the intensity and poleward edge of regional

HC over South America We investigate the impacts ofthe regional HC intensification in Section 3 whereas inSection 4 the impacts of the regionalHCpoleward expansionare examined Finally conclusions highlighting the mainresults of the paper are given in Section 5

2 Methodology

21 RegCM4 Details Giorgi et al [11] described the RegCM4model details so we give only a brief overview here RegCM4is an evolution of its previous version RegCM3 described byPal et al [13] Thus it is a hydrostatic compressible sigma-p vertical coordinate model run on an Arakawa B-grid inwhich wind and thermodynamical variables are horizontallystaggered A time-splitting explicit integration scheme is usedinwhich the two fastest gravitymodes are first separated fromthe model solution and then integrated with smaller timesteps This allows the use of a longer time step for the restof the model

Radiative transfer calculations in RegCM4 are carriedout with the radiative transfer scheme of the global modelCCM3 [14]This includes calculations for the short-wave andinfrared parts of the spectrum including both atmosphericgases and aerosols The scheme includes contributions fromall main greenhouse gases that is H

2O CO

2 O3 CH4 N2O

and CFCs and solar radiative processes are treated using adelta-Eddington formulation Scattering and absorption ofsolar radiation by aerosols are also included based on theaerosol optical properties (absorption coefficient and singlescattering albedo)

Land surface processes are described via the Biosphere-Atmosphere Transfer Scheme (BATS) of Dickinson et al [15]This scheme which has been used for many years includesa 1-layer vegetation module a 1-layer snow module a force-restore model for soil temperatures a 3-layer soil schemeand a simple surface runoff parameterization The schemeincludes twenty surface types and twelve soil color and soiltexture types

RegCM4 includes three options for representing cumulusconvection The first is a simplified version of the Kuo-typescheme of Anthes [16] as described by Anthes et al [17] thesecond is that of Grell [18] in the implementation of Giorgiet al [19] and the third which is the so-calledMassachusettsInstitute of Technology (MIT) scheme [20] is introduced inthe RegCM3 by Pal et al [13]

22 Simulation Details and Observed Data The daily datawith 6-hourly temporal resolution used to run the model areobtained from the RegCM4 data website (httpusersictpitsimpubregcmRegCM4globedathtm)Themodel version usedhere is the 4357 The lateral meteorological boundary con-ditions are provided by theEra-Interim reanalysis of ECMWF[12] The Weekly Optimum Interpolation Sea Surface Tem-perature (OI WK) version 2 [21] of National Oceanic andAtmospheric Administration (NOAA) which is availableweekly on a 10∘times 10∘ grid is used to force the model Table 1shows the technical details of the simulation performed

Advances in Meteorology 3

Table 1 Technical details of the simulation performed

Simulation parametersDomain cartographic projection(iproj)

Rotated Mercator(ROTMER)

Grid point horizontal resolution inkm (ds) 600 km

Number of points in the NSdirection (iy) 152

Number of points in the EWdirection (jx) 192

Number of vertical levels (kz) 18Central latitude of model domain indegrees (clat) minus2300

Central longitude of model domainin degrees (clon) minus3300

Cumulus convection scheme (icup) 1-Kuo

Moisture scheme (ipptls) 1-Explicit moisture(SUBEX Pal et al [42])

Lateral boundary conditionsscheme (iboudy)

5-Relaxationexponential technique

Boundary layer scheme (ibltyp) 1-Holtslag PBL [43]

Ocean flux scheme (iocnflx) 2-Zeng et al [44]

including the simulation domain The model is run for theperiod 1995ndash2011 The first year is discarded to avoid prob-lems related to the spin-up

The Kuo-type cumulus parameterization is used here Inthis scheme convection is triggered in a convectively unstablelow troposphere when the column moisture convergenceexceeds a threshold value [11] Some studies found thatthe Kuo scheme produces satisfactory seasonal and annualprecipitation [22 23] We will show that this scheme couldreasonably reproduce the precipitation anomalies especiallyin the southern region associated with the modifications inthe HC over South America

The Era-Interim monthly data the Global PrecipitationClimatology Centre dataset (GPCC) (GPCC dataset is avail-able on httpwwwesrlnoaagovpsddatagriddeddatagpcchtml) [24] and the SST monthly mean data of the HadleyCentre Global Sea Ice and Sea Surface Temperature (HAD-ISST) (HADISST dataset is available on httpwwwmetofficegovukhadobshadisstdatadownloadhtml) [25] are used forvalidation of the model results

23 Methods Horizontal wind field can be expressed bythe sum of a nondivergent (or rotational) component and adivergent (or irrotational) component [26 27]

V = V120595+ V120594=

119896 times nabla120595 + nabla120594 (1)

where 120595 is stream function and 120594 is velocity potential Thedivergent component is fundamental to study the atmo-spheric divergence-convergence that drives the verticalmotion and circulation in the tropics [28]

The methodology used here to define the intensityindex and poleward edge of the regional HC is similar toChen et al [6] The regional HC intensity is defined throughan index based on the vertical shear of divergent meridionalwinds at 200 hPa and 850 hPa whereas the poleward edge ofthe regional HC is identified as the latitude where the OLRis equal to 240Wmminus2 (found through linear interpolation)The OLR data is obtained from the Climate DiagnosticsCenter of the National Oceanic and Atmospheric Admin-istration (NOAA-CDC) (OLRdata is available on httpwwwesrlnoaagovpsddatagriddeddatainterp OLRhtml) in a25∘times 25∘ grid [29] Annual means are constructed by aver-aging the monthly data from January to December

Initially two domains are selected for the regional HCsthe SouthAmerica (80∘Wndash45∘W)andAtlanticOcean (38∘Wndash10∘E) regionsThe regionalHCs in these locations can have animportant influence on synoptic weather and climate systems[6] The annual mean of the intensity index for the regionalHC over South America (hereafter denoted as HCSA) isdefined as the difference in the area-averaged (80∘Wndash45∘W20∘Sndash0) divergent meridional winds between 200 hPa and850 hPa This latitudinal band is chosen according to theclimatological distributions of divergent winds (Figure 1(a))Definition of the intensity index for the regional HC overAtlantic Ocean (hereafter denoted HCAO) is similar to thoseof the HCSA except that the divergent meridional winds areaveraged over 38∘Wndash10∘E 5∘Nndash20∘S (Figure 1(b)) Chen et al[6] conducted a series of tests and found that the result doesnot change much if other latitudinal bands are used

The normalized time series for the annual mean HCSAintensity index (Figure 2(a)) presents a significant negativetrend (minus012 yearminus1) This indicates an intensification of theHCSA as the HC in SH is negative corresponding to acounterclockwise circulationThus cases of strongHCSA areselected based on the years that present HCSA normalizedintensity index equal to or less than minus1 in the period 1996ndash2011 The HCSA strong cases are 2005 2008 2010 and 2011Figure 3(a) shows the HCSA anomaly for these strong casesIt is possible to note an intense ascending branch from 0 to5∘N and a descending branch around 30∘S

It is worth highlighting that we are using the Era-Interimreanalysis data to calculate the intensity indexThis dataset ischosen since it is the latest global atmospheric reanalysis pro-duced by ECMWF and thus incorporates many importantmodel improvements such as resolution and physics changesthe use of four-dimensional variational data assimilation andvarious other changes in the analysis methodology [30] Sev-eral studies found that Era-Interim has the best performanceamong reanalysis datasets [31 32] Also some studies foundthat the performance skills of the RegCM driven by Era-Interim are better than by other reanalyses [33 34]

Asmentioned earlier the HCSA poleward edge is definedby the latitude where the OLR annual mean is equal to240Wmminus2 Slight deviations from this OLR value such as236 or 245Wmminus2 lead to similar results Thus a linearinterpolation of the OLR annual mean values averaged over80∘Wndash45∘Wwas performed to obtain the latitude that defines

4 Advances in Meteorology

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

EQLatitude

2

45∘S 40

∘S 35∘S 30

∘S 25∘S 20

∘S 15∘S 10

∘S 5∘S 5

∘N 10∘N

(a)

2

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(b)

Figure 1 Annual mean of the regional HC in the period 1996ndash2011 over (a) South America averaged between 80∘W and 45∘W (HCSA) and(b) Atlantic Ocean averaged between 38∘W and 10∘E (HCAO) Vertical circulations are described by the vertical velocities (Pa sminus1) and thedivergent meridional winds (m sminus1) based on the Era-Interim reanalysis The vertical velocity has been multiplied by minus100 The color scalerepresents the vectorrsquos magnitude

minus3

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

HCS

A in

tens

ity in

dex

Year

(a)

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

Year

HCS

A p

olew

ard

edge

(b)

Figure 2 Normalized time series for the annual mean HCSA (a) intensity index and (b) poleward edge in the period 1996ndash2011 with thelinear trend (red dotted line)

the HCSA poleward edge The normalized time series of theannual mean HCSA poleward edge (Figure 2(b)) presents asignificant negative trend (minus013 degrees of latitude yearminus1)which indicates a poleward expansion of the HCSA Thuscases of HCSA poleward expansion are selected based on theyears that present HCSA normalized poleward edge equal toor less than minus1 in the period 1996ndash2011 The HCSA polewardexpansion cases are 1996 2008 2010 and 2011 Interestinglythese three last years are also part of the HCSA strong casesFigure 3(b) shows the HCSA anomaly for these polewardexpansion cases It is possible to note an ascending branchfrom 0 to 5∘N and a descending branch around 35∘S Hereas expected the descending branch is stronger and is shiftedtowards the pole (Figure 3(b)) compared to the HCSA strongcases (Figure 3(a)) However the ascending branch is weaker(Figure 3(b)) than for the HCSA strong cases (Figure 3(a))Therefore an intense upward motion which induces more

divergence in the upper troposphere distinguishes a HCSAstrong case whereas a HCSA poleward expansion case ischaracterized by a strong and poleward shifted descendingbranch

The normalized time series for the annual mean HCAOintensity index also shows a significant negative trend (minus013yearminus1) (Figure 4(a)) but only the years 2008 and 2010presented HCAO normalized intensity index equal to or lessthan minus1 in the period 1996ndash2011 Moreover the observedtrend for the annual mean HCAO poleward edge normalizedtime series (minus011 yearminus1 119901 value = 004) (Figure 4(b)) is notas significant as theHCSA poleward edge trend (minus013 yearminus1119901 value = 001) Therefore we will focus on the impacts ofthe HCSA intensification and poleward expansion on thehorizontal wind precipitation temperature and geopotentialheight fields over South America However this investigationalso encompasses the Ocean Atlantic region

Advances in Meteorology 5

05

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(a)

05

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(b)

Figure 3 Annual mean of the HCSA anomaly based on the Era-Interim reanalysis for (a) strong cases and (b) poleward expansion casesThe vertical velocity has been multiplied by minus100 The color scale represents the vectorrsquos magnitude

minus3

minus2

minus1

1996 1998 2000 2002 2004 2006 2008 2010

HCA

O in

tens

ity in

dex

Year

0

1

2

(a)

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

Year

HCA

O p

olew

ard

edge

(b)

Figure 4 Normalized time series for the annual mean HCAO (a) intensity index and (b) poleward edge in the period 1996ndash2011 with thelinear trend (red dotted line)

The performance of the regional climate models variesgreatly depending on the combination of physical parame-terization schemes simulation region boundary conditionsand the horizontal resolution [18 20] The skill of theRegCM4 over different regions covering the South Americaand Atlantic Ocean (Figure 5) during the 1996ndash2011 periodis evaluated quantitatively by analyzing the root mean squareerror (RMSE) bias and spatial correlation coefficient of thesimulation in relation to observations for the annual meananomalies We selected the regions in which the impactsof the HCSA intensification and poleward expansion aremore prominent The RMSE measures the overall magnitudeof deviations regardless of their individual signs while thebias measures systematic or predominant deviations in givendirections [35] The closer the values of RMSE and biasget to zero the better the agreement between observed andsimulated data is The RMSE is calculated according to (2)where 1199091015840119872 and 1199091015840119874 are the anomalies of the model and theobservations respectivelyThe summation is carried out over

119873 grid points in a predefined regionThe bias and the spatialcorrelation coefficient are calculated according to (3) and (4)respectively Consider

RMSE = radic 1119873

119873

sum

119894=1

(119909

1015840119872

119894

minus 119909

1015840119874

119894

)

2 (2)

bias = 1119873

119873

sum

119894=1

(119909

1015840119872

119894

minus 119909

1015840119874

119894

) (3)

120588 =

sum

119873

119894=1

(119909

1015840119872

119894

minus 119909

1015840119872

) (119909

1015840119874

119894

minus 119909

1015840119874

)

radic

sum

119873

119894=1

(119909

1015840119872

119894

minus 119909

1015840119872

)

2

radicsum

119873

119894=1

(119909

1015840119874

119894

minus 119909

1015840119874

)

2

(4)

3 Impacts of the HCSA Intensification

The vertical velocity and the divergent winds are the mainfields analyzed when the regional HC is studied However

6 Advances in Meteorology

Table 2 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSAstrong cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 042 031 152 026 012 012 052 minus010 078 035 minus015 0112 091 018 329 023 minus067 002 318 004 068 050 087 0403 051 019 167 011 039 013 155 006 083 088 098 0854 065 025 313 019 minus011 011 225 minus001 084 073 091 055

Latit

ude

Longitude

1

2

3

4

5

6

7

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

Figure 5 South America regions selected for statistical evaluationof model performance (1) northern (80∘Wndash35∘W 15∘Sndash10∘N) (2)central (80∘Wndash35∘W 35∘Sndash15∘S) (3) southern (80∘Wndash55∘W 55∘Sndash35∘S) (4) entire domain (80∘Wndash10∘W 55∘Sndash10∘N) (5) southeast(70∘Wndash50∘W 40∘Sndash30∘S) (6) northeast (60∘Wndash35∘W 10∘Sndash5∘N)and (7) all South America (80∘Wndash35∘W 55∘Sndash10∘S)

there is a remarkable impact of the changes in the HC onthe horizontal winds at low and high troposphere as they arelinked to the vertical movement of the air Indeed the tradewinds are the expression of the HC at surface

Figure 6 shows the anomalies of the annual mean hor-izontal wind at 200 and 850 hPa for the HCSA strongcases based on the Era-Interim reanalysis and the modelsimulation Strong anomalous easterly winds coming fromthe Atlantic Ocean are seen over the northern South America(SA) at 200 hPa in the reanalysis (Figure 6(a)) and especiallyin model simulation (Figure 6(c)) which exhibits a positivebias (012) for the zonal wind in this region (Table 2) Whenentering the mainland these winds rotate counterclockwiseStrong anomalous easterly winds coming from the Atlanticalso hits the southeast of the continent and then they rotate

clockwise resulting in an anomalous trough which is seenat 200 hPa (Figure not shown) and 500 hPa (Figure 7) Theanomalous wind pattern at 850 hPa is characterized by thestrengthening of the trade winds around the equator whichspin in northeast direction before entering the northern SAand by a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 6(b)) and model simu-lation (Figure 6(d))This counterclockwise circulation resultsin a significant anomalous ridge which is seen at 850 hPa(Figure not shown) and 500 hPa in the reanalysis and modelsimulation (Figure 7)

To investigate the quantitative link between the HCSAintensity and the annual changes in the zonal wind (strongercomponent of the horizontal wind) correlation coefficientsbetween the HCSA annual intensity index and the annualmean zonal wind at 200 and 850 hPa are displayed in Figure 8The model simulation (Figure 8(c)) is able to reproduce thesigns (positives and negatives) of the correlation coefficientsobserved in the reanalysis (Figure 8(a)) Three regions withsignificant correlations (above 05) in the high troposphereare seen in the reanalysis (Figure 8(a)) Figures 6(a) and 6(c)showed that for HCSA strong cases the zonal component ofthe anomalous wind at 200 hPa is easterly (negative) in thenorthern and southern SA regions whereas in the centralarea of SA it is westerly (positive) In anotherway Figures 8(a)and 8(c) show that the anomalous easterly (westerly) windscoming from (towards) the Atlantic presents negative (pos-itive) correlations with the HCSA intensity At 850 hPasignificant positive correlations are seen in the reanalysison the edge of northwest SA and negative correlations onthe central-south area of the continent (Figure 8(b)) Themodel simulation shows significant positive correlations ona great part of the northwest SA (Figure 8(d)) probably dueto the misrepresentation of the horizontal wind in this region(Figure 6(d)) At 850 hPa it can also be seen that regions withanomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA intensity

The three regions (northern central and southern) withsignificant correlations seen in Figure 8(a) and the entiredomain considered in the composite analyses were selectedfor evaluation of model performance in reproducing theanomalies of zonal wind geopotential height and temper-ature (Figure 5) Table 2 shows that for the annual meanzonal wind anomalies at 200 hPa lower values of RMSE andbias (absolute value) and higher values of spatial correlation

Advances in Meteorology 7La

titud

e

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 6 Annual mean horizontal wind (m sminus1) anomaly for HCSA strong cases in the period 1996ndash2011 based on the Era-Interim reanalysisat (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents the vectorrsquos magnitude

8 Advances in MeteorologyLa

titud

e

Longitude

18

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

0

0

0

0

minus4

minus4

minus4

minus4

minus2

minus2

minus2

minus2

2

4

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus2

minus2

0

0

0

2

2

2

2

1820

1614

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 7 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

are found mainly in the northern and southern SA andwhen we consider the entire domain (region 4 of Figure 5)Thus at 200 hPa the central region of SA exhibits weakeragreement between the model and observation for the zonalwind anomalies Table 2 shows that themodel underestimatesthe zonal wind anomalies in this region (bias = minus067) Aswell Figures 6(c) and 7(b) show that the model simulatesa strong anticyclonic anomalous circulation in the central-western region Interestingly at 850 hPa the central region ofSA exhibits the lowest value of RMSE and bias (018 and 002resp) but the spatial correlation is only 05 Mid- and upper-tropospheric winds influence low-level circulations thusthe biases in low and high levels are somewhat connectedIncreasing resolution can improve the results in the centralregion of SA due to better representation of orography in thisregion in a high-resolution simulation The highest spatialcorrelation for the zonal wind anomalies at 850 hPa is foundin the southern SA (088) The geopotential height is thevariable that exhibits the highest values of RMSE and bias(mostly greater than 1) for all regions (Table 2) The lowervalues of RMSE and bias are found in the northern andsouthern SA but the spatial correlation is weaker in thenorthern and stronger in the southern

Figure 9 shows the annual mean temperature anomaliesfor HCSA strong cases The reanalysis data presents signif-icant positive anomalies on a great part of northern SA andequatorial Atlantic whereas the negative anomalies are foundover the east Pacific around 20∘S (Figure 9(a)) The model

shows significant positive anomalies only over the northeastedge of SA and over the equatorial Atlantic (Figure 9(b))Indeed themodel underestimates the temperature anomaliesin the northern region as indicated by the bias value in thisregion (minus011 Table 2) A positive (negative) bias in surfaceair temperature is associated with a negative (positive) biasin precipitation as an underestimation (overestimation) ofprecipitation is usually associated with an underestimation(overestimation) of clouds which enhances (reduces) thenet short-wave radiation budget at surface and consequentlyenhances (reduces) the net energy budget at surface [36]Indeed the model exhibits a positive bias (019) for theprecipitation anomalies in the northern region The lowest(highest) value of the RMSE (correlation) for the annualmean temperature anomalies is found in the southern regionindicating a good agreement between the model and obser-vation

Figure 10 displays the correlation between the HCSAannual intensity index and the annual mean temperaturein the period 1996ndash2011 As in Figure 8(a) three signifi-cant regions (northern central and southern SA) are seenHowever the signs of the correlation coefficients in thesethree regions of Figure 10 are opposite in relation to Figure 8Thus positive (negative) correlations of the annual meantemperature (zonal wind) with the HCSA intensity are seenover the northern and southern SA in the 1996ndash2011 periodThe opposite signal between the annual mean temperatureand the zonal wind is also seen over the central regions of

Advances in Meteorology 9

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus06

minus06

minus06

minus04

minus04

minus04

minus04

minus02

minus02

minus020

0

0

02

02

02

02

04

04

0404

06

06

08

minus020

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

02

02 02

02

02

02

02

04

04

04

06

08

0

0

0

0

0

0

0

0

0

Latit

ude

Longitude

minus06

minus06

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02minus02

minus02

minus02

minus02

04

04

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

minus02

minus02

minus02

minus02

minus02

minus08

minus08

minus04

minus04

minus04

minus04

minus06

minus06

minus06

0202

02

02

040

0

0

0

0

Latit

ude

Longitude

08

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04minus04

minus04

minus04

minus04

minus06

02

02

02

02

02

02

0204

04

04

04

04

04

06 06

08

08

08

0

00

0

0

0

0

Latit

ude

Longitude

06

minus02

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 8 Correlation between the HCSA annual intensity index and the annual mean zonal wind (m sminus1) in the period 1996ndash2011 based onthe Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervals are 02and values above 05 are shaded and are statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSAannual intensity index time series have been multiplied by (minus1)

10 Advances in Meteorology

02

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus05

minus04

minus03

minus03

minus03

minus02

minus02

minus01

minus01

minus01

minus01

0

0

0

0

01

01

01

01

02

02

02

02

02

02

02

02

02

03

03

03

03

03

04

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

minus02

minus02

minus02

minus02

minus02

minus01

minus01

minus01

minus01

minus01

minus01

minus01

00

0

0

0

0

0

01

01

01

01

01

02

02

02

02

02

03

03

04

05

03

03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 9 Annual mean temperature (K) anomaly at 850 hPa for HCSA strong cases in the period 1996ndash2011 based on the (a) Era-Interimreanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level from two-tailed Studentrsquos119905-test are shaded

minus06minus06

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

02

02

02

02

02

04

04

04

04

04

06

06

06

06

06

Latit

ude

Longitude

0

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

04

04

04

02

02

02

02

04

04

06

0606

08

08

00

0

0

0

0

Latit

ude

Longitude

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 10 Correlation between the HCSA annual intensity index and the annual mean temperature (K) at 850 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSA annual intensity index has beenmultiplied by (minus1)

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 2: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

2 Advances in Meteorology

in theHC should be different in various regions [6] Althoughthe HC is referred to in some studies as a global symmetriczonally meridional circulation [7] it is possible to analyze itsimpact and variability in a regional level However it is goodto highlight that in this case the atmospheric circulationsmay not be closed cells in particular if we also consider therotational component of wind [8]

Some studies have demonstrated that alterations inregional HCs could have an important influence on theanomalies of the regional climates Ambrizzi et al [9]analyzed the influence of the El Nino-Southern Oscillation(ENSO) events on the regional Hadley and Walker cells andtheir respective impacts on the South American seasonalrainfall As not all ENSO events follow a canonical patternthe authors demonstrated that depending on the conditionsof the Atlantic Sea Surface Temperature (SST) anomalies theposition of the Intertropical Convergence Zone (ITCZ) canbe modified the ascending and descending branches of therelated regional HC may vary and the South American rainyseason may be strongly affected Chen et al [6] investigatedlong-term trends in the intensity and poleward edge of theregional HCs in six selected regions using six reanalysesOutgoing Longwave Radiation (OLR) and precipitationdatasets In the Southern Hemisphere (SH) the authorsfound a significant poleward expansion trend of the HClocalized over the South America with significant impact onprecipitation anomalies in this region The results found bythe authors also indicated that the poleward expansion ofthe zonal mean HC identified by several studies could beattributed mainly to the poleward expansion of the regionalHC over South America Zeng et al [10] demonstrated thatchanges in theHC intensity in the western and eastern Pacificare associated with anomalous East Asian winter monsoonThus changes in regional HCs are of great importance for theregional climate variability [6]

In the present study composite analyses are performed toexamine the impact of the changes in intensity and polewardedge of the regional HC over South America on the patternsof wind geopotential height precipitation and temperatureduring the 1996ndash2011 period This is addressed using mainlythe regional climate model version 4 (RegCM4) [11] and theEra-Interim reanalysis [12] of European Centre for MediumRange Forecasting (ECMWF)

The studies about regional HCs cited above used obser-vational datasets Thus as far as we know little attentionhas been paid to the investigation of the impacts of changesin regional HCs using regional climate modelling It isimportant to assess the performance of the regional climatemodels in simulating local anomalies associatedwith changesin regional HCs Local changes in temperature and precipita-tion could be quite different from the coarse-scale changesprojected by global models Thus this assessment may provevery useful in the search to understand the future of thesecirculations in a changing climate and the local anomaliesassociated

Details of the model and simulation are given in theSections 21 and 22 In Section 23 we describe our method-ology and show how we set up the composites that representchanges in the intensity and poleward edge of regional

HC over South America We investigate the impacts ofthe regional HC intensification in Section 3 whereas inSection 4 the impacts of the regionalHCpoleward expansionare examined Finally conclusions highlighting the mainresults of the paper are given in Section 5

2 Methodology

21 RegCM4 Details Giorgi et al [11] described the RegCM4model details so we give only a brief overview here RegCM4is an evolution of its previous version RegCM3 described byPal et al [13] Thus it is a hydrostatic compressible sigma-p vertical coordinate model run on an Arakawa B-grid inwhich wind and thermodynamical variables are horizontallystaggered A time-splitting explicit integration scheme is usedinwhich the two fastest gravitymodes are first separated fromthe model solution and then integrated with smaller timesteps This allows the use of a longer time step for the restof the model

Radiative transfer calculations in RegCM4 are carriedout with the radiative transfer scheme of the global modelCCM3 [14]This includes calculations for the short-wave andinfrared parts of the spectrum including both atmosphericgases and aerosols The scheme includes contributions fromall main greenhouse gases that is H

2O CO

2 O3 CH4 N2O

and CFCs and solar radiative processes are treated using adelta-Eddington formulation Scattering and absorption ofsolar radiation by aerosols are also included based on theaerosol optical properties (absorption coefficient and singlescattering albedo)

Land surface processes are described via the Biosphere-Atmosphere Transfer Scheme (BATS) of Dickinson et al [15]This scheme which has been used for many years includesa 1-layer vegetation module a 1-layer snow module a force-restore model for soil temperatures a 3-layer soil schemeand a simple surface runoff parameterization The schemeincludes twenty surface types and twelve soil color and soiltexture types

RegCM4 includes three options for representing cumulusconvection The first is a simplified version of the Kuo-typescheme of Anthes [16] as described by Anthes et al [17] thesecond is that of Grell [18] in the implementation of Giorgiet al [19] and the third which is the so-calledMassachusettsInstitute of Technology (MIT) scheme [20] is introduced inthe RegCM3 by Pal et al [13]

22 Simulation Details and Observed Data The daily datawith 6-hourly temporal resolution used to run the model areobtained from the RegCM4 data website (httpusersictpitsimpubregcmRegCM4globedathtm)Themodel version usedhere is the 4357 The lateral meteorological boundary con-ditions are provided by theEra-Interim reanalysis of ECMWF[12] The Weekly Optimum Interpolation Sea Surface Tem-perature (OI WK) version 2 [21] of National Oceanic andAtmospheric Administration (NOAA) which is availableweekly on a 10∘times 10∘ grid is used to force the model Table 1shows the technical details of the simulation performed

Advances in Meteorology 3

Table 1 Technical details of the simulation performed

Simulation parametersDomain cartographic projection(iproj)

Rotated Mercator(ROTMER)

Grid point horizontal resolution inkm (ds) 600 km

Number of points in the NSdirection (iy) 152

Number of points in the EWdirection (jx) 192

Number of vertical levels (kz) 18Central latitude of model domain indegrees (clat) minus2300

Central longitude of model domainin degrees (clon) minus3300

Cumulus convection scheme (icup) 1-Kuo

Moisture scheme (ipptls) 1-Explicit moisture(SUBEX Pal et al [42])

Lateral boundary conditionsscheme (iboudy)

5-Relaxationexponential technique

Boundary layer scheme (ibltyp) 1-Holtslag PBL [43]

Ocean flux scheme (iocnflx) 2-Zeng et al [44]

including the simulation domain The model is run for theperiod 1995ndash2011 The first year is discarded to avoid prob-lems related to the spin-up

The Kuo-type cumulus parameterization is used here Inthis scheme convection is triggered in a convectively unstablelow troposphere when the column moisture convergenceexceeds a threshold value [11] Some studies found thatthe Kuo scheme produces satisfactory seasonal and annualprecipitation [22 23] We will show that this scheme couldreasonably reproduce the precipitation anomalies especiallyin the southern region associated with the modifications inthe HC over South America

The Era-Interim monthly data the Global PrecipitationClimatology Centre dataset (GPCC) (GPCC dataset is avail-able on httpwwwesrlnoaagovpsddatagriddeddatagpcchtml) [24] and the SST monthly mean data of the HadleyCentre Global Sea Ice and Sea Surface Temperature (HAD-ISST) (HADISST dataset is available on httpwwwmetofficegovukhadobshadisstdatadownloadhtml) [25] are used forvalidation of the model results

23 Methods Horizontal wind field can be expressed bythe sum of a nondivergent (or rotational) component and adivergent (or irrotational) component [26 27]

V = V120595+ V120594=

119896 times nabla120595 + nabla120594 (1)

where 120595 is stream function and 120594 is velocity potential Thedivergent component is fundamental to study the atmo-spheric divergence-convergence that drives the verticalmotion and circulation in the tropics [28]

The methodology used here to define the intensityindex and poleward edge of the regional HC is similar toChen et al [6] The regional HC intensity is defined throughan index based on the vertical shear of divergent meridionalwinds at 200 hPa and 850 hPa whereas the poleward edge ofthe regional HC is identified as the latitude where the OLRis equal to 240Wmminus2 (found through linear interpolation)The OLR data is obtained from the Climate DiagnosticsCenter of the National Oceanic and Atmospheric Admin-istration (NOAA-CDC) (OLRdata is available on httpwwwesrlnoaagovpsddatagriddeddatainterp OLRhtml) in a25∘times 25∘ grid [29] Annual means are constructed by aver-aging the monthly data from January to December

Initially two domains are selected for the regional HCsthe SouthAmerica (80∘Wndash45∘W)andAtlanticOcean (38∘Wndash10∘E) regionsThe regionalHCs in these locations can have animportant influence on synoptic weather and climate systems[6] The annual mean of the intensity index for the regionalHC over South America (hereafter denoted as HCSA) isdefined as the difference in the area-averaged (80∘Wndash45∘W20∘Sndash0) divergent meridional winds between 200 hPa and850 hPa This latitudinal band is chosen according to theclimatological distributions of divergent winds (Figure 1(a))Definition of the intensity index for the regional HC overAtlantic Ocean (hereafter denoted HCAO) is similar to thoseof the HCSA except that the divergent meridional winds areaveraged over 38∘Wndash10∘E 5∘Nndash20∘S (Figure 1(b)) Chen et al[6] conducted a series of tests and found that the result doesnot change much if other latitudinal bands are used

The normalized time series for the annual mean HCSAintensity index (Figure 2(a)) presents a significant negativetrend (minus012 yearminus1) This indicates an intensification of theHCSA as the HC in SH is negative corresponding to acounterclockwise circulationThus cases of strongHCSA areselected based on the years that present HCSA normalizedintensity index equal to or less than minus1 in the period 1996ndash2011 The HCSA strong cases are 2005 2008 2010 and 2011Figure 3(a) shows the HCSA anomaly for these strong casesIt is possible to note an intense ascending branch from 0 to5∘N and a descending branch around 30∘S

It is worth highlighting that we are using the Era-Interimreanalysis data to calculate the intensity indexThis dataset ischosen since it is the latest global atmospheric reanalysis pro-duced by ECMWF and thus incorporates many importantmodel improvements such as resolution and physics changesthe use of four-dimensional variational data assimilation andvarious other changes in the analysis methodology [30] Sev-eral studies found that Era-Interim has the best performanceamong reanalysis datasets [31 32] Also some studies foundthat the performance skills of the RegCM driven by Era-Interim are better than by other reanalyses [33 34]

Asmentioned earlier the HCSA poleward edge is definedby the latitude where the OLR annual mean is equal to240Wmminus2 Slight deviations from this OLR value such as236 or 245Wmminus2 lead to similar results Thus a linearinterpolation of the OLR annual mean values averaged over80∘Wndash45∘Wwas performed to obtain the latitude that defines

4 Advances in Meteorology

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

EQLatitude

2

45∘S 40

∘S 35∘S 30

∘S 25∘S 20

∘S 15∘S 10

∘S 5∘S 5

∘N 10∘N

(a)

2

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(b)

Figure 1 Annual mean of the regional HC in the period 1996ndash2011 over (a) South America averaged between 80∘W and 45∘W (HCSA) and(b) Atlantic Ocean averaged between 38∘W and 10∘E (HCAO) Vertical circulations are described by the vertical velocities (Pa sminus1) and thedivergent meridional winds (m sminus1) based on the Era-Interim reanalysis The vertical velocity has been multiplied by minus100 The color scalerepresents the vectorrsquos magnitude

minus3

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

HCS

A in

tens

ity in

dex

Year

(a)

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

Year

HCS

A p

olew

ard

edge

(b)

Figure 2 Normalized time series for the annual mean HCSA (a) intensity index and (b) poleward edge in the period 1996ndash2011 with thelinear trend (red dotted line)

the HCSA poleward edge The normalized time series of theannual mean HCSA poleward edge (Figure 2(b)) presents asignificant negative trend (minus013 degrees of latitude yearminus1)which indicates a poleward expansion of the HCSA Thuscases of HCSA poleward expansion are selected based on theyears that present HCSA normalized poleward edge equal toor less than minus1 in the period 1996ndash2011 The HCSA polewardexpansion cases are 1996 2008 2010 and 2011 Interestinglythese three last years are also part of the HCSA strong casesFigure 3(b) shows the HCSA anomaly for these polewardexpansion cases It is possible to note an ascending branchfrom 0 to 5∘N and a descending branch around 35∘S Hereas expected the descending branch is stronger and is shiftedtowards the pole (Figure 3(b)) compared to the HCSA strongcases (Figure 3(a)) However the ascending branch is weaker(Figure 3(b)) than for the HCSA strong cases (Figure 3(a))Therefore an intense upward motion which induces more

divergence in the upper troposphere distinguishes a HCSAstrong case whereas a HCSA poleward expansion case ischaracterized by a strong and poleward shifted descendingbranch

The normalized time series for the annual mean HCAOintensity index also shows a significant negative trend (minus013yearminus1) (Figure 4(a)) but only the years 2008 and 2010presented HCAO normalized intensity index equal to or lessthan minus1 in the period 1996ndash2011 Moreover the observedtrend for the annual mean HCAO poleward edge normalizedtime series (minus011 yearminus1 119901 value = 004) (Figure 4(b)) is notas significant as theHCSA poleward edge trend (minus013 yearminus1119901 value = 001) Therefore we will focus on the impacts ofthe HCSA intensification and poleward expansion on thehorizontal wind precipitation temperature and geopotentialheight fields over South America However this investigationalso encompasses the Ocean Atlantic region

Advances in Meteorology 5

05

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(a)

05

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(b)

Figure 3 Annual mean of the HCSA anomaly based on the Era-Interim reanalysis for (a) strong cases and (b) poleward expansion casesThe vertical velocity has been multiplied by minus100 The color scale represents the vectorrsquos magnitude

minus3

minus2

minus1

1996 1998 2000 2002 2004 2006 2008 2010

HCA

O in

tens

ity in

dex

Year

0

1

2

(a)

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

Year

HCA

O p

olew

ard

edge

(b)

Figure 4 Normalized time series for the annual mean HCAO (a) intensity index and (b) poleward edge in the period 1996ndash2011 with thelinear trend (red dotted line)

The performance of the regional climate models variesgreatly depending on the combination of physical parame-terization schemes simulation region boundary conditionsand the horizontal resolution [18 20] The skill of theRegCM4 over different regions covering the South Americaand Atlantic Ocean (Figure 5) during the 1996ndash2011 periodis evaluated quantitatively by analyzing the root mean squareerror (RMSE) bias and spatial correlation coefficient of thesimulation in relation to observations for the annual meananomalies We selected the regions in which the impactsof the HCSA intensification and poleward expansion aremore prominent The RMSE measures the overall magnitudeof deviations regardless of their individual signs while thebias measures systematic or predominant deviations in givendirections [35] The closer the values of RMSE and biasget to zero the better the agreement between observed andsimulated data is The RMSE is calculated according to (2)where 1199091015840119872 and 1199091015840119874 are the anomalies of the model and theobservations respectivelyThe summation is carried out over

119873 grid points in a predefined regionThe bias and the spatialcorrelation coefficient are calculated according to (3) and (4)respectively Consider

RMSE = radic 1119873

119873

sum

119894=1

(119909

1015840119872

119894

minus 119909

1015840119874

119894

)

2 (2)

bias = 1119873

119873

sum

119894=1

(119909

1015840119872

119894

minus 119909

1015840119874

119894

) (3)

120588 =

sum

119873

119894=1

(119909

1015840119872

119894

minus 119909

1015840119872

) (119909

1015840119874

119894

minus 119909

1015840119874

)

radic

sum

119873

119894=1

(119909

1015840119872

119894

minus 119909

1015840119872

)

2

radicsum

119873

119894=1

(119909

1015840119874

119894

minus 119909

1015840119874

)

2

(4)

3 Impacts of the HCSA Intensification

The vertical velocity and the divergent winds are the mainfields analyzed when the regional HC is studied However

6 Advances in Meteorology

Table 2 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSAstrong cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 042 031 152 026 012 012 052 minus010 078 035 minus015 0112 091 018 329 023 minus067 002 318 004 068 050 087 0403 051 019 167 011 039 013 155 006 083 088 098 0854 065 025 313 019 minus011 011 225 minus001 084 073 091 055

Latit

ude

Longitude

1

2

3

4

5

6

7

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

Figure 5 South America regions selected for statistical evaluationof model performance (1) northern (80∘Wndash35∘W 15∘Sndash10∘N) (2)central (80∘Wndash35∘W 35∘Sndash15∘S) (3) southern (80∘Wndash55∘W 55∘Sndash35∘S) (4) entire domain (80∘Wndash10∘W 55∘Sndash10∘N) (5) southeast(70∘Wndash50∘W 40∘Sndash30∘S) (6) northeast (60∘Wndash35∘W 10∘Sndash5∘N)and (7) all South America (80∘Wndash35∘W 55∘Sndash10∘S)

there is a remarkable impact of the changes in the HC onthe horizontal winds at low and high troposphere as they arelinked to the vertical movement of the air Indeed the tradewinds are the expression of the HC at surface

Figure 6 shows the anomalies of the annual mean hor-izontal wind at 200 and 850 hPa for the HCSA strongcases based on the Era-Interim reanalysis and the modelsimulation Strong anomalous easterly winds coming fromthe Atlantic Ocean are seen over the northern South America(SA) at 200 hPa in the reanalysis (Figure 6(a)) and especiallyin model simulation (Figure 6(c)) which exhibits a positivebias (012) for the zonal wind in this region (Table 2) Whenentering the mainland these winds rotate counterclockwiseStrong anomalous easterly winds coming from the Atlanticalso hits the southeast of the continent and then they rotate

clockwise resulting in an anomalous trough which is seenat 200 hPa (Figure not shown) and 500 hPa (Figure 7) Theanomalous wind pattern at 850 hPa is characterized by thestrengthening of the trade winds around the equator whichspin in northeast direction before entering the northern SAand by a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 6(b)) and model simu-lation (Figure 6(d))This counterclockwise circulation resultsin a significant anomalous ridge which is seen at 850 hPa(Figure not shown) and 500 hPa in the reanalysis and modelsimulation (Figure 7)

To investigate the quantitative link between the HCSAintensity and the annual changes in the zonal wind (strongercomponent of the horizontal wind) correlation coefficientsbetween the HCSA annual intensity index and the annualmean zonal wind at 200 and 850 hPa are displayed in Figure 8The model simulation (Figure 8(c)) is able to reproduce thesigns (positives and negatives) of the correlation coefficientsobserved in the reanalysis (Figure 8(a)) Three regions withsignificant correlations (above 05) in the high troposphereare seen in the reanalysis (Figure 8(a)) Figures 6(a) and 6(c)showed that for HCSA strong cases the zonal component ofthe anomalous wind at 200 hPa is easterly (negative) in thenorthern and southern SA regions whereas in the centralarea of SA it is westerly (positive) In anotherway Figures 8(a)and 8(c) show that the anomalous easterly (westerly) windscoming from (towards) the Atlantic presents negative (pos-itive) correlations with the HCSA intensity At 850 hPasignificant positive correlations are seen in the reanalysison the edge of northwest SA and negative correlations onthe central-south area of the continent (Figure 8(b)) Themodel simulation shows significant positive correlations ona great part of the northwest SA (Figure 8(d)) probably dueto the misrepresentation of the horizontal wind in this region(Figure 6(d)) At 850 hPa it can also be seen that regions withanomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA intensity

The three regions (northern central and southern) withsignificant correlations seen in Figure 8(a) and the entiredomain considered in the composite analyses were selectedfor evaluation of model performance in reproducing theanomalies of zonal wind geopotential height and temper-ature (Figure 5) Table 2 shows that for the annual meanzonal wind anomalies at 200 hPa lower values of RMSE andbias (absolute value) and higher values of spatial correlation

Advances in Meteorology 7La

titud

e

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 6 Annual mean horizontal wind (m sminus1) anomaly for HCSA strong cases in the period 1996ndash2011 based on the Era-Interim reanalysisat (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents the vectorrsquos magnitude

8 Advances in MeteorologyLa

titud

e

Longitude

18

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

0

0

0

0

minus4

minus4

minus4

minus4

minus2

minus2

minus2

minus2

2

4

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus2

minus2

0

0

0

2

2

2

2

1820

1614

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 7 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

are found mainly in the northern and southern SA andwhen we consider the entire domain (region 4 of Figure 5)Thus at 200 hPa the central region of SA exhibits weakeragreement between the model and observation for the zonalwind anomalies Table 2 shows that themodel underestimatesthe zonal wind anomalies in this region (bias = minus067) Aswell Figures 6(c) and 7(b) show that the model simulatesa strong anticyclonic anomalous circulation in the central-western region Interestingly at 850 hPa the central region ofSA exhibits the lowest value of RMSE and bias (018 and 002resp) but the spatial correlation is only 05 Mid- and upper-tropospheric winds influence low-level circulations thusthe biases in low and high levels are somewhat connectedIncreasing resolution can improve the results in the centralregion of SA due to better representation of orography in thisregion in a high-resolution simulation The highest spatialcorrelation for the zonal wind anomalies at 850 hPa is foundin the southern SA (088) The geopotential height is thevariable that exhibits the highest values of RMSE and bias(mostly greater than 1) for all regions (Table 2) The lowervalues of RMSE and bias are found in the northern andsouthern SA but the spatial correlation is weaker in thenorthern and stronger in the southern

Figure 9 shows the annual mean temperature anomaliesfor HCSA strong cases The reanalysis data presents signif-icant positive anomalies on a great part of northern SA andequatorial Atlantic whereas the negative anomalies are foundover the east Pacific around 20∘S (Figure 9(a)) The model

shows significant positive anomalies only over the northeastedge of SA and over the equatorial Atlantic (Figure 9(b))Indeed themodel underestimates the temperature anomaliesin the northern region as indicated by the bias value in thisregion (minus011 Table 2) A positive (negative) bias in surfaceair temperature is associated with a negative (positive) biasin precipitation as an underestimation (overestimation) ofprecipitation is usually associated with an underestimation(overestimation) of clouds which enhances (reduces) thenet short-wave radiation budget at surface and consequentlyenhances (reduces) the net energy budget at surface [36]Indeed the model exhibits a positive bias (019) for theprecipitation anomalies in the northern region The lowest(highest) value of the RMSE (correlation) for the annualmean temperature anomalies is found in the southern regionindicating a good agreement between the model and obser-vation

Figure 10 displays the correlation between the HCSAannual intensity index and the annual mean temperaturein the period 1996ndash2011 As in Figure 8(a) three signifi-cant regions (northern central and southern SA) are seenHowever the signs of the correlation coefficients in thesethree regions of Figure 10 are opposite in relation to Figure 8Thus positive (negative) correlations of the annual meantemperature (zonal wind) with the HCSA intensity are seenover the northern and southern SA in the 1996ndash2011 periodThe opposite signal between the annual mean temperatureand the zonal wind is also seen over the central regions of

Advances in Meteorology 9

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus06

minus06

minus06

minus04

minus04

minus04

minus04

minus02

minus02

minus020

0

0

02

02

02

02

04

04

0404

06

06

08

minus020

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

02

02 02

02

02

02

02

04

04

04

06

08

0

0

0

0

0

0

0

0

0

Latit

ude

Longitude

minus06

minus06

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02minus02

minus02

minus02

minus02

04

04

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

minus02

minus02

minus02

minus02

minus02

minus08

minus08

minus04

minus04

minus04

minus04

minus06

minus06

minus06

0202

02

02

040

0

0

0

0

Latit

ude

Longitude

08

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04minus04

minus04

minus04

minus04

minus06

02

02

02

02

02

02

0204

04

04

04

04

04

06 06

08

08

08

0

00

0

0

0

0

Latit

ude

Longitude

06

minus02

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 8 Correlation between the HCSA annual intensity index and the annual mean zonal wind (m sminus1) in the period 1996ndash2011 based onthe Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervals are 02and values above 05 are shaded and are statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSAannual intensity index time series have been multiplied by (minus1)

10 Advances in Meteorology

02

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus05

minus04

minus03

minus03

minus03

minus02

minus02

minus01

minus01

minus01

minus01

0

0

0

0

01

01

01

01

02

02

02

02

02

02

02

02

02

03

03

03

03

03

04

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

minus02

minus02

minus02

minus02

minus02

minus01

minus01

minus01

minus01

minus01

minus01

minus01

00

0

0

0

0

0

01

01

01

01

01

02

02

02

02

02

03

03

04

05

03

03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 9 Annual mean temperature (K) anomaly at 850 hPa for HCSA strong cases in the period 1996ndash2011 based on the (a) Era-Interimreanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level from two-tailed Studentrsquos119905-test are shaded

minus06minus06

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

02

02

02

02

02

04

04

04

04

04

06

06

06

06

06

Latit

ude

Longitude

0

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

04

04

04

02

02

02

02

04

04

06

0606

08

08

00

0

0

0

0

Latit

ude

Longitude

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 10 Correlation between the HCSA annual intensity index and the annual mean temperature (K) at 850 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSA annual intensity index has beenmultiplied by (minus1)

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 3: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Advances in Meteorology 3

Table 1 Technical details of the simulation performed

Simulation parametersDomain cartographic projection(iproj)

Rotated Mercator(ROTMER)

Grid point horizontal resolution inkm (ds) 600 km

Number of points in the NSdirection (iy) 152

Number of points in the EWdirection (jx) 192

Number of vertical levels (kz) 18Central latitude of model domain indegrees (clat) minus2300

Central longitude of model domainin degrees (clon) minus3300

Cumulus convection scheme (icup) 1-Kuo

Moisture scheme (ipptls) 1-Explicit moisture(SUBEX Pal et al [42])

Lateral boundary conditionsscheme (iboudy)

5-Relaxationexponential technique

Boundary layer scheme (ibltyp) 1-Holtslag PBL [43]

Ocean flux scheme (iocnflx) 2-Zeng et al [44]

including the simulation domain The model is run for theperiod 1995ndash2011 The first year is discarded to avoid prob-lems related to the spin-up

The Kuo-type cumulus parameterization is used here Inthis scheme convection is triggered in a convectively unstablelow troposphere when the column moisture convergenceexceeds a threshold value [11] Some studies found thatthe Kuo scheme produces satisfactory seasonal and annualprecipitation [22 23] We will show that this scheme couldreasonably reproduce the precipitation anomalies especiallyin the southern region associated with the modifications inthe HC over South America

The Era-Interim monthly data the Global PrecipitationClimatology Centre dataset (GPCC) (GPCC dataset is avail-able on httpwwwesrlnoaagovpsddatagriddeddatagpcchtml) [24] and the SST monthly mean data of the HadleyCentre Global Sea Ice and Sea Surface Temperature (HAD-ISST) (HADISST dataset is available on httpwwwmetofficegovukhadobshadisstdatadownloadhtml) [25] are used forvalidation of the model results

23 Methods Horizontal wind field can be expressed bythe sum of a nondivergent (or rotational) component and adivergent (or irrotational) component [26 27]

V = V120595+ V120594=

119896 times nabla120595 + nabla120594 (1)

where 120595 is stream function and 120594 is velocity potential Thedivergent component is fundamental to study the atmo-spheric divergence-convergence that drives the verticalmotion and circulation in the tropics [28]

The methodology used here to define the intensityindex and poleward edge of the regional HC is similar toChen et al [6] The regional HC intensity is defined throughan index based on the vertical shear of divergent meridionalwinds at 200 hPa and 850 hPa whereas the poleward edge ofthe regional HC is identified as the latitude where the OLRis equal to 240Wmminus2 (found through linear interpolation)The OLR data is obtained from the Climate DiagnosticsCenter of the National Oceanic and Atmospheric Admin-istration (NOAA-CDC) (OLRdata is available on httpwwwesrlnoaagovpsddatagriddeddatainterp OLRhtml) in a25∘times 25∘ grid [29] Annual means are constructed by aver-aging the monthly data from January to December

Initially two domains are selected for the regional HCsthe SouthAmerica (80∘Wndash45∘W)andAtlanticOcean (38∘Wndash10∘E) regionsThe regionalHCs in these locations can have animportant influence on synoptic weather and climate systems[6] The annual mean of the intensity index for the regionalHC over South America (hereafter denoted as HCSA) isdefined as the difference in the area-averaged (80∘Wndash45∘W20∘Sndash0) divergent meridional winds between 200 hPa and850 hPa This latitudinal band is chosen according to theclimatological distributions of divergent winds (Figure 1(a))Definition of the intensity index for the regional HC overAtlantic Ocean (hereafter denoted HCAO) is similar to thoseof the HCSA except that the divergent meridional winds areaveraged over 38∘Wndash10∘E 5∘Nndash20∘S (Figure 1(b)) Chen et al[6] conducted a series of tests and found that the result doesnot change much if other latitudinal bands are used

The normalized time series for the annual mean HCSAintensity index (Figure 2(a)) presents a significant negativetrend (minus012 yearminus1) This indicates an intensification of theHCSA as the HC in SH is negative corresponding to acounterclockwise circulationThus cases of strongHCSA areselected based on the years that present HCSA normalizedintensity index equal to or less than minus1 in the period 1996ndash2011 The HCSA strong cases are 2005 2008 2010 and 2011Figure 3(a) shows the HCSA anomaly for these strong casesIt is possible to note an intense ascending branch from 0 to5∘N and a descending branch around 30∘S

It is worth highlighting that we are using the Era-Interimreanalysis data to calculate the intensity indexThis dataset ischosen since it is the latest global atmospheric reanalysis pro-duced by ECMWF and thus incorporates many importantmodel improvements such as resolution and physics changesthe use of four-dimensional variational data assimilation andvarious other changes in the analysis methodology [30] Sev-eral studies found that Era-Interim has the best performanceamong reanalysis datasets [31 32] Also some studies foundthat the performance skills of the RegCM driven by Era-Interim are better than by other reanalyses [33 34]

Asmentioned earlier the HCSA poleward edge is definedby the latitude where the OLR annual mean is equal to240Wmminus2 Slight deviations from this OLR value such as236 or 245Wmminus2 lead to similar results Thus a linearinterpolation of the OLR annual mean values averaged over80∘Wndash45∘Wwas performed to obtain the latitude that defines

4 Advances in Meteorology

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

EQLatitude

2

45∘S 40

∘S 35∘S 30

∘S 25∘S 20

∘S 15∘S 10

∘S 5∘S 5

∘N 10∘N

(a)

2

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(b)

Figure 1 Annual mean of the regional HC in the period 1996ndash2011 over (a) South America averaged between 80∘W and 45∘W (HCSA) and(b) Atlantic Ocean averaged between 38∘W and 10∘E (HCAO) Vertical circulations are described by the vertical velocities (Pa sminus1) and thedivergent meridional winds (m sminus1) based on the Era-Interim reanalysis The vertical velocity has been multiplied by minus100 The color scalerepresents the vectorrsquos magnitude

minus3

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

HCS

A in

tens

ity in

dex

Year

(a)

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

Year

HCS

A p

olew

ard

edge

(b)

Figure 2 Normalized time series for the annual mean HCSA (a) intensity index and (b) poleward edge in the period 1996ndash2011 with thelinear trend (red dotted line)

the HCSA poleward edge The normalized time series of theannual mean HCSA poleward edge (Figure 2(b)) presents asignificant negative trend (minus013 degrees of latitude yearminus1)which indicates a poleward expansion of the HCSA Thuscases of HCSA poleward expansion are selected based on theyears that present HCSA normalized poleward edge equal toor less than minus1 in the period 1996ndash2011 The HCSA polewardexpansion cases are 1996 2008 2010 and 2011 Interestinglythese three last years are also part of the HCSA strong casesFigure 3(b) shows the HCSA anomaly for these polewardexpansion cases It is possible to note an ascending branchfrom 0 to 5∘N and a descending branch around 35∘S Hereas expected the descending branch is stronger and is shiftedtowards the pole (Figure 3(b)) compared to the HCSA strongcases (Figure 3(a)) However the ascending branch is weaker(Figure 3(b)) than for the HCSA strong cases (Figure 3(a))Therefore an intense upward motion which induces more

divergence in the upper troposphere distinguishes a HCSAstrong case whereas a HCSA poleward expansion case ischaracterized by a strong and poleward shifted descendingbranch

The normalized time series for the annual mean HCAOintensity index also shows a significant negative trend (minus013yearminus1) (Figure 4(a)) but only the years 2008 and 2010presented HCAO normalized intensity index equal to or lessthan minus1 in the period 1996ndash2011 Moreover the observedtrend for the annual mean HCAO poleward edge normalizedtime series (minus011 yearminus1 119901 value = 004) (Figure 4(b)) is notas significant as theHCSA poleward edge trend (minus013 yearminus1119901 value = 001) Therefore we will focus on the impacts ofthe HCSA intensification and poleward expansion on thehorizontal wind precipitation temperature and geopotentialheight fields over South America However this investigationalso encompasses the Ocean Atlantic region

Advances in Meteorology 5

05

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(a)

05

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(b)

Figure 3 Annual mean of the HCSA anomaly based on the Era-Interim reanalysis for (a) strong cases and (b) poleward expansion casesThe vertical velocity has been multiplied by minus100 The color scale represents the vectorrsquos magnitude

minus3

minus2

minus1

1996 1998 2000 2002 2004 2006 2008 2010

HCA

O in

tens

ity in

dex

Year

0

1

2

(a)

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

Year

HCA

O p

olew

ard

edge

(b)

Figure 4 Normalized time series for the annual mean HCAO (a) intensity index and (b) poleward edge in the period 1996ndash2011 with thelinear trend (red dotted line)

The performance of the regional climate models variesgreatly depending on the combination of physical parame-terization schemes simulation region boundary conditionsand the horizontal resolution [18 20] The skill of theRegCM4 over different regions covering the South Americaand Atlantic Ocean (Figure 5) during the 1996ndash2011 periodis evaluated quantitatively by analyzing the root mean squareerror (RMSE) bias and spatial correlation coefficient of thesimulation in relation to observations for the annual meananomalies We selected the regions in which the impactsof the HCSA intensification and poleward expansion aremore prominent The RMSE measures the overall magnitudeof deviations regardless of their individual signs while thebias measures systematic or predominant deviations in givendirections [35] The closer the values of RMSE and biasget to zero the better the agreement between observed andsimulated data is The RMSE is calculated according to (2)where 1199091015840119872 and 1199091015840119874 are the anomalies of the model and theobservations respectivelyThe summation is carried out over

119873 grid points in a predefined regionThe bias and the spatialcorrelation coefficient are calculated according to (3) and (4)respectively Consider

RMSE = radic 1119873

119873

sum

119894=1

(119909

1015840119872

119894

minus 119909

1015840119874

119894

)

2 (2)

bias = 1119873

119873

sum

119894=1

(119909

1015840119872

119894

minus 119909

1015840119874

119894

) (3)

120588 =

sum

119873

119894=1

(119909

1015840119872

119894

minus 119909

1015840119872

) (119909

1015840119874

119894

minus 119909

1015840119874

)

radic

sum

119873

119894=1

(119909

1015840119872

119894

minus 119909

1015840119872

)

2

radicsum

119873

119894=1

(119909

1015840119874

119894

minus 119909

1015840119874

)

2

(4)

3 Impacts of the HCSA Intensification

The vertical velocity and the divergent winds are the mainfields analyzed when the regional HC is studied However

6 Advances in Meteorology

Table 2 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSAstrong cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 042 031 152 026 012 012 052 minus010 078 035 minus015 0112 091 018 329 023 minus067 002 318 004 068 050 087 0403 051 019 167 011 039 013 155 006 083 088 098 0854 065 025 313 019 minus011 011 225 minus001 084 073 091 055

Latit

ude

Longitude

1

2

3

4

5

6

7

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

Figure 5 South America regions selected for statistical evaluationof model performance (1) northern (80∘Wndash35∘W 15∘Sndash10∘N) (2)central (80∘Wndash35∘W 35∘Sndash15∘S) (3) southern (80∘Wndash55∘W 55∘Sndash35∘S) (4) entire domain (80∘Wndash10∘W 55∘Sndash10∘N) (5) southeast(70∘Wndash50∘W 40∘Sndash30∘S) (6) northeast (60∘Wndash35∘W 10∘Sndash5∘N)and (7) all South America (80∘Wndash35∘W 55∘Sndash10∘S)

there is a remarkable impact of the changes in the HC onthe horizontal winds at low and high troposphere as they arelinked to the vertical movement of the air Indeed the tradewinds are the expression of the HC at surface

Figure 6 shows the anomalies of the annual mean hor-izontal wind at 200 and 850 hPa for the HCSA strongcases based on the Era-Interim reanalysis and the modelsimulation Strong anomalous easterly winds coming fromthe Atlantic Ocean are seen over the northern South America(SA) at 200 hPa in the reanalysis (Figure 6(a)) and especiallyin model simulation (Figure 6(c)) which exhibits a positivebias (012) for the zonal wind in this region (Table 2) Whenentering the mainland these winds rotate counterclockwiseStrong anomalous easterly winds coming from the Atlanticalso hits the southeast of the continent and then they rotate

clockwise resulting in an anomalous trough which is seenat 200 hPa (Figure not shown) and 500 hPa (Figure 7) Theanomalous wind pattern at 850 hPa is characterized by thestrengthening of the trade winds around the equator whichspin in northeast direction before entering the northern SAand by a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 6(b)) and model simu-lation (Figure 6(d))This counterclockwise circulation resultsin a significant anomalous ridge which is seen at 850 hPa(Figure not shown) and 500 hPa in the reanalysis and modelsimulation (Figure 7)

To investigate the quantitative link between the HCSAintensity and the annual changes in the zonal wind (strongercomponent of the horizontal wind) correlation coefficientsbetween the HCSA annual intensity index and the annualmean zonal wind at 200 and 850 hPa are displayed in Figure 8The model simulation (Figure 8(c)) is able to reproduce thesigns (positives and negatives) of the correlation coefficientsobserved in the reanalysis (Figure 8(a)) Three regions withsignificant correlations (above 05) in the high troposphereare seen in the reanalysis (Figure 8(a)) Figures 6(a) and 6(c)showed that for HCSA strong cases the zonal component ofthe anomalous wind at 200 hPa is easterly (negative) in thenorthern and southern SA regions whereas in the centralarea of SA it is westerly (positive) In anotherway Figures 8(a)and 8(c) show that the anomalous easterly (westerly) windscoming from (towards) the Atlantic presents negative (pos-itive) correlations with the HCSA intensity At 850 hPasignificant positive correlations are seen in the reanalysison the edge of northwest SA and negative correlations onthe central-south area of the continent (Figure 8(b)) Themodel simulation shows significant positive correlations ona great part of the northwest SA (Figure 8(d)) probably dueto the misrepresentation of the horizontal wind in this region(Figure 6(d)) At 850 hPa it can also be seen that regions withanomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA intensity

The three regions (northern central and southern) withsignificant correlations seen in Figure 8(a) and the entiredomain considered in the composite analyses were selectedfor evaluation of model performance in reproducing theanomalies of zonal wind geopotential height and temper-ature (Figure 5) Table 2 shows that for the annual meanzonal wind anomalies at 200 hPa lower values of RMSE andbias (absolute value) and higher values of spatial correlation

Advances in Meteorology 7La

titud

e

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 6 Annual mean horizontal wind (m sminus1) anomaly for HCSA strong cases in the period 1996ndash2011 based on the Era-Interim reanalysisat (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents the vectorrsquos magnitude

8 Advances in MeteorologyLa

titud

e

Longitude

18

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

0

0

0

0

minus4

minus4

minus4

minus4

minus2

minus2

minus2

minus2

2

4

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus2

minus2

0

0

0

2

2

2

2

1820

1614

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 7 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

are found mainly in the northern and southern SA andwhen we consider the entire domain (region 4 of Figure 5)Thus at 200 hPa the central region of SA exhibits weakeragreement between the model and observation for the zonalwind anomalies Table 2 shows that themodel underestimatesthe zonal wind anomalies in this region (bias = minus067) Aswell Figures 6(c) and 7(b) show that the model simulatesa strong anticyclonic anomalous circulation in the central-western region Interestingly at 850 hPa the central region ofSA exhibits the lowest value of RMSE and bias (018 and 002resp) but the spatial correlation is only 05 Mid- and upper-tropospheric winds influence low-level circulations thusthe biases in low and high levels are somewhat connectedIncreasing resolution can improve the results in the centralregion of SA due to better representation of orography in thisregion in a high-resolution simulation The highest spatialcorrelation for the zonal wind anomalies at 850 hPa is foundin the southern SA (088) The geopotential height is thevariable that exhibits the highest values of RMSE and bias(mostly greater than 1) for all regions (Table 2) The lowervalues of RMSE and bias are found in the northern andsouthern SA but the spatial correlation is weaker in thenorthern and stronger in the southern

Figure 9 shows the annual mean temperature anomaliesfor HCSA strong cases The reanalysis data presents signif-icant positive anomalies on a great part of northern SA andequatorial Atlantic whereas the negative anomalies are foundover the east Pacific around 20∘S (Figure 9(a)) The model

shows significant positive anomalies only over the northeastedge of SA and over the equatorial Atlantic (Figure 9(b))Indeed themodel underestimates the temperature anomaliesin the northern region as indicated by the bias value in thisregion (minus011 Table 2) A positive (negative) bias in surfaceair temperature is associated with a negative (positive) biasin precipitation as an underestimation (overestimation) ofprecipitation is usually associated with an underestimation(overestimation) of clouds which enhances (reduces) thenet short-wave radiation budget at surface and consequentlyenhances (reduces) the net energy budget at surface [36]Indeed the model exhibits a positive bias (019) for theprecipitation anomalies in the northern region The lowest(highest) value of the RMSE (correlation) for the annualmean temperature anomalies is found in the southern regionindicating a good agreement between the model and obser-vation

Figure 10 displays the correlation between the HCSAannual intensity index and the annual mean temperaturein the period 1996ndash2011 As in Figure 8(a) three signifi-cant regions (northern central and southern SA) are seenHowever the signs of the correlation coefficients in thesethree regions of Figure 10 are opposite in relation to Figure 8Thus positive (negative) correlations of the annual meantemperature (zonal wind) with the HCSA intensity are seenover the northern and southern SA in the 1996ndash2011 periodThe opposite signal between the annual mean temperatureand the zonal wind is also seen over the central regions of

Advances in Meteorology 9

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus06

minus06

minus06

minus04

minus04

minus04

minus04

minus02

minus02

minus020

0

0

02

02

02

02

04

04

0404

06

06

08

minus020

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

02

02 02

02

02

02

02

04

04

04

06

08

0

0

0

0

0

0

0

0

0

Latit

ude

Longitude

minus06

minus06

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02minus02

minus02

minus02

minus02

04

04

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

minus02

minus02

minus02

minus02

minus02

minus08

minus08

minus04

minus04

minus04

minus04

minus06

minus06

minus06

0202

02

02

040

0

0

0

0

Latit

ude

Longitude

08

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04minus04

minus04

minus04

minus04

minus06

02

02

02

02

02

02

0204

04

04

04

04

04

06 06

08

08

08

0

00

0

0

0

0

Latit

ude

Longitude

06

minus02

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 8 Correlation between the HCSA annual intensity index and the annual mean zonal wind (m sminus1) in the period 1996ndash2011 based onthe Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervals are 02and values above 05 are shaded and are statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSAannual intensity index time series have been multiplied by (minus1)

10 Advances in Meteorology

02

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus05

minus04

minus03

minus03

minus03

minus02

minus02

minus01

minus01

minus01

minus01

0

0

0

0

01

01

01

01

02

02

02

02

02

02

02

02

02

03

03

03

03

03

04

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

minus02

minus02

minus02

minus02

minus02

minus01

minus01

minus01

minus01

minus01

minus01

minus01

00

0

0

0

0

0

01

01

01

01

01

02

02

02

02

02

03

03

04

05

03

03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 9 Annual mean temperature (K) anomaly at 850 hPa for HCSA strong cases in the period 1996ndash2011 based on the (a) Era-Interimreanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level from two-tailed Studentrsquos119905-test are shaded

minus06minus06

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

02

02

02

02

02

04

04

04

04

04

06

06

06

06

06

Latit

ude

Longitude

0

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

04

04

04

02

02

02

02

04

04

06

0606

08

08

00

0

0

0

0

Latit

ude

Longitude

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 10 Correlation between the HCSA annual intensity index and the annual mean temperature (K) at 850 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSA annual intensity index has beenmultiplied by (minus1)

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 4: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

4 Advances in Meteorology

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

EQLatitude

2

45∘S 40

∘S 35∘S 30

∘S 25∘S 20

∘S 15∘S 10

∘S 5∘S 5

∘N 10∘N

(a)

2

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(b)

Figure 1 Annual mean of the regional HC in the period 1996ndash2011 over (a) South America averaged between 80∘W and 45∘W (HCSA) and(b) Atlantic Ocean averaged between 38∘W and 10∘E (HCAO) Vertical circulations are described by the vertical velocities (Pa sminus1) and thedivergent meridional winds (m sminus1) based on the Era-Interim reanalysis The vertical velocity has been multiplied by minus100 The color scalerepresents the vectorrsquos magnitude

minus3

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

HCS

A in

tens

ity in

dex

Year

(a)

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

Year

HCS

A p

olew

ard

edge

(b)

Figure 2 Normalized time series for the annual mean HCSA (a) intensity index and (b) poleward edge in the period 1996ndash2011 with thelinear trend (red dotted line)

the HCSA poleward edge The normalized time series of theannual mean HCSA poleward edge (Figure 2(b)) presents asignificant negative trend (minus013 degrees of latitude yearminus1)which indicates a poleward expansion of the HCSA Thuscases of HCSA poleward expansion are selected based on theyears that present HCSA normalized poleward edge equal toor less than minus1 in the period 1996ndash2011 The HCSA polewardexpansion cases are 1996 2008 2010 and 2011 Interestinglythese three last years are also part of the HCSA strong casesFigure 3(b) shows the HCSA anomaly for these polewardexpansion cases It is possible to note an ascending branchfrom 0 to 5∘N and a descending branch around 35∘S Hereas expected the descending branch is stronger and is shiftedtowards the pole (Figure 3(b)) compared to the HCSA strongcases (Figure 3(a)) However the ascending branch is weaker(Figure 3(b)) than for the HCSA strong cases (Figure 3(a))Therefore an intense upward motion which induces more

divergence in the upper troposphere distinguishes a HCSAstrong case whereas a HCSA poleward expansion case ischaracterized by a strong and poleward shifted descendingbranch

The normalized time series for the annual mean HCAOintensity index also shows a significant negative trend (minus013yearminus1) (Figure 4(a)) but only the years 2008 and 2010presented HCAO normalized intensity index equal to or lessthan minus1 in the period 1996ndash2011 Moreover the observedtrend for the annual mean HCAO poleward edge normalizedtime series (minus011 yearminus1 119901 value = 004) (Figure 4(b)) is notas significant as theHCSA poleward edge trend (minus013 yearminus1119901 value = 001) Therefore we will focus on the impacts ofthe HCSA intensification and poleward expansion on thehorizontal wind precipitation temperature and geopotentialheight fields over South America However this investigationalso encompasses the Ocean Atlantic region

Advances in Meteorology 5

05

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(a)

05

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(b)

Figure 3 Annual mean of the HCSA anomaly based on the Era-Interim reanalysis for (a) strong cases and (b) poleward expansion casesThe vertical velocity has been multiplied by minus100 The color scale represents the vectorrsquos magnitude

minus3

minus2

minus1

1996 1998 2000 2002 2004 2006 2008 2010

HCA

O in

tens

ity in

dex

Year

0

1

2

(a)

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

Year

HCA

O p

olew

ard

edge

(b)

Figure 4 Normalized time series for the annual mean HCAO (a) intensity index and (b) poleward edge in the period 1996ndash2011 with thelinear trend (red dotted line)

The performance of the regional climate models variesgreatly depending on the combination of physical parame-terization schemes simulation region boundary conditionsand the horizontal resolution [18 20] The skill of theRegCM4 over different regions covering the South Americaand Atlantic Ocean (Figure 5) during the 1996ndash2011 periodis evaluated quantitatively by analyzing the root mean squareerror (RMSE) bias and spatial correlation coefficient of thesimulation in relation to observations for the annual meananomalies We selected the regions in which the impactsof the HCSA intensification and poleward expansion aremore prominent The RMSE measures the overall magnitudeof deviations regardless of their individual signs while thebias measures systematic or predominant deviations in givendirections [35] The closer the values of RMSE and biasget to zero the better the agreement between observed andsimulated data is The RMSE is calculated according to (2)where 1199091015840119872 and 1199091015840119874 are the anomalies of the model and theobservations respectivelyThe summation is carried out over

119873 grid points in a predefined regionThe bias and the spatialcorrelation coefficient are calculated according to (3) and (4)respectively Consider

RMSE = radic 1119873

119873

sum

119894=1

(119909

1015840119872

119894

minus 119909

1015840119874

119894

)

2 (2)

bias = 1119873

119873

sum

119894=1

(119909

1015840119872

119894

minus 119909

1015840119874

119894

) (3)

120588 =

sum

119873

119894=1

(119909

1015840119872

119894

minus 119909

1015840119872

) (119909

1015840119874

119894

minus 119909

1015840119874

)

radic

sum

119873

119894=1

(119909

1015840119872

119894

minus 119909

1015840119872

)

2

radicsum

119873

119894=1

(119909

1015840119874

119894

minus 119909

1015840119874

)

2

(4)

3 Impacts of the HCSA Intensification

The vertical velocity and the divergent winds are the mainfields analyzed when the regional HC is studied However

6 Advances in Meteorology

Table 2 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSAstrong cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 042 031 152 026 012 012 052 minus010 078 035 minus015 0112 091 018 329 023 minus067 002 318 004 068 050 087 0403 051 019 167 011 039 013 155 006 083 088 098 0854 065 025 313 019 minus011 011 225 minus001 084 073 091 055

Latit

ude

Longitude

1

2

3

4

5

6

7

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

Figure 5 South America regions selected for statistical evaluationof model performance (1) northern (80∘Wndash35∘W 15∘Sndash10∘N) (2)central (80∘Wndash35∘W 35∘Sndash15∘S) (3) southern (80∘Wndash55∘W 55∘Sndash35∘S) (4) entire domain (80∘Wndash10∘W 55∘Sndash10∘N) (5) southeast(70∘Wndash50∘W 40∘Sndash30∘S) (6) northeast (60∘Wndash35∘W 10∘Sndash5∘N)and (7) all South America (80∘Wndash35∘W 55∘Sndash10∘S)

there is a remarkable impact of the changes in the HC onthe horizontal winds at low and high troposphere as they arelinked to the vertical movement of the air Indeed the tradewinds are the expression of the HC at surface

Figure 6 shows the anomalies of the annual mean hor-izontal wind at 200 and 850 hPa for the HCSA strongcases based on the Era-Interim reanalysis and the modelsimulation Strong anomalous easterly winds coming fromthe Atlantic Ocean are seen over the northern South America(SA) at 200 hPa in the reanalysis (Figure 6(a)) and especiallyin model simulation (Figure 6(c)) which exhibits a positivebias (012) for the zonal wind in this region (Table 2) Whenentering the mainland these winds rotate counterclockwiseStrong anomalous easterly winds coming from the Atlanticalso hits the southeast of the continent and then they rotate

clockwise resulting in an anomalous trough which is seenat 200 hPa (Figure not shown) and 500 hPa (Figure 7) Theanomalous wind pattern at 850 hPa is characterized by thestrengthening of the trade winds around the equator whichspin in northeast direction before entering the northern SAand by a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 6(b)) and model simu-lation (Figure 6(d))This counterclockwise circulation resultsin a significant anomalous ridge which is seen at 850 hPa(Figure not shown) and 500 hPa in the reanalysis and modelsimulation (Figure 7)

To investigate the quantitative link between the HCSAintensity and the annual changes in the zonal wind (strongercomponent of the horizontal wind) correlation coefficientsbetween the HCSA annual intensity index and the annualmean zonal wind at 200 and 850 hPa are displayed in Figure 8The model simulation (Figure 8(c)) is able to reproduce thesigns (positives and negatives) of the correlation coefficientsobserved in the reanalysis (Figure 8(a)) Three regions withsignificant correlations (above 05) in the high troposphereare seen in the reanalysis (Figure 8(a)) Figures 6(a) and 6(c)showed that for HCSA strong cases the zonal component ofthe anomalous wind at 200 hPa is easterly (negative) in thenorthern and southern SA regions whereas in the centralarea of SA it is westerly (positive) In anotherway Figures 8(a)and 8(c) show that the anomalous easterly (westerly) windscoming from (towards) the Atlantic presents negative (pos-itive) correlations with the HCSA intensity At 850 hPasignificant positive correlations are seen in the reanalysison the edge of northwest SA and negative correlations onthe central-south area of the continent (Figure 8(b)) Themodel simulation shows significant positive correlations ona great part of the northwest SA (Figure 8(d)) probably dueto the misrepresentation of the horizontal wind in this region(Figure 6(d)) At 850 hPa it can also be seen that regions withanomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA intensity

The three regions (northern central and southern) withsignificant correlations seen in Figure 8(a) and the entiredomain considered in the composite analyses were selectedfor evaluation of model performance in reproducing theanomalies of zonal wind geopotential height and temper-ature (Figure 5) Table 2 shows that for the annual meanzonal wind anomalies at 200 hPa lower values of RMSE andbias (absolute value) and higher values of spatial correlation

Advances in Meteorology 7La

titud

e

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 6 Annual mean horizontal wind (m sminus1) anomaly for HCSA strong cases in the period 1996ndash2011 based on the Era-Interim reanalysisat (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents the vectorrsquos magnitude

8 Advances in MeteorologyLa

titud

e

Longitude

18

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

0

0

0

0

minus4

minus4

minus4

minus4

minus2

minus2

minus2

minus2

2

4

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus2

minus2

0

0

0

2

2

2

2

1820

1614

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 7 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

are found mainly in the northern and southern SA andwhen we consider the entire domain (region 4 of Figure 5)Thus at 200 hPa the central region of SA exhibits weakeragreement between the model and observation for the zonalwind anomalies Table 2 shows that themodel underestimatesthe zonal wind anomalies in this region (bias = minus067) Aswell Figures 6(c) and 7(b) show that the model simulatesa strong anticyclonic anomalous circulation in the central-western region Interestingly at 850 hPa the central region ofSA exhibits the lowest value of RMSE and bias (018 and 002resp) but the spatial correlation is only 05 Mid- and upper-tropospheric winds influence low-level circulations thusthe biases in low and high levels are somewhat connectedIncreasing resolution can improve the results in the centralregion of SA due to better representation of orography in thisregion in a high-resolution simulation The highest spatialcorrelation for the zonal wind anomalies at 850 hPa is foundin the southern SA (088) The geopotential height is thevariable that exhibits the highest values of RMSE and bias(mostly greater than 1) for all regions (Table 2) The lowervalues of RMSE and bias are found in the northern andsouthern SA but the spatial correlation is weaker in thenorthern and stronger in the southern

Figure 9 shows the annual mean temperature anomaliesfor HCSA strong cases The reanalysis data presents signif-icant positive anomalies on a great part of northern SA andequatorial Atlantic whereas the negative anomalies are foundover the east Pacific around 20∘S (Figure 9(a)) The model

shows significant positive anomalies only over the northeastedge of SA and over the equatorial Atlantic (Figure 9(b))Indeed themodel underestimates the temperature anomaliesin the northern region as indicated by the bias value in thisregion (minus011 Table 2) A positive (negative) bias in surfaceair temperature is associated with a negative (positive) biasin precipitation as an underestimation (overestimation) ofprecipitation is usually associated with an underestimation(overestimation) of clouds which enhances (reduces) thenet short-wave radiation budget at surface and consequentlyenhances (reduces) the net energy budget at surface [36]Indeed the model exhibits a positive bias (019) for theprecipitation anomalies in the northern region The lowest(highest) value of the RMSE (correlation) for the annualmean temperature anomalies is found in the southern regionindicating a good agreement between the model and obser-vation

Figure 10 displays the correlation between the HCSAannual intensity index and the annual mean temperaturein the period 1996ndash2011 As in Figure 8(a) three signifi-cant regions (northern central and southern SA) are seenHowever the signs of the correlation coefficients in thesethree regions of Figure 10 are opposite in relation to Figure 8Thus positive (negative) correlations of the annual meantemperature (zonal wind) with the HCSA intensity are seenover the northern and southern SA in the 1996ndash2011 periodThe opposite signal between the annual mean temperatureand the zonal wind is also seen over the central regions of

Advances in Meteorology 9

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus06

minus06

minus06

minus04

minus04

minus04

minus04

minus02

minus02

minus020

0

0

02

02

02

02

04

04

0404

06

06

08

minus020

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

02

02 02

02

02

02

02

04

04

04

06

08

0

0

0

0

0

0

0

0

0

Latit

ude

Longitude

minus06

minus06

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02minus02

minus02

minus02

minus02

04

04

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

minus02

minus02

minus02

minus02

minus02

minus08

minus08

minus04

minus04

minus04

minus04

minus06

minus06

minus06

0202

02

02

040

0

0

0

0

Latit

ude

Longitude

08

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04minus04

minus04

minus04

minus04

minus06

02

02

02

02

02

02

0204

04

04

04

04

04

06 06

08

08

08

0

00

0

0

0

0

Latit

ude

Longitude

06

minus02

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 8 Correlation between the HCSA annual intensity index and the annual mean zonal wind (m sminus1) in the period 1996ndash2011 based onthe Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervals are 02and values above 05 are shaded and are statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSAannual intensity index time series have been multiplied by (minus1)

10 Advances in Meteorology

02

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus05

minus04

minus03

minus03

minus03

minus02

minus02

minus01

minus01

minus01

minus01

0

0

0

0

01

01

01

01

02

02

02

02

02

02

02

02

02

03

03

03

03

03

04

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

minus02

minus02

minus02

minus02

minus02

minus01

minus01

minus01

minus01

minus01

minus01

minus01

00

0

0

0

0

0

01

01

01

01

01

02

02

02

02

02

03

03

04

05

03

03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 9 Annual mean temperature (K) anomaly at 850 hPa for HCSA strong cases in the period 1996ndash2011 based on the (a) Era-Interimreanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level from two-tailed Studentrsquos119905-test are shaded

minus06minus06

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

02

02

02

02

02

04

04

04

04

04

06

06

06

06

06

Latit

ude

Longitude

0

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

04

04

04

02

02

02

02

04

04

06

0606

08

08

00

0

0

0

0

Latit

ude

Longitude

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 10 Correlation between the HCSA annual intensity index and the annual mean temperature (K) at 850 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSA annual intensity index has beenmultiplied by (minus1)

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 5: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Advances in Meteorology 5

05

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(a)

05

1000

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

LatitudeEQ45

∘S 40∘S 35

∘S 30∘S 25

∘S 20∘S 15

∘S 10∘S 5

∘S 5∘N 10

∘N

(b)

Figure 3 Annual mean of the HCSA anomaly based on the Era-Interim reanalysis for (a) strong cases and (b) poleward expansion casesThe vertical velocity has been multiplied by minus100 The color scale represents the vectorrsquos magnitude

minus3

minus2

minus1

1996 1998 2000 2002 2004 2006 2008 2010

HCA

O in

tens

ity in

dex

Year

0

1

2

(a)

minus2

minus1

0

1

2

1996 1998 2000 2002 2004 2006 2008 2010

Year

HCA

O p

olew

ard

edge

(b)

Figure 4 Normalized time series for the annual mean HCAO (a) intensity index and (b) poleward edge in the period 1996ndash2011 with thelinear trend (red dotted line)

The performance of the regional climate models variesgreatly depending on the combination of physical parame-terization schemes simulation region boundary conditionsand the horizontal resolution [18 20] The skill of theRegCM4 over different regions covering the South Americaand Atlantic Ocean (Figure 5) during the 1996ndash2011 periodis evaluated quantitatively by analyzing the root mean squareerror (RMSE) bias and spatial correlation coefficient of thesimulation in relation to observations for the annual meananomalies We selected the regions in which the impactsof the HCSA intensification and poleward expansion aremore prominent The RMSE measures the overall magnitudeof deviations regardless of their individual signs while thebias measures systematic or predominant deviations in givendirections [35] The closer the values of RMSE and biasget to zero the better the agreement between observed andsimulated data is The RMSE is calculated according to (2)where 1199091015840119872 and 1199091015840119874 are the anomalies of the model and theobservations respectivelyThe summation is carried out over

119873 grid points in a predefined regionThe bias and the spatialcorrelation coefficient are calculated according to (3) and (4)respectively Consider

RMSE = radic 1119873

119873

sum

119894=1

(119909

1015840119872

119894

minus 119909

1015840119874

119894

)

2 (2)

bias = 1119873

119873

sum

119894=1

(119909

1015840119872

119894

minus 119909

1015840119874

119894

) (3)

120588 =

sum

119873

119894=1

(119909

1015840119872

119894

minus 119909

1015840119872

) (119909

1015840119874

119894

minus 119909

1015840119874

)

radic

sum

119873

119894=1

(119909

1015840119872

119894

minus 119909

1015840119872

)

2

radicsum

119873

119894=1

(119909

1015840119874

119894

minus 119909

1015840119874

)

2

(4)

3 Impacts of the HCSA Intensification

The vertical velocity and the divergent winds are the mainfields analyzed when the regional HC is studied However

6 Advances in Meteorology

Table 2 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSAstrong cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 042 031 152 026 012 012 052 minus010 078 035 minus015 0112 091 018 329 023 minus067 002 318 004 068 050 087 0403 051 019 167 011 039 013 155 006 083 088 098 0854 065 025 313 019 minus011 011 225 minus001 084 073 091 055

Latit

ude

Longitude

1

2

3

4

5

6

7

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

Figure 5 South America regions selected for statistical evaluationof model performance (1) northern (80∘Wndash35∘W 15∘Sndash10∘N) (2)central (80∘Wndash35∘W 35∘Sndash15∘S) (3) southern (80∘Wndash55∘W 55∘Sndash35∘S) (4) entire domain (80∘Wndash10∘W 55∘Sndash10∘N) (5) southeast(70∘Wndash50∘W 40∘Sndash30∘S) (6) northeast (60∘Wndash35∘W 10∘Sndash5∘N)and (7) all South America (80∘Wndash35∘W 55∘Sndash10∘S)

there is a remarkable impact of the changes in the HC onthe horizontal winds at low and high troposphere as they arelinked to the vertical movement of the air Indeed the tradewinds are the expression of the HC at surface

Figure 6 shows the anomalies of the annual mean hor-izontal wind at 200 and 850 hPa for the HCSA strongcases based on the Era-Interim reanalysis and the modelsimulation Strong anomalous easterly winds coming fromthe Atlantic Ocean are seen over the northern South America(SA) at 200 hPa in the reanalysis (Figure 6(a)) and especiallyin model simulation (Figure 6(c)) which exhibits a positivebias (012) for the zonal wind in this region (Table 2) Whenentering the mainland these winds rotate counterclockwiseStrong anomalous easterly winds coming from the Atlanticalso hits the southeast of the continent and then they rotate

clockwise resulting in an anomalous trough which is seenat 200 hPa (Figure not shown) and 500 hPa (Figure 7) Theanomalous wind pattern at 850 hPa is characterized by thestrengthening of the trade winds around the equator whichspin in northeast direction before entering the northern SAand by a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 6(b)) and model simu-lation (Figure 6(d))This counterclockwise circulation resultsin a significant anomalous ridge which is seen at 850 hPa(Figure not shown) and 500 hPa in the reanalysis and modelsimulation (Figure 7)

To investigate the quantitative link between the HCSAintensity and the annual changes in the zonal wind (strongercomponent of the horizontal wind) correlation coefficientsbetween the HCSA annual intensity index and the annualmean zonal wind at 200 and 850 hPa are displayed in Figure 8The model simulation (Figure 8(c)) is able to reproduce thesigns (positives and negatives) of the correlation coefficientsobserved in the reanalysis (Figure 8(a)) Three regions withsignificant correlations (above 05) in the high troposphereare seen in the reanalysis (Figure 8(a)) Figures 6(a) and 6(c)showed that for HCSA strong cases the zonal component ofthe anomalous wind at 200 hPa is easterly (negative) in thenorthern and southern SA regions whereas in the centralarea of SA it is westerly (positive) In anotherway Figures 8(a)and 8(c) show that the anomalous easterly (westerly) windscoming from (towards) the Atlantic presents negative (pos-itive) correlations with the HCSA intensity At 850 hPasignificant positive correlations are seen in the reanalysison the edge of northwest SA and negative correlations onthe central-south area of the continent (Figure 8(b)) Themodel simulation shows significant positive correlations ona great part of the northwest SA (Figure 8(d)) probably dueto the misrepresentation of the horizontal wind in this region(Figure 6(d)) At 850 hPa it can also be seen that regions withanomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA intensity

The three regions (northern central and southern) withsignificant correlations seen in Figure 8(a) and the entiredomain considered in the composite analyses were selectedfor evaluation of model performance in reproducing theanomalies of zonal wind geopotential height and temper-ature (Figure 5) Table 2 shows that for the annual meanzonal wind anomalies at 200 hPa lower values of RMSE andbias (absolute value) and higher values of spatial correlation

Advances in Meteorology 7La

titud

e

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 6 Annual mean horizontal wind (m sminus1) anomaly for HCSA strong cases in the period 1996ndash2011 based on the Era-Interim reanalysisat (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents the vectorrsquos magnitude

8 Advances in MeteorologyLa

titud

e

Longitude

18

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

0

0

0

0

minus4

minus4

minus4

minus4

minus2

minus2

minus2

minus2

2

4

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus2

minus2

0

0

0

2

2

2

2

1820

1614

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 7 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

are found mainly in the northern and southern SA andwhen we consider the entire domain (region 4 of Figure 5)Thus at 200 hPa the central region of SA exhibits weakeragreement between the model and observation for the zonalwind anomalies Table 2 shows that themodel underestimatesthe zonal wind anomalies in this region (bias = minus067) Aswell Figures 6(c) and 7(b) show that the model simulatesa strong anticyclonic anomalous circulation in the central-western region Interestingly at 850 hPa the central region ofSA exhibits the lowest value of RMSE and bias (018 and 002resp) but the spatial correlation is only 05 Mid- and upper-tropospheric winds influence low-level circulations thusthe biases in low and high levels are somewhat connectedIncreasing resolution can improve the results in the centralregion of SA due to better representation of orography in thisregion in a high-resolution simulation The highest spatialcorrelation for the zonal wind anomalies at 850 hPa is foundin the southern SA (088) The geopotential height is thevariable that exhibits the highest values of RMSE and bias(mostly greater than 1) for all regions (Table 2) The lowervalues of RMSE and bias are found in the northern andsouthern SA but the spatial correlation is weaker in thenorthern and stronger in the southern

Figure 9 shows the annual mean temperature anomaliesfor HCSA strong cases The reanalysis data presents signif-icant positive anomalies on a great part of northern SA andequatorial Atlantic whereas the negative anomalies are foundover the east Pacific around 20∘S (Figure 9(a)) The model

shows significant positive anomalies only over the northeastedge of SA and over the equatorial Atlantic (Figure 9(b))Indeed themodel underestimates the temperature anomaliesin the northern region as indicated by the bias value in thisregion (minus011 Table 2) A positive (negative) bias in surfaceair temperature is associated with a negative (positive) biasin precipitation as an underestimation (overestimation) ofprecipitation is usually associated with an underestimation(overestimation) of clouds which enhances (reduces) thenet short-wave radiation budget at surface and consequentlyenhances (reduces) the net energy budget at surface [36]Indeed the model exhibits a positive bias (019) for theprecipitation anomalies in the northern region The lowest(highest) value of the RMSE (correlation) for the annualmean temperature anomalies is found in the southern regionindicating a good agreement between the model and obser-vation

Figure 10 displays the correlation between the HCSAannual intensity index and the annual mean temperaturein the period 1996ndash2011 As in Figure 8(a) three signifi-cant regions (northern central and southern SA) are seenHowever the signs of the correlation coefficients in thesethree regions of Figure 10 are opposite in relation to Figure 8Thus positive (negative) correlations of the annual meantemperature (zonal wind) with the HCSA intensity are seenover the northern and southern SA in the 1996ndash2011 periodThe opposite signal between the annual mean temperatureand the zonal wind is also seen over the central regions of

Advances in Meteorology 9

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus06

minus06

minus06

minus04

minus04

minus04

minus04

minus02

minus02

minus020

0

0

02

02

02

02

04

04

0404

06

06

08

minus020

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

02

02 02

02

02

02

02

04

04

04

06

08

0

0

0

0

0

0

0

0

0

Latit

ude

Longitude

minus06

minus06

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02minus02

minus02

minus02

minus02

04

04

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

minus02

minus02

minus02

minus02

minus02

minus08

minus08

minus04

minus04

minus04

minus04

minus06

minus06

minus06

0202

02

02

040

0

0

0

0

Latit

ude

Longitude

08

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04minus04

minus04

minus04

minus04

minus06

02

02

02

02

02

02

0204

04

04

04

04

04

06 06

08

08

08

0

00

0

0

0

0

Latit

ude

Longitude

06

minus02

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 8 Correlation between the HCSA annual intensity index and the annual mean zonal wind (m sminus1) in the period 1996ndash2011 based onthe Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervals are 02and values above 05 are shaded and are statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSAannual intensity index time series have been multiplied by (minus1)

10 Advances in Meteorology

02

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus05

minus04

minus03

minus03

minus03

minus02

minus02

minus01

minus01

minus01

minus01

0

0

0

0

01

01

01

01

02

02

02

02

02

02

02

02

02

03

03

03

03

03

04

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

minus02

minus02

minus02

minus02

minus02

minus01

minus01

minus01

minus01

minus01

minus01

minus01

00

0

0

0

0

0

01

01

01

01

01

02

02

02

02

02

03

03

04

05

03

03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 9 Annual mean temperature (K) anomaly at 850 hPa for HCSA strong cases in the period 1996ndash2011 based on the (a) Era-Interimreanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level from two-tailed Studentrsquos119905-test are shaded

minus06minus06

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

02

02

02

02

02

04

04

04

04

04

06

06

06

06

06

Latit

ude

Longitude

0

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

04

04

04

02

02

02

02

04

04

06

0606

08

08

00

0

0

0

0

Latit

ude

Longitude

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 10 Correlation between the HCSA annual intensity index and the annual mean temperature (K) at 850 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSA annual intensity index has beenmultiplied by (minus1)

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 6: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

6 Advances in Meteorology

Table 2 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSAstrong cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 042 031 152 026 012 012 052 minus010 078 035 minus015 0112 091 018 329 023 minus067 002 318 004 068 050 087 0403 051 019 167 011 039 013 155 006 083 088 098 0854 065 025 313 019 minus011 011 225 minus001 084 073 091 055

Latit

ude

Longitude

1

2

3

4

5

6

7

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

Figure 5 South America regions selected for statistical evaluationof model performance (1) northern (80∘Wndash35∘W 15∘Sndash10∘N) (2)central (80∘Wndash35∘W 35∘Sndash15∘S) (3) southern (80∘Wndash55∘W 55∘Sndash35∘S) (4) entire domain (80∘Wndash10∘W 55∘Sndash10∘N) (5) southeast(70∘Wndash50∘W 40∘Sndash30∘S) (6) northeast (60∘Wndash35∘W 10∘Sndash5∘N)and (7) all South America (80∘Wndash35∘W 55∘Sndash10∘S)

there is a remarkable impact of the changes in the HC onthe horizontal winds at low and high troposphere as they arelinked to the vertical movement of the air Indeed the tradewinds are the expression of the HC at surface

Figure 6 shows the anomalies of the annual mean hor-izontal wind at 200 and 850 hPa for the HCSA strongcases based on the Era-Interim reanalysis and the modelsimulation Strong anomalous easterly winds coming fromthe Atlantic Ocean are seen over the northern South America(SA) at 200 hPa in the reanalysis (Figure 6(a)) and especiallyin model simulation (Figure 6(c)) which exhibits a positivebias (012) for the zonal wind in this region (Table 2) Whenentering the mainland these winds rotate counterclockwiseStrong anomalous easterly winds coming from the Atlanticalso hits the southeast of the continent and then they rotate

clockwise resulting in an anomalous trough which is seenat 200 hPa (Figure not shown) and 500 hPa (Figure 7) Theanomalous wind pattern at 850 hPa is characterized by thestrengthening of the trade winds around the equator whichspin in northeast direction before entering the northern SAand by a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 6(b)) and model simu-lation (Figure 6(d))This counterclockwise circulation resultsin a significant anomalous ridge which is seen at 850 hPa(Figure not shown) and 500 hPa in the reanalysis and modelsimulation (Figure 7)

To investigate the quantitative link between the HCSAintensity and the annual changes in the zonal wind (strongercomponent of the horizontal wind) correlation coefficientsbetween the HCSA annual intensity index and the annualmean zonal wind at 200 and 850 hPa are displayed in Figure 8The model simulation (Figure 8(c)) is able to reproduce thesigns (positives and negatives) of the correlation coefficientsobserved in the reanalysis (Figure 8(a)) Three regions withsignificant correlations (above 05) in the high troposphereare seen in the reanalysis (Figure 8(a)) Figures 6(a) and 6(c)showed that for HCSA strong cases the zonal component ofthe anomalous wind at 200 hPa is easterly (negative) in thenorthern and southern SA regions whereas in the centralarea of SA it is westerly (positive) In anotherway Figures 8(a)and 8(c) show that the anomalous easterly (westerly) windscoming from (towards) the Atlantic presents negative (pos-itive) correlations with the HCSA intensity At 850 hPasignificant positive correlations are seen in the reanalysison the edge of northwest SA and negative correlations onthe central-south area of the continent (Figure 8(b)) Themodel simulation shows significant positive correlations ona great part of the northwest SA (Figure 8(d)) probably dueto the misrepresentation of the horizontal wind in this region(Figure 6(d)) At 850 hPa it can also be seen that regions withanomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA intensity

The three regions (northern central and southern) withsignificant correlations seen in Figure 8(a) and the entiredomain considered in the composite analyses were selectedfor evaluation of model performance in reproducing theanomalies of zonal wind geopotential height and temper-ature (Figure 5) Table 2 shows that for the annual meanzonal wind anomalies at 200 hPa lower values of RMSE andbias (absolute value) and higher values of spatial correlation

Advances in Meteorology 7La

titud

e

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 6 Annual mean horizontal wind (m sminus1) anomaly for HCSA strong cases in the period 1996ndash2011 based on the Era-Interim reanalysisat (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents the vectorrsquos magnitude

8 Advances in MeteorologyLa

titud

e

Longitude

18

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

0

0

0

0

minus4

minus4

minus4

minus4

minus2

minus2

minus2

minus2

2

4

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus2

minus2

0

0

0

2

2

2

2

1820

1614

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 7 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

are found mainly in the northern and southern SA andwhen we consider the entire domain (region 4 of Figure 5)Thus at 200 hPa the central region of SA exhibits weakeragreement between the model and observation for the zonalwind anomalies Table 2 shows that themodel underestimatesthe zonal wind anomalies in this region (bias = minus067) Aswell Figures 6(c) and 7(b) show that the model simulatesa strong anticyclonic anomalous circulation in the central-western region Interestingly at 850 hPa the central region ofSA exhibits the lowest value of RMSE and bias (018 and 002resp) but the spatial correlation is only 05 Mid- and upper-tropospheric winds influence low-level circulations thusthe biases in low and high levels are somewhat connectedIncreasing resolution can improve the results in the centralregion of SA due to better representation of orography in thisregion in a high-resolution simulation The highest spatialcorrelation for the zonal wind anomalies at 850 hPa is foundin the southern SA (088) The geopotential height is thevariable that exhibits the highest values of RMSE and bias(mostly greater than 1) for all regions (Table 2) The lowervalues of RMSE and bias are found in the northern andsouthern SA but the spatial correlation is weaker in thenorthern and stronger in the southern

Figure 9 shows the annual mean temperature anomaliesfor HCSA strong cases The reanalysis data presents signif-icant positive anomalies on a great part of northern SA andequatorial Atlantic whereas the negative anomalies are foundover the east Pacific around 20∘S (Figure 9(a)) The model

shows significant positive anomalies only over the northeastedge of SA and over the equatorial Atlantic (Figure 9(b))Indeed themodel underestimates the temperature anomaliesin the northern region as indicated by the bias value in thisregion (minus011 Table 2) A positive (negative) bias in surfaceair temperature is associated with a negative (positive) biasin precipitation as an underestimation (overestimation) ofprecipitation is usually associated with an underestimation(overestimation) of clouds which enhances (reduces) thenet short-wave radiation budget at surface and consequentlyenhances (reduces) the net energy budget at surface [36]Indeed the model exhibits a positive bias (019) for theprecipitation anomalies in the northern region The lowest(highest) value of the RMSE (correlation) for the annualmean temperature anomalies is found in the southern regionindicating a good agreement between the model and obser-vation

Figure 10 displays the correlation between the HCSAannual intensity index and the annual mean temperaturein the period 1996ndash2011 As in Figure 8(a) three signifi-cant regions (northern central and southern SA) are seenHowever the signs of the correlation coefficients in thesethree regions of Figure 10 are opposite in relation to Figure 8Thus positive (negative) correlations of the annual meantemperature (zonal wind) with the HCSA intensity are seenover the northern and southern SA in the 1996ndash2011 periodThe opposite signal between the annual mean temperatureand the zonal wind is also seen over the central regions of

Advances in Meteorology 9

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus06

minus06

minus06

minus04

minus04

minus04

minus04

minus02

minus02

minus020

0

0

02

02

02

02

04

04

0404

06

06

08

minus020

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

02

02 02

02

02

02

02

04

04

04

06

08

0

0

0

0

0

0

0

0

0

Latit

ude

Longitude

minus06

minus06

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02minus02

minus02

minus02

minus02

04

04

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

minus02

minus02

minus02

minus02

minus02

minus08

minus08

minus04

minus04

minus04

minus04

minus06

minus06

minus06

0202

02

02

040

0

0

0

0

Latit

ude

Longitude

08

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04minus04

minus04

minus04

minus04

minus06

02

02

02

02

02

02

0204

04

04

04

04

04

06 06

08

08

08

0

00

0

0

0

0

Latit

ude

Longitude

06

minus02

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 8 Correlation between the HCSA annual intensity index and the annual mean zonal wind (m sminus1) in the period 1996ndash2011 based onthe Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervals are 02and values above 05 are shaded and are statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSAannual intensity index time series have been multiplied by (minus1)

10 Advances in Meteorology

02

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus05

minus04

minus03

minus03

minus03

minus02

minus02

minus01

minus01

minus01

minus01

0

0

0

0

01

01

01

01

02

02

02

02

02

02

02

02

02

03

03

03

03

03

04

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

minus02

minus02

minus02

minus02

minus02

minus01

minus01

minus01

minus01

minus01

minus01

minus01

00

0

0

0

0

0

01

01

01

01

01

02

02

02

02

02

03

03

04

05

03

03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 9 Annual mean temperature (K) anomaly at 850 hPa for HCSA strong cases in the period 1996ndash2011 based on the (a) Era-Interimreanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level from two-tailed Studentrsquos119905-test are shaded

minus06minus06

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

02

02

02

02

02

04

04

04

04

04

06

06

06

06

06

Latit

ude

Longitude

0

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

04

04

04

02

02

02

02

04

04

06

0606

08

08

00

0

0

0

0

Latit

ude

Longitude

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 10 Correlation between the HCSA annual intensity index and the annual mean temperature (K) at 850 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSA annual intensity index has beenmultiplied by (minus1)

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 7: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Advances in Meteorology 7La

titud

e

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 6 Annual mean horizontal wind (m sminus1) anomaly for HCSA strong cases in the period 1996ndash2011 based on the Era-Interim reanalysisat (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents the vectorrsquos magnitude

8 Advances in MeteorologyLa

titud

e

Longitude

18

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

0

0

0

0

minus4

minus4

minus4

minus4

minus2

minus2

minus2

minus2

2

4

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus2

minus2

0

0

0

2

2

2

2

1820

1614

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 7 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

are found mainly in the northern and southern SA andwhen we consider the entire domain (region 4 of Figure 5)Thus at 200 hPa the central region of SA exhibits weakeragreement between the model and observation for the zonalwind anomalies Table 2 shows that themodel underestimatesthe zonal wind anomalies in this region (bias = minus067) Aswell Figures 6(c) and 7(b) show that the model simulatesa strong anticyclonic anomalous circulation in the central-western region Interestingly at 850 hPa the central region ofSA exhibits the lowest value of RMSE and bias (018 and 002resp) but the spatial correlation is only 05 Mid- and upper-tropospheric winds influence low-level circulations thusthe biases in low and high levels are somewhat connectedIncreasing resolution can improve the results in the centralregion of SA due to better representation of orography in thisregion in a high-resolution simulation The highest spatialcorrelation for the zonal wind anomalies at 850 hPa is foundin the southern SA (088) The geopotential height is thevariable that exhibits the highest values of RMSE and bias(mostly greater than 1) for all regions (Table 2) The lowervalues of RMSE and bias are found in the northern andsouthern SA but the spatial correlation is weaker in thenorthern and stronger in the southern

Figure 9 shows the annual mean temperature anomaliesfor HCSA strong cases The reanalysis data presents signif-icant positive anomalies on a great part of northern SA andequatorial Atlantic whereas the negative anomalies are foundover the east Pacific around 20∘S (Figure 9(a)) The model

shows significant positive anomalies only over the northeastedge of SA and over the equatorial Atlantic (Figure 9(b))Indeed themodel underestimates the temperature anomaliesin the northern region as indicated by the bias value in thisregion (minus011 Table 2) A positive (negative) bias in surfaceair temperature is associated with a negative (positive) biasin precipitation as an underestimation (overestimation) ofprecipitation is usually associated with an underestimation(overestimation) of clouds which enhances (reduces) thenet short-wave radiation budget at surface and consequentlyenhances (reduces) the net energy budget at surface [36]Indeed the model exhibits a positive bias (019) for theprecipitation anomalies in the northern region The lowest(highest) value of the RMSE (correlation) for the annualmean temperature anomalies is found in the southern regionindicating a good agreement between the model and obser-vation

Figure 10 displays the correlation between the HCSAannual intensity index and the annual mean temperaturein the period 1996ndash2011 As in Figure 8(a) three signifi-cant regions (northern central and southern SA) are seenHowever the signs of the correlation coefficients in thesethree regions of Figure 10 are opposite in relation to Figure 8Thus positive (negative) correlations of the annual meantemperature (zonal wind) with the HCSA intensity are seenover the northern and southern SA in the 1996ndash2011 periodThe opposite signal between the annual mean temperatureand the zonal wind is also seen over the central regions of

Advances in Meteorology 9

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus06

minus06

minus06

minus04

minus04

minus04

minus04

minus02

minus02

minus020

0

0

02

02

02

02

04

04

0404

06

06

08

minus020

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

02

02 02

02

02

02

02

04

04

04

06

08

0

0

0

0

0

0

0

0

0

Latit

ude

Longitude

minus06

minus06

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02minus02

minus02

minus02

minus02

04

04

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

minus02

minus02

minus02

minus02

minus02

minus08

minus08

minus04

minus04

minus04

minus04

minus06

minus06

minus06

0202

02

02

040

0

0

0

0

Latit

ude

Longitude

08

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04minus04

minus04

minus04

minus04

minus06

02

02

02

02

02

02

0204

04

04

04

04

04

06 06

08

08

08

0

00

0

0

0

0

Latit

ude

Longitude

06

minus02

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 8 Correlation between the HCSA annual intensity index and the annual mean zonal wind (m sminus1) in the period 1996ndash2011 based onthe Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervals are 02and values above 05 are shaded and are statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSAannual intensity index time series have been multiplied by (minus1)

10 Advances in Meteorology

02

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus05

minus04

minus03

minus03

minus03

minus02

minus02

minus01

minus01

minus01

minus01

0

0

0

0

01

01

01

01

02

02

02

02

02

02

02

02

02

03

03

03

03

03

04

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

minus02

minus02

minus02

minus02

minus02

minus01

minus01

minus01

minus01

minus01

minus01

minus01

00

0

0

0

0

0

01

01

01

01

01

02

02

02

02

02

03

03

04

05

03

03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 9 Annual mean temperature (K) anomaly at 850 hPa for HCSA strong cases in the period 1996ndash2011 based on the (a) Era-Interimreanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level from two-tailed Studentrsquos119905-test are shaded

minus06minus06

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

02

02

02

02

02

04

04

04

04

04

06

06

06

06

06

Latit

ude

Longitude

0

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

04

04

04

02

02

02

02

04

04

06

0606

08

08

00

0

0

0

0

Latit

ude

Longitude

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 10 Correlation between the HCSA annual intensity index and the annual mean temperature (K) at 850 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSA annual intensity index has beenmultiplied by (minus1)

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 8: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

8 Advances in MeteorologyLa

titud

e

Longitude

18

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

0

0

0

0

minus4

minus4

minus4

minus4

minus2

minus2

minus2

minus2

2

4

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus2

minus2

0

0

0

2

2

2

2

1820

1614

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 7 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

are found mainly in the northern and southern SA andwhen we consider the entire domain (region 4 of Figure 5)Thus at 200 hPa the central region of SA exhibits weakeragreement between the model and observation for the zonalwind anomalies Table 2 shows that themodel underestimatesthe zonal wind anomalies in this region (bias = minus067) Aswell Figures 6(c) and 7(b) show that the model simulatesa strong anticyclonic anomalous circulation in the central-western region Interestingly at 850 hPa the central region ofSA exhibits the lowest value of RMSE and bias (018 and 002resp) but the spatial correlation is only 05 Mid- and upper-tropospheric winds influence low-level circulations thusthe biases in low and high levels are somewhat connectedIncreasing resolution can improve the results in the centralregion of SA due to better representation of orography in thisregion in a high-resolution simulation The highest spatialcorrelation for the zonal wind anomalies at 850 hPa is foundin the southern SA (088) The geopotential height is thevariable that exhibits the highest values of RMSE and bias(mostly greater than 1) for all regions (Table 2) The lowervalues of RMSE and bias are found in the northern andsouthern SA but the spatial correlation is weaker in thenorthern and stronger in the southern

Figure 9 shows the annual mean temperature anomaliesfor HCSA strong cases The reanalysis data presents signif-icant positive anomalies on a great part of northern SA andequatorial Atlantic whereas the negative anomalies are foundover the east Pacific around 20∘S (Figure 9(a)) The model

shows significant positive anomalies only over the northeastedge of SA and over the equatorial Atlantic (Figure 9(b))Indeed themodel underestimates the temperature anomaliesin the northern region as indicated by the bias value in thisregion (minus011 Table 2) A positive (negative) bias in surfaceair temperature is associated with a negative (positive) biasin precipitation as an underestimation (overestimation) ofprecipitation is usually associated with an underestimation(overestimation) of clouds which enhances (reduces) thenet short-wave radiation budget at surface and consequentlyenhances (reduces) the net energy budget at surface [36]Indeed the model exhibits a positive bias (019) for theprecipitation anomalies in the northern region The lowest(highest) value of the RMSE (correlation) for the annualmean temperature anomalies is found in the southern regionindicating a good agreement between the model and obser-vation

Figure 10 displays the correlation between the HCSAannual intensity index and the annual mean temperaturein the period 1996ndash2011 As in Figure 8(a) three signifi-cant regions (northern central and southern SA) are seenHowever the signs of the correlation coefficients in thesethree regions of Figure 10 are opposite in relation to Figure 8Thus positive (negative) correlations of the annual meantemperature (zonal wind) with the HCSA intensity are seenover the northern and southern SA in the 1996ndash2011 periodThe opposite signal between the annual mean temperatureand the zonal wind is also seen over the central regions of

Advances in Meteorology 9

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus06

minus06

minus06

minus04

minus04

minus04

minus04

minus02

minus02

minus020

0

0

02

02

02

02

04

04

0404

06

06

08

minus020

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

02

02 02

02

02

02

02

04

04

04

06

08

0

0

0

0

0

0

0

0

0

Latit

ude

Longitude

minus06

minus06

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02minus02

minus02

minus02

minus02

04

04

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

minus02

minus02

minus02

minus02

minus02

minus08

minus08

minus04

minus04

minus04

minus04

minus06

minus06

minus06

0202

02

02

040

0

0

0

0

Latit

ude

Longitude

08

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04minus04

minus04

minus04

minus04

minus06

02

02

02

02

02

02

0204

04

04

04

04

04

06 06

08

08

08

0

00

0

0

0

0

Latit

ude

Longitude

06

minus02

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 8 Correlation between the HCSA annual intensity index and the annual mean zonal wind (m sminus1) in the period 1996ndash2011 based onthe Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervals are 02and values above 05 are shaded and are statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSAannual intensity index time series have been multiplied by (minus1)

10 Advances in Meteorology

02

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus05

minus04

minus03

minus03

minus03

minus02

minus02

minus01

minus01

minus01

minus01

0

0

0

0

01

01

01

01

02

02

02

02

02

02

02

02

02

03

03

03

03

03

04

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

minus02

minus02

minus02

minus02

minus02

minus01

minus01

minus01

minus01

minus01

minus01

minus01

00

0

0

0

0

0

01

01

01

01

01

02

02

02

02

02

03

03

04

05

03

03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 9 Annual mean temperature (K) anomaly at 850 hPa for HCSA strong cases in the period 1996ndash2011 based on the (a) Era-Interimreanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level from two-tailed Studentrsquos119905-test are shaded

minus06minus06

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

02

02

02

02

02

04

04

04

04

04

06

06

06

06

06

Latit

ude

Longitude

0

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

04

04

04

02

02

02

02

04

04

06

0606

08

08

00

0

0

0

0

Latit

ude

Longitude

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 10 Correlation between the HCSA annual intensity index and the annual mean temperature (K) at 850 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSA annual intensity index has beenmultiplied by (minus1)

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 9: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Advances in Meteorology 9

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus06

minus06

minus06

minus04

minus04

minus04

minus04

minus02

minus02

minus020

0

0

02

02

02

02

04

04

0404

06

06

08

minus020

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

02

02 02

02

02

02

02

04

04

04

06

08

0

0

0

0

0

0

0

0

0

Latit

ude

Longitude

minus06

minus06

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02minus02

minus02

minus02

minus02

04

04

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

minus02

minus02

minus02

minus02

minus02

minus08

minus08

minus04

minus04

minus04

minus04

minus06

minus06

minus06

0202

02

02

040

0

0

0

0

Latit

ude

Longitude

08

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04minus04

minus04

minus04

minus04

minus06

02

02

02

02

02

02

0204

04

04

04

04

04

06 06

08

08

08

0

00

0

0

0

0

Latit

ude

Longitude

06

minus02

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 8 Correlation between the HCSA annual intensity index and the annual mean zonal wind (m sminus1) in the period 1996ndash2011 based onthe Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervals are 02and values above 05 are shaded and are statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSAannual intensity index time series have been multiplied by (minus1)

10 Advances in Meteorology

02

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus05

minus04

minus03

minus03

minus03

minus02

minus02

minus01

minus01

minus01

minus01

0

0

0

0

01

01

01

01

02

02

02

02

02

02

02

02

02

03

03

03

03

03

04

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

minus02

minus02

minus02

minus02

minus02

minus01

minus01

minus01

minus01

minus01

minus01

minus01

00

0

0

0

0

0

01

01

01

01

01

02

02

02

02

02

03

03

04

05

03

03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 9 Annual mean temperature (K) anomaly at 850 hPa for HCSA strong cases in the period 1996ndash2011 based on the (a) Era-Interimreanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level from two-tailed Studentrsquos119905-test are shaded

minus06minus06

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

02

02

02

02

02

04

04

04

04

04

06

06

06

06

06

Latit

ude

Longitude

0

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

04

04

04

02

02

02

02

04

04

06

0606

08

08

00

0

0

0

0

Latit

ude

Longitude

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 10 Correlation between the HCSA annual intensity index and the annual mean temperature (K) at 850 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSA annual intensity index has beenmultiplied by (minus1)

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 10: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

10 Advances in Meteorology

02

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus06

minus05

minus04

minus03

minus03

minus03

minus02

minus02

minus01

minus01

minus01

minus01

0

0

0

0

01

01

01

01

02

02

02

02

02

02

02

02

02

03

03

03

03

03

04

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

minus02

minus02

minus02

minus02

minus02

minus01

minus01

minus01

minus01

minus01

minus01

minus01

00

0

0

0

0

0

01

01

01

01

01

02

02

02

02

02

03

03

04

05

03

03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 9 Annual mean temperature (K) anomaly at 850 hPa for HCSA strong cases in the period 1996ndash2011 based on the (a) Era-Interimreanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level from two-tailed Studentrsquos119905-test are shaded

minus06minus06

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

02

02

02

02

02

04

04

04

04

04

06

06

06

06

06

Latit

ude

Longitude

0

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

04

04

04

02

02

02

02

04

04

06

0606

08

08

00

0

0

0

0

Latit

ude

Longitude

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 10 Correlation between the HCSA annual intensity index and the annual mean temperature (K) at 850 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level For clarity purposes it should be noted that the HCSA annual intensity index has beenmultiplied by (minus1)

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 11: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Advances in Meteorology 11

180 0

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

Longitude

Latit

ude

60∘N

30∘N

60∘S

30∘S

0

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 11 Annual mean sea surface temperature (K) anomaly forHCSA strong cases in the period 1996ndash2011 based on the HADISSTdataset Contour intervals are 01 K with stippling indicating regionsstatistically significant above the 95 confidence level as determinedby two-tailed Studentrsquos 119905-test

SA and South Atlantic (Figures 10 and 8) Then we can statethat a cold trough is observed in the central regions of SA andSouth Atlantic and a warm ridge is seen over the southern SAand South Atlantic around 50∘S when the HCSA strengthens(Figures 7 8 and 10)

The significant positive temperature anomalies seen on agreat part of northern SA (Figure 9(a)) indicates a strength-ening of the Amazon heat source which is linked to astrong HCSA ascending branch On the other hand thetemperature anomalies seen over the Atlantic Ocean andeastern Pacific (Figure 9) are closely related to the SSTanomalies for the HCSA strong cases (Figure 11) A La Ninapattern is observed in the equatorial Pacific region whereasover the Atlantic Ocean from 0 to 20∘N positive SSTanomalies are seen The HCSA strong cases are 2005 20082010 and 2011 years Weak La Nina episodes were observedduring 2005-2006 and 2008-2009 and moderate to strongLa Nina events were observed during 2010-2011 (histori-cal ENSO episodes (1950ndashpresent) httpwwwcpcnoaagovproductsanalysis monitoringensostuffensoyearsshtml)

The Pacific Walker circulation for the HCSA strong cases(Figure 12(b)) shows some anomalous features compared tothe climatology (Figure 12(a)) which are also seen during LaNina events [9] For instance around 180∘ intense subsidenceis seen whereas in the eastern Pacific (around 90∘W) thedownward motion is weaker (Figure 12(b)) Upward motionis observed in the western Pacific (around 130∘E) as also seenin the climatology (Figure 12(a))

The strong ascending branch of the Walker circulationover northern SA (around 70∘W) is seen in the clima-tology (Figure 12(a)) and also for the HCSA strong cases(Figure 12(b)) Anomalous upward motion is also seen alongthe Atlantic basin (strongest at higher levels) (Figure 12(b))due to positive SST anomalies from 0 to 20∘N (Figure 11)These positive SST anomalies centered at 15∘N are asso-ciated with the weakening of the northern trade winds(Figure 6(b)) which usually blow southwestward toward theequator On the other hand the trade winds on the equatorintensify (Figure 6(b)) These features seem to be related to a

Table 3 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA strong cases over the lastthree regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 011 0746 025 minus010 0637 052 014 034

ldquonew-typerdquo Atlantic Ninos events described by Richter et al[37] During ldquonormalrdquo Atlantic Ninos ocean temperatureswarm up right on the equator due to the weakening ofthe surface winds On the other hand during the ldquonew-typerdquo positive SST anomalies are found north of the equatordue to the weakening of the northern trade winds and thestrengthening of thewestward-directedwinds on the equatorWe can also see that the anomalous ascending branch of theHCSA is shifted northward (Figure 3(a)) compared to theclimatology (Figure 1(a)) due to these positive SST anomaliesfound north of the equator

The maximum of the annual total rainfall observed overSA during the period 1996ndash2011 occurs mainly over thenorthern region from 7∘N to 10∘S and over the south ofBrazil (figure not shown) The importance of the ENSOin modulating South American precipitation is well knownthrough its associated changes in atmospheric circulation[38] The SA rainfall distribution during La Nina events seenin the summer and autumn seasons is characterized by aboveaverage precipitation in the north and northeast part of SAand belownormal values in the southern part of the continent[9] Figure 13 shows that the annual anomalous rainfalldistribution during the HCSA strong cases analyzed followsapproximately the La Nina canonical impacts especially overthe south of Brazil andUruguay (negative rainfall anomalies)but also presents the influence of the positive SST anomaliesin Atlantic (ldquonew-typerdquo Atlantic Ninos) especially over theedge of northeast SA (mostly negative rainfall anomalies)

Thus the main regions of SA affected by the precipitationanomalies which were selected for model evaluation are thesoutheast which encompasses the south of Brazil Uruguayand Argentina (region 5 of Figure 5) and the northeast(region 6 of Figure 5) Figure 13 and Table 3 show that themodel overestimates the negative rainfall anomalies (bias =011) over the southeast SA However this region presentsthe highest correlation between the model and observation(074) On the other hand Figure 13 and Table 3 show that therainfall anomalies observed in the northeast SA are underes-timated by themodel (bias = minus010)Whenwe consider all SA(region 7 of Figure 5) it is possible to note in Table 3 a weakeragreement between themodel and observation for the precip-itation anomalies (higher values of RMSE and absolute biasand lower value of correlation) Particularly Figure 13 showsa weaker agreement between the model and observation inthe region of Andes cordillera indicating an overestimationof the precipitation which is also documented in severalstudies (eg [39 40]) The complex topography of theAndes with a high degree of convective precipitation provides

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 12: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

12 Advances in Meteorology

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

5 Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(a)

1

925

850

700

500

400

300

250200

150

100

Pres

sure

(hPa

)

1000

Longitude

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(b)

1 Longitude

925

850

700

500

400

300

250

200

150

100

Pres

sure

(hPa

)

1000

120∘E

140∘E

160∘E

180

160∘W

140∘W

120∘W

100∘W

80∘W

60∘W

40∘W

20∘W 0

(c)

Figure 12 Walker circulation by averaging divergent wind and vertical velocity in the latitudinal band between 5∘S and 5∘N in the period1996ndash2011 based on the Era-Interim reanalysis for (a) climatological annual mean and anomalies for (b) HCSA strong cases and (c) HCSApoleward expansion cases The color scale represents the vectorrsquos magnitude The reference vector in (a) corresponds to 5m sminus1 whereas in(b) and (c) it is 1m sminus1

a particularly difficult challenge for model performanceBesides the Andes region is characterized by very sparsedata availability which is one of the major shortcomingsfor evaluating model performance over the South Americancontinent Increasing the model resolution can improve theresults Rojas [40] found that for climate change studies thehorizontal resolution required to reproduce adequate rainfallpatterns over the central Andes region is around 30 to 40 kmAlso it is important to remember that these model results aresensitive to the choice of the convective scheme In this paperthe Kuo scheme was chosen but Fernandez et al [39] foundthat the precipitation is in general better simulated with theGrell scheme than the Kuo scheme

Chen et al [6] found disagreements in the regional HCsintensity trends among six different reanalyses (Era-InterimERA-40 NCEP1 NCEP2 JRA25 and 20CR) For the HCover SA the authors showed that all six reanalyses presentan intensification trend although not all are statisticallysignificant The authors also found high correlation valuesfor the intensity index of the HC over SA between the Era-Interim and ERA-40 andNCEP2 and JRA25 reanalyses (089076 and 064 resp) whereas lower correlation values areseen with the 20CR and NCEP1 reanalyses (047 and 042resp) It is worth remembering that these both reanalyses(20CR and NCEP1) present some well-known issues In the20CR only surface pressure observations are assimilated

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 13: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Advances in Meteorology 13

Latit

ude

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)La

titud

e

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 13 Precipitation anomaly (mmdayminus1) for HCSA strong cases in the period 1996ndash2011 based on the (a) GPCC dataset and (b) RegCM4simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95 confidence level asdetermined by two-tailed Studentrsquos 119905-test

and therefore there is a potential for large differencesespecially in vertical structure between this dataset and otherreanalyses On the other hand in the NCEP1 reanalysis thepotential for differences rises due to the change in quantityand spatial coverage of the observed data especially from thebeginning of the major satellite era Thus the results foundhere for the HCSA intensification present some sensitivity tothe reanalysis used However we believe that they are reliableas the Era-Interim represents a newer-generation reanalysisthat incorporates many important model improvements andaccording to several studies (eg [31 32]) has the bestperformance among reanalysis datasets

4 Impacts of the HCSA Poleward Expansion

Figure 14 shows the anomalies of the annual mean horizontalwind at 200 and 850 hPa for the HCSA poleward expan-sion cases based on the Era-Interim reanalysis and modelsimulation Here as for the HCSA strong cases anomalouseasterly winds coming from the Atlantic Ocean are seenover northern SA at 200 hPa in the reanalysis (Figure 14(a))andmodel simulation (Figure 14(c))These anomalous windspresent in northern SA are weaker compared to those seenfor HCSA strong cases (Figures 6(a) and 6(c)) Around40∘S strong anomalous winds forHCSA poleward expansioncases are observed coming from the Atlantic Ocean andhitting the southeast SA (Figures 14(a) and 14(c)) Howeverwhen entering the mainland these winds become weakerand rotate clockwise This results in a significant anomalous

trough which is seen in reanalysis and model simulationat 200 hPa (Figure not shown) and 500 hPa over the centralregion of SA (Figure 15) The center of this trough is over thecentral Atlantic basin where the anomalous easterly windsare stronger (Figures 14(a) and 14(c)) This trough is moreintense compared to that seen for the HCSA strong cases(Figure 7)

The anomalous wind pattern at 850 hPa is character-ized by the weakening of the trade winds around equatorand a counterclockwise circulation in the South Atlanticaround 50∘S in the reanalysis (Figure 14(b)) and simulation(Figure 14(d)) This counterclockwise circulation results in asignificant anomalous ridge which is observed at 850 hPa(Figure not shown) and at 500 hPa over the southern SA andespecially over the southern of Atlantic basin in the reanal-ysis and simulation (Figure 15) Strong anomalous westerlywinds coming from the eastern Pacific cross the southern SA(Figures 14(b) and 14(d)) and contribute to the formation ofthis ridgeThis ridge ismore intense over the southern SA andAtlantic Ocean compared to that seen for the HCSA strongcases (Figure 7) and is associated with the robust descendingbranch of the HCSA (Figure 3(b))

The correlation between theHCSA annual poleward edgetime series and the annual mean zonal wind at 200 and850 hPa in the period 1996ndash2011 is displayed in Figure 16The correlation signs (positives and negatives) are the sameobserved for the HCSA strong cases (Figure 8) thoughthe statistically significant regions are different At 200 hPathe reanalysis dataset presents two regions with significantnegative correlations (over Peru and Atlantic Ocean around

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 14: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

14 Advances in Meteorology

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

1La

titud

eLongitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Latit

ude

Longitude2

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(c)

Latit

ude

Longitude1

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(d)

Figure 14 Annual mean horizontal wind (m sminus1) anomaly for HCSA poleward expansion cases in the period 1996ndash2011 based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa The color scale represents thevectorrsquos magnitude

45∘S) and two other regions with significant positive cor-relations (over the southern SA and the central area ofSouthAtlantic) (Figure 16(a))Themodel simulation presentssmaller and not significant correlations for the central areas ofSA and South Atlantic (Figure 16(c)) The correlation map at

850 hPa shows regions with significant positive and negativecorrelations over the southern SA and over the south ofBrazil and south of Atlantic around 15∘W respectively in thereanalysis and model simulation (Figures 16(b) and 16(d))Here as also seen for the HCSA strong cases regions with

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 15: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Advances in Meteorology 15

Table 4 Area averaged root mean square error (RMSE) and bias and spatial correlation coefficient (CC) for the zonal wind (m sminus1) at 200 hPa(ZW200) and 850 hPa (ZW850) geopotential height (m) at 500 hPa (GH) and air temperature (K) at 925 hPa (AT) anomalies for HCSApoleward expansion cases over the first four regions of Figure 5

Region RMSE BIAS CCZW200 ZW850 GH AT ZW200 ZW850 GH AT ZW200 ZW850 GH AT

1 044 026 224 024 minus013 009 159 minus013 072 045 020 0312 082 015 393 017 minus060 minus002 378 minus004 079 075 086 0393 061 022 214 010 050 013 203 003 096 097 099 0814 069 024 334 017 minus023 009 265 minus002 087 085 094 070

minus12

minus6

minus6

minus8

minus4

minus4

minus4

minus4

minus2

minus2

minus20

810

12

14

Latit

ude

Longitude

4

246

minus10

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

Latit

ude

Longitude

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

minus4

minus2

minus2

0

0

0

0

0

2

246

810

1416

18

20

6

12

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 15 Annual mean geopotential height (m) anomaly for HCSA strong cases at 500 hPa in the period 1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 2m and significant values above 90 confidence level from two-tailedStudentrsquos 119905-test are shaded

anomalous easterly (westerly) winds coming from (towards)the Atlantic Ocean have negative (positive) correlations withthe HCSA poleward edge

Table 4 shows that for the zonal wind anomalies at200 hPa the northern SA exhibits weaker spatial correlationbetween the model and observation (072) although thevalues of RMSE and bias (absolute value) are the lowestcompared to the other regions The central region exhibitsthe highest values of RMSE and bias for the anomalies ofzonal wind at 200 hPa and geopotential height at 500 hPawhereas the opposite is found for the zonal wind anomaliesat 850 hPa Theses aspects were also verified for the HCSAstrong cases (Table 2) The southern SA exhibits the highestspatial correlation for the anomalies of zonal wind at 200 and850 hPa and geopotential height at 500 hPa (096 097 and099 resp) As well these correlation values are greater thanfor the HCSA strong cases

Figure 17 displays the annual mean temperature anomalyfor the HCSA poleward expansion cases Significant positiveanomalies are seen in the reanalysis over northern SA(between 0 and 5∘S and 60∘ and 45∘W) (Figure 17(a)) whereasnegative anomalies are found over the east Pacific around20∘S in the reanalysis and model simulation (Figures 17(a)and 17(b)) Table 4 shows that the model underestimatesthe temperature anomalies in the northern region (bias =minus013) As well this region exhibits the highest values ofRMSE and bias (absolute value) and the lowest value ofcorrelation indicating a weaker agreement between themodel and observation compared to the other regions Itis worth remembering that the observational data are likelyaffected by significant uncertainties particularly in remoteareas of the Amazon basin and this makes a rigorous modelassessment over this region rather difficult Moreover localprocesses involving land-atmosphere interactions influence

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 16: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

16 Advances in Meteorology

minus06

minus04

minus04

minus04

minus04

minus04

minus02

minus02

minus02

minus02

0

0

0

0

02

02

02

02

04

04

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus04

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus06

0

0

0

0

0

0

0

0

0

0

02

0202

02

02

04

04

04

06

04

04

Latit

ude

Longitude

minus04

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Latit

ude

Longitude

minus04

minus04

minus04

minus04

minus04

minus04

minus06

minus06

minus02

minus02

minus02

minus02

minus02

0

0

0

02

02

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(c)

minus04

minus04

minus06

minus02

minus02

minus02

minus02

minus02minus0

minus02

minus02

00

00

0

0

02

02

02

0202

02

0204

04

04

04

04

06

06

Latit

ude

Longitude

02

minus02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(d)

Figure 16 Correlation between the HCSA annual poleward edge time series and the annual mean zonal wind (m sminus1) in the period 1996ndash2011based on the Era-Interim reanalysis at (a) 200 hPa and (b) 850 hPa and RegCM4 simulation at (c) 200 hPa and (d) 850 hPa Contour intervalsare 02 and values above 05 are shaded and are statistically significant at 95 confidence level Here the poleward edge is defined as thelatitude where the OLR is equal to 240Wmminus2 For clarity purposes it should be noted that the HCSA annual poleward edge time series havebeen multiplied by (minus1)

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 17: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Advances in Meteorology 17

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus04

minus04

minus04minus03

minus03

minus03

minus03

minus02

minus02

minus02

minus02

minus02

minus02

0

0

00

0

0

01

01

01

01

01

01

0102

02

02

02

03

Latit

ude

Longitude

minus03

01

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

Latit

ude

Longitude

minus01

minus01

minus01

minus01

minus01

minus01

minus01

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus02

minus04

minus04

minus04minus04

minus04

minus02

0

0

0

0

0

01

01

01

01

01

01

01

02

02

02

02

02

03

minus03

minus03

minus03

minus03

minus03

minus03

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 17 Annual mean temperature (K) anomaly at 925 hPa for HCSA poleward expansion cases in the period 1996ndash2011 based on the(a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 01 K and significant values above 90 confidence level fromtwo-tailed Studentrsquos 119905-test are shaded

the surface climate of the Amazon basin thus the treatmentof surface processes is crucial [11] On the other hand thesouthern region exhibits good agreement between the modeland observation with low values for RMSE and bias and highvalues for spatial correlation

The temperature anomalies for the HCSA polewardexpansion cases seem to be very similar to those foundfor the strong cases however there are differences in thestatistically significant regions andor the anomalies intensityCompared to the HCSA poleward expansion cases yearsthat present HCSA intensification show significant positiveanomalies over a great part of the northern region reducedanomalies over the southern SA and the central area of theSouth Atlantic and more intense positive anomalies over thenorthern regions of SA and South Atlantic (Figure 9(a))

From the combination of Figures 15 and 17 one canconclude that a strong cold trough is observed in the centralregions of SA and South Atlantic and a strong warm ridgeis seen over the southern SA and South Atlantic around50∘S when the HCSA expands poleward Here as mentionedearlier the trough and the ridge are more intense than for theHCSA strong cases (Figure 7)

Figure 18 displays the correlation between the HCSAannual poleward edge time series and the annual meantemperature at 925 hPa in the period 1996ndash2011This figure issimilar to Figure 10 which shows the correlation between theHCSA intensity and the annualmean temperature at 850 hPaHowever the results show that in the HCSA strong cases

high correlations are seen over the northern regions of SAand South Atlantic (above 06) The correlations are smallerover these regions in the HCSA poleward expansion cases

The temperature anomalies seen over the Atlantic Oceanand eastern Pacific (Figure 17) are connected with theSST anomalies for the HCSA poleward expansion cases(Figure 19) The La Nina pattern is also observed in thePacific region but with greater intensity in the equatorialregion and weaker anomalies in the eastern Pacific around20∘S compared to the HCSA strong cases (Figure 11) Dueto these facts the Pacific Walker circulation here presentsweaker downward motion in eastern Pacific (around 90∘W)stronger upward motion in western Pacific (around 130∘E)and more intense subsidence around 180∘ compared to theHCSA strong cases (Figures 12(b) and 12(c)) The HCSApoleward expansion cases are 1996 2008 2010 and 2011 yearsThe three last years are common to both composites thatrepresent the HCSA intensification and poleward expansionThus the 1996 year is the differential for the HCSA polewardexpansion composite and a La Nina event was also observedduring 1995-1996 (Historical ENSO episodes (1950ndashpresent)httpwwwcpcnoaagovproductsanalysis monitoringen-sostuffensoyearsshtml)

Over the Atlantic Ocean positive SST anomalies are nowseen from 20∘S to 20∘N Thus here we see a southwardextension and reduced intensity at the north of equator ofthe positive SST anomalies compared to HCSA strong cases(Figures 11 and 19)TheHCSA ascending branch for poleward

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 18: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

18 Advances in MeteorologyLa

titud

e

Longitude

minus04minus04

minus02

minus02minus02

minus02

minus02

minus02

0

0

0

0

0

0

02

02

02

02

02

02

02

02

04

04

04

04

06

04

04

04

04

04

0

0

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(a)

minus04

minus02 minus02

minus02

minus02

minus02

minus02

minus02

02

02

02

02

02

02

02 04

04

04

04

04

06

04

04

04

0

0 0

0

0

0

0

0

Latit

ude

Longitude

02

10∘W

15∘W

20∘W

25∘W

30∘W

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(b)

Figure 18 Correlation between the HCSA annual poleward edge time series and the annual mean temperature (K) at 925 hPa in the period1996ndash2011 based on the (a) Era-Interim reanalysis and (b) RegCM4 simulation Contour intervals are 02 and values above 05 are shaded andare statistically significant at 95 confidence level Here the poleward edge is defined as the latitude where the OLR is equal to 240Wmminus2For clarity purposes it should be noted that the HCSA annual poleward edge time series have been multiplied by (minus1)

0

Longitude

Latit

ude

minus05

minus04

minus03

minus02

minus01

0

01

02

03

04

05

180 0

60∘N

30∘N

60∘S

30∘S

150∘W 120

∘W 90∘W 30

∘W60∘W

Figure 19 Annual mean sea surface temperature (K) anomaly forHCSA poleward expansion cases in the period 1996ndash2011 based onthe HADISST dataset Contour intervals are 01 K with stipplingindicating regions statistically significant above the 95 confidencelevel as determined by two-tailed Studentrsquos 119905-test

expansion cases is weaker compared to the intensificationcases probably associated with reduced temperature anoma-lies over northern SA (Figure 17(a)) however the descend-ing branch is stronger and shifted poleward as expected(Figures 3(a) and 3(b)) The ascending branch of the Walkercirculation for HCSA poleward expansion cases over thenorthern SA (around 70∘W) is also weaker compared to thestrong cases however the anomalous upward motion seenalong the Atlantic basin is stronger (Figures 12(b) and 12(c))

due to the southward extension of the positive SST anomalies(Figures 11 and 19) This southward extension is associatedwith theweakening of the tradewinds on the equator (Figures14(b) and 14(d)) in contrast to their intensification in theHCSA strong cases (Figure 6(b))

The annual anomalous rainfall distribution during theHCSA poleward expansion cases analyzed (Figure 20) issimilar to that seen for strong cases (Figure 13) However itfollows more closely the La Nina canonical impacts due tothe influence of the southward extension of the positive SSTanomalies in Atlantic especially over the edge of northeastSA (mostly positive rainfall anomalies) (Figure 20(a)) Chenet al [6] found that when the poleward edge of the regionalHC is located to the south of its climatological locationsignificant descending motion and negative precipitationanomalies can be observed over the regions to the south ofthe respective regional HC climatological poleward edgeThesignificant descending motion around 35∘S can be seen inFigure 3(b) and the negative precipitation anomalies associ-ated are observed over southeast (south of Brazil Uruguayand Argentina) and southwestern coast of SA (from 35∘ to45∘S) in Figure 20 The model overestimates the negativerainfall anomalies over south of Brazil and southwesterncoast of SA (from 33∘ to 45∘S) However Table 5 showsthat the highest spatial correlation between the model andobservation is found in the southeast of SA In the northeastSA (region 6 of Figure 5) the observed rainfall anomaliesare underestimated by the model (bias = minus014) as seen for

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 19: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Advances in Meteorology 19

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

(a)

2

15

1

05

025

0

minus025

minus05

minus1

minus15

Latit

ude

Longitude

EQ

45∘S

50∘S

55∘S

40∘S

35∘S

30∘S

25∘S

20∘S

15∘S

10∘S

5∘S

5∘N

10∘N

35∘W

40∘W

45∘W

50∘W

55∘W

60∘W

65∘W

70∘W

75∘W

80∘W

(b)

Figure 20 Precipitation anomaly (mmdayminus1) for HCSA poleward expansion cases in the period 1996ndash2011 based on the (a) GPCC datasetand (b) RegCM4 simulation Contour intervals are 025mmdayminus1 with hatching indicating regions statistically significant above the 95confidence level as determined by two-tailed Studentrsquos 119905-test

Table 5 Area averaged root mean square error (RMSE) and biasand spatial correlation coefficient (CC) for the annual mean pre-cipitation (mmdayminus1) anomalies for HCSA poleward expansioncases over the last three regions of Figure 5 in the period 1996ndash2011

Region RMSE (mmdayminus1) BIAS (mmdayminus1) CC5 047 006 0656 032 minus014 0197 050 005 026

the HCSA strong cases As well in this region is found thelowest correlation between the model and observation

Several studies found that the performance of the regionalclimate models varies regionally and there is no singleparameter setting that performs best in all domains tested[11 39] Therefore the models must be adequately tuned inorder to give reliable predictions in the different regions of SA[39] In both cases (intensification and poleward expansion ofthe HCSA) the best combination of low values for RMSE andbias (absolute value) and high values for spatial correlationare found in the southern region for the anomalies of zonalwind geopotential height and temperatureTheprecipitationanomalies were reasonably reproduced by the model in thesoutheast regionThis indicates that the southern SA exhibitsgood agreement between themodel and observationsThere-fore the current model setup is considered satisfactory forapplication in future climate studies in this region

5 Discussion and Conclusions

This paper investigated the impacts of changes in the intensityand poleward edge of regional HC over SA on the patternsof wind geopotential height precipitation and temperatureusing the regional climate model (RegCM4) and observa-tional datasets during the 1996ndash2011 period

Several studies indicate aweakening andpoleward expan-sion of the zonal HC in a global warming scenario [4 41]However the changes in the regional HC can be different andthey are still lesser known As well zonally averaged changesmay reflect strong regional circulation changes

Our results have shown that during the 1996ndash2011 periodthere are trends of intensification and poleward expansionof the HCSA Therefore composites were constructed toevaluate the impacts of these changes in the HCSA onthe patterns of wind geopotential height precipitation andtemperature Significant correlations are seen between thetemperature zonal wind and the HCSA intensity over thenorthern central and southern regions of SA and SouthAtlantic Results have also shown that in both compositesregions with anomalous easterly (westerly) winds comingfrom (towards) the Atlantic Ocean have negative (positive)correlations with the HCSA intensity and poleward edge

In summary the following outcomes have been seen forthe HCSA strong cases analyzed

(i) A cold trough in the central regions of SA and SouthAtlantic and a warm ridge over the southern SA andSouth Atlantic around 50∘S

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 20: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

20 Advances in Meteorology

(ii) An intense ascending branch of the Hadley andWalker circulations due to Amazon heat sourceintensification

(iii) A La Nina pattern in the equatorial Pacific region anda ldquonew-typerdquo Atlantic Nino with positive SST anom-alies centered at 15∘N due to the weakening of thenorthern trade winds and strengthening of the tradewinds on the equator

(iv) Annual anomalous rainfall distribution followingapproximately the LaNina canonical impacts but alsopresenting the influence of the positive SST anomaliesin Atlantic especially over the edge of northeast SA(mostly negative rainfall anomalies)

The following outcomes have been seen for the HCSApoleward expansion cases compared to the strong casesanalyzed

(i) A stronger cold trough in the central regions of SAand South Atlantic and a stronger warm ridge overthe southern SA and South Atlantic around 50∘S

(ii) Aweaker ascending branch and a stronger and shiftedpoleward descending branch of the HCSA

(iii) A weaker ascending branch over the northern SA andan anomalous upwardmotion seen along the Atlanticbasin for the Walker circulation

(iv) A stronger La Nina pattern in the equatorial Pacificregion and a southward extension of the positive SSTanomalies in the Atlantic due to weaker trade windson the equator

(v) Annual anomalous rainfall distribution followingmore closely the LaNina canonical impacts due to theinfluence of the southward extension of the positiveSST anomalies in Atlantic especially over the edge ofnortheast SA (mostly positive rainfall anomalies)

The performance of the RegCM4 in simulating theclimate anomalies over different regions of SA associatedwith changes in the HCSA was assessed The results haveshown that in general the model is capable of simulatingthemain features of the circulation anomalies associated withthe intensification and poleward expansion of the HCSATheperformance of the model varies regionally and the southernSA exhibited better agreement between the model and obser-vations for the anomalies of zonal wind geopotential heightand temperature in the 1996ndash2011 period The precipitationanomalies were reasonably reproduced by the model inthe southeast region Studies have shown that the regionalclimate models show a significant sensitivity to differentparameterizations and parameter settings which can be usedto optimize the model performance over different regionsThus further studies using other cumulus parametrizationdifferent domain size and period of simulations will bepursued in the future

It is important to highlight that the composites represent-ing theHCSA intensification and poleward expansion are dif-ferent by only one yearThis is why notable similarity is foundin the results for the intensification and poleward expansion

cases which could indicate that in most circumstances thechanges in the intensity and poleward edge of the HCSA areoccurring simultaneously

The main differences between the results for the inten-sification and poleward expansion cases are seen in thestatistically significant regions andor anomalies intensityThe changes in the SST anomalies between the cases helpto explain these differences However it is always importantto have in mind that many mechanisms and interactionscan be responsible for controlling the intensity and polewardedge of the regional HC This is an issue of continuallygrowing knowledge due to complexity of the coupled ocean-atmosphere-land systemWe hope that this paper contributesto a better understanding about the local impacts of thechanges in regional HC and helps to develop insights intothe mechanisms that lead to these spatially heterogeneouschanges and into the future of this circulation in a changingclimate

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

Acknowledgments

The authors thank FAPESP (Fundacao de Amparo a Pesquisado Estado de Sao Paulo) for financial support (Grants 201113669-0 and 200858101-9) for the development of thisresearch CNPQ (Conselho Nacional de DesenvolvimentoCientıfico e Tecnologico) and ITV (Vale Institute of Technol-ogy) have also partially funded Tercio Ambrizzi

References

[1] S Hastenrath Climate and Circulation of the Tropics D RiedelDordrecht The Netherlands 1985

[2] C M Johanson and Q Fu ldquoHadley cell widening modelsimulations versus observationsrdquo Journal of Climate vol 22 no10 pp 2713ndash2725 2009

[3] J Lu G A Vecchi and T Reichler ldquoExpansion of the Hadleycell under global warmingrdquo Geophysical Research Letters vol34 Article ID L06805 2007

[4] D M W Frierson J Lu and G Chen ldquoWidth of the Hadleycell in simple and comprehensive general circulation modelsrdquoGeophysical Research Letters vol 34 Article ID L18804 2007

[5] X Quan H F Diaz andM P Hoerling ldquoChange in the tropicalHadley cell since 1950rdquo inThe Hadley Circulation Present Pastand Future H F Diaz and R S Bradley Eds vol 21 ofAdvancesin Global Change Research pp 85ndash120 Springer DordrechtTheNetherlands 2004

[6] S Chen K Wei W Chen and L Song ldquoRegional changes inthe annual mean Hadley circulation in recent decadesrdquo Journalof Geophysical Research Atmospheres vol 119 no 13 pp 7815ndash7832 2014

[7] A H Oort and J J Yienger ldquoObserved interannual variabilityin the Hadley circulation and its connection to ENSOrdquo Journalof Climate vol 9 no 11 pp 2751ndash2767 1996

[8] C Wang ldquoENSO Atlantic climate variability and the Walkerand Hadley circulationsrdquo in The Hadley Circulation Present

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 21: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Advances in Meteorology 21

Past and Future H F Diaz and R S Bradley Eds pp 173ndash202Kluwer Academic 2005

[9] T Ambrizzi E B de Souza and R S Pulwarty ldquoTheHadley andWalker regional circulations and associated ENSO impacts onSouth American seasonal rainfallrdquo in The Hadley CirculationPresent Past and Future H F Diaz and R S Bradley Eds vol21 of Advances in Global Change Research pp 203ndash235 KluwerAcademic New York NY USA 2004

[10] G Zeng W-C Wang Z B Sun and Z X Li ldquoAtmosphericcirculation cells associated with anomalous east Asian wintermonsoonrdquo Advances in Atmospheric Sciences vol 28 no 4 pp913ndash926 2011

[11] F Giorgi E Coppola F Solmon et al ldquoRegCM4 modeldescription and preliminary tests over multiple CORDEXdomainsrdquo Climate Research vol 52 no 1 pp 7ndash29 2012

[12] D P Dee S M Uppala A J Simmons et al ldquoThe ERA-Interimreanalysis configuration and performance of the data assim-ilation systemrdquo Quarterly Journal of the Royal MeteorologicalSociety vol 137 no 656 pp 553ndash597 2011

[13] J S Pal F Giorgi X Bi et al ldquoRegional climatemodeling for thedeveloping world the ICTP RegCM3 and RegCNETrdquo Bulletinof the American Meteorological Society vol 88 no 9 pp 1395ndash1409 2007

[14] J T Kiehl J J Hack G B Bonan et al ldquoDescription of theNCAR community climate model (CCM3)rdquo NCAR TechnicalNote NCARTN-420+STR National Center for AtmosphericResearch Boulder Colo USA 1996

[15] R E Dickinson A Henderson-Sellers and P KennedyldquoBiosphere-atmosphere transfer scheme (BATS) Version 1e ascoupled to the NCAR community climate modelrdquo TechnicalReport National Center for Atmospheric Research TechnicalNote TN-387+STR NCAR Boulder Colo USA 1993

[16] R A Anthes ldquoA cumulus parameterization scheme utilizing aone-dimensional cloud modelrdquo Monthly Weather Review vol105 no 3 pp 270ndash286 1977

[17] R Anthes E-Y Hsie and Y-H Kuo ldquoDescription of thePenn StateNCARMesoscale model version 4 (MM4)rdquo NCARTechicalNoteNCARTN-282+STRUniversityCorporation forAtmospheric Research Boulder Colo USA 1987

[18] G A Grell ldquoPrognostic evaluation of assumptions used bycumulus parameterizationsrdquoMonthly Weather Review vol 121no 3 pp 764ndash787 1993

[19] F Giorgi M R Marinucci G T Bates and G de CanioldquoDevelopment of a second-generation regional climate model(RegCM2) Part II convective processes and assimilation oflateral boundary conditionsrdquoMonthly Weather Review vol 121no 10 pp 2814ndash2832 1993

[20] K A Emanuel ldquoA scheme for representing cumulus convectionin large-scale modelsrdquo Journal of the Atmospheric Sciences vol48 no 21 pp 2313ndash2335 1991

[21] R W Reynolds N A Rayner T M Smith D C Stokes andW Wang ldquoAn improved in situ and satellite SST analysis forclimaterdquo Journal of Climate vol 15 no 13 pp 1609ndash1625 2002

[22] M S Reboita L M M Dutra and R P da Rocha ldquoRegCM4dynamical downscaling of seasonal climate predictions over theSoutheast of Brazilrdquo in Geophysical Research Abstracts vol 15EGU2013-6061 2013

[23] M P Marcella and E A B Eltahir ldquoModeling the summertimeclimate of Southwest Asia the role of land surface processes inshaping the climate of semiarid regionsrdquo Journal of Climate vol25 no 2 pp 704ndash719 2012

[24] U Schneider A Becker P Finger A Meyer-Christoffer MZiese and B Rudolf ldquoGPCCrsquos new land surface precipitationclimatology based on quality-controlled in situ data and its rolein quantifying the global water cyclerdquo Theoretical and AppliedClimatology vol 115 no 1-2 pp 15ndash40 2014

[25] N A Rayner D E Parker E B Horton et al ldquoGlobal anal-yses of sea surface temperature sea ice and night marine airtemperature since the late nineteenth centuryrdquo Journal of Geo-physical Research vol 108 no 14 pp 1ndash37 2003

[26] T N Krishnamurti ldquoTropical east-west circulations during thenorthern summerrdquo Journal of the Atmospheric Sciences vol 28no 8 pp 1342ndash1347 1971

[27] T N Krishnamurti M Kanamitsu W J Koss and J D LeeldquoTropical east-west circulations during the northern winterrdquoJournal of the Atmospheric Sciences vol 30 no 5 pp 780ndash7871973

[28] S Hastenrath ldquoIn search of zonal circulations in the equatorialatlantic sector from the NCEP-NCAR reanalysisrdquo InternationalJournal of Climatology vol 21 no 1 pp 37ndash47 2001

[29] B Liebmann and C A Smith ldquoDescription of a complete(interpolated) outgoing longwave radiation datasetrdquo Bulletin ofthe AmericanMeteorological Society vol 77 pp 1275ndash1277 1996

[30] K E Trenberth RDole Y Xue et al ldquoAtmospheric reanalyses amajor resource for ocean product development and modelingrdquoinProceedings of the SustainedOceanObservations and Informa-tion for Society (OceanObs rsquo09) J Hall D E Harrison and DStammer Eds vol 2 of WPP-306 pp 21ndash25 ESA PublicationVenice Italy September 2009

[31] R Lin T Zhou and Y Qian ldquoEvaluation of global monsoonprecipitation changes based on five reanalysis datasetsrdquo Journalof Climate vol 27 no 3 pp 1271ndash1289 2014

[32] E Jakobson T Vihma T Palo L Jakobson H Keernik and JJaagus ldquoValidation of atmospheric reanalyses over the centralArctic Oceanrdquo Geophysical Research Letters vol 39 no 10Article ID L10802 2012

[33] M B Sylla E Coppola L Mariotti et al ldquoMultiyear simulationof theAfrican climate using a regional climatemodel (RegCM3)with the high resolution ERA-interim reanalysisrdquo ClimateDynamics vol 35 no 1 pp 231ndash247 2010

[34] J-H Park S G Oh M S Suh and H S Kang ldquoImpact ofboundary conditions on RegCM4 20-year-long precipitationsimulations over CORDEX-East Asia domainrdquo in Proceedingsof the EGU General Assembly p 4475 Vienna Austria April2012

[35] R J Bombardi L M V Carvalho and C Jones ldquoSimulating theinfluence of the SouthAtlantic dipole on the SouthAtlantic con-vergence zone during neutral ENSOrdquo Theoretical and AppliedClimatology vol 118 no 1-2 pp 251ndash269 2014

[36] F Cabre S Solman and M Nunez ldquoClimate downscalingover southern South America for present-day climate (1970ndash1989) using the MM5 model Mean interannual variability andinternal variabilityrdquo Atmosfera vol 27 no 2 pp 117ndash140 2014

[37] I Richter S K Behera YMasumoto B Taguchi H Sasaki andT Yamagata ldquoMultiple causes of interannual sea surface tem-perature variability in the equatorial Atlantic Oceanrdquo NatureGeoscience vol 6 no 1 pp 43ndash47 2013

[38] C F Ropelewski and M S Halpert ldquoGlobal and regional scaleprecipitation patterns associated with the El NinoSouthernOscillationrdquoMonthly Weather Review vol 115 no 8 pp 1606ndash1626 1987

[39] J P R Fernandez S H Franchito and V B Rao ldquoSimulationof the summer circulation over South America by two regional

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 22: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

22 Advances in Meteorology

climate models Part I mean climatologyrdquo Theoretical andApplied Climatology vol 86 no 1-4 pp 247ndash260 2006

[40] M Rojas ldquoMultiply nested regional climate simulation forsouthern South America sensitivity to model resolutionrdquoMonthly Weather Review vol 134 no 8 pp 2208ndash2223 2006

[41] J Lu G Chen and D M W Frierson ldquoResponse of the zonalmean atmospheric circulation to El Nino versus global warm-ingrdquo Journal of Climate vol 21 no 22 pp 5835ndash5851 2008

[42] J S Pal E E Small and E A B Eltahir ldquoSimulation of regional-scale water and energy budgets representation of subgridcloud and precipitation processes within RegCMrdquo Journal ofGeophysical Research D Atmospheres vol 105 no 24 pp29579ndash29594 2000

[43] A A M Holtslag E I F De Bruijn and H-L Pan ldquoAhigh resolution air mass transformation model for short-rangeweather forecastingrdquo Monthly Weather Review vol 118 no 8pp 1561ndash1575 1990

[44] X Zeng M Zhao and R E Dickinson ldquoIntercomparison ofbulk aerodynamic algorithms for the computation of sea surfacefluxes using TOGA COARE and TAO datardquo Journal of Climatevol 11 no 10 pp 2628ndash2644 1998

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in

Page 23: Research Article Recent Changes in the Annual Mean ...downloads.hindawi.com/journals/amete/2015/780205.pdfAdvances in Meteorology 1000 925 850 700 500 400 300 250 200 150 100 Pressure

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

EarthquakesJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MineralogyInternational Journal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geological ResearchJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Geology Advances in