This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 128.97.194.87

This content was downloaded on 14/04/2016 at 00:07

Please note that terms and conditions apply.

Spring land temperature anomalies in northwestern US and the summer drought over

Southern Plains and adjacent areas

View the table of contents for this issue, or go to the journal homepage for more

2016 Environ. Res. Lett. 11 044018

(http://iopscience.iop.org/1748-9326/11/4/044018)

Home Search Collections Journals About Contact us My IOPscience

Environ. Res. Lett. 11 (2016) 044018 doi:10.1088/1748-9326/11/4/044018

LETTER

Spring land temperature anomalies in northwestern US and thesummer drought over Southern Plains and adjacent areas

YongkangXue1,2, CatalinaMOaida2, IsmailaDiallo1, J DavidNeelin2, Suosuo Li1,3, FernandoDe Sales1,4,YuGu2, DavidARobinson5, RatkoVasic6 and LanYi7

1 Department ofGeography, University of California, Los Angeles, CA 90095,USA2 Department of Atmospheric &Oceanic Sciences, University of California, Los Angeles, CA 90095,USA3 Cold andArid Regions Environmental andEngineering Research Institute, Chinese Academy of Sciences, Lanzhou, 730000, People’s

Republic of China4 SanDiego StateUniversity, SanDiego, CA 921825 Department ofGeography, RutgersUniversity, Piscataway,NJ 08854,USA6 National Center for Environmental Prediction, College Park,MD20740,USA7 Chinese Academy ofMeteorological sciences, Beijing, 10081, People’s Republic of China

E-mail: yxue@geog.ucla.edu

Keywords: land temperature, US drought and heat, regional climatemodel, southern great plains, sea surface temperature

Supplementarymaterial for this article is available online

AbstractRecurrent drought and associated heatwave episodes are important features of theUS climate.Manystudies have examined the connection between ocean surface temperature changes and conterminousUS droughts. However, remote effects of large-scale land surface temperature variability, over shorterbut still considerable distances, onUS regional droughts have been largely ignored. The present studycombines two types of evidence to address these effects: climate observations andmodel simulations.Our analysis of observational data shows that springtime land temperature in northwest US issignificantly correlatedwith summer rainfall and surface temperature changes in theUS SouthernPlains and its adjacent areas. Ourmodel simulations of the 2011 Southern Plains drought using ageneral circulationmodel and a regional climatemodel confirm the observed relationship betweenland temperature anomaly and drought, and suggest that the long-distance effect of land temperaturechanges in the northwest US on Southern Plains droughts is probably as large as themore familiareffects of ocean surface temperatures and atmospheric internal variability.We conclude that the cool2011 springtime climate conditions in the northwest US increased the probability of summer droughtand abnormal heat in the Southern Plains. The present study suggests a strong potential formoreskillful intra-seasonal predictions ofUS Southern Plains droughts when such facts as ones presentedhere are considered.

1. Introduction

Recurrent drought and associated heatwave episodesare important features of the conterminous UnitedStates climate [1]. Many analyses of US drought datashow that the Great Plains is one of the most drought-prone regions across the nation [2–4]. US droughtshave been attributed to various mechanisms, includ-ing natural variability of the Great Plains climate (i.e.the great droughts there seem to occur once or twice acentury [5, 6]), the presence of an atmospheric

circulation pattern at the beginning of a period leadingto anomalies and/or preferred atmospheric condi-tions for maintenance of the drought [7, 8], effects ofsoil moisture and land conditions [8–12], and tele-connection with sea surface temperature (SST)anomalies. Of these, SST anomalies have been studiedthe most as a factor responsible for drought. The keyfinding is that the El Niño–Southern Oscillation andthe Pacific decadal oscillation (PDO) in their coldphases and the Atlantic Multidecadal Oscillation in itswarm phase tend to cause droughts over the US, with

OPEN ACCESS

RECEIVED

9November 2015

REVISED

25 February 2016

ACCEPTED FOR PUBLICATION

24March 2016

PUBLISHED

13April 2016

Original content from thisworkmay be used underthe terms of the CreativeCommonsAttribution 3.0licence.

Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.

© 2016 IOPPublishing Ltd

the Pacific Ocean playing the dominant role [2, 13–17]. The studies showed statistically significant corre-lations above∼0.3 at the 0.05 level between Pacific SSTand droughts in the Great Plains and the adjacentsouthwest [18–20]. Modeling studies confirm obser-vational findings in support of the important role SSTsplay in US drought. However, SST fluctuations aloneonly partially account for drought severity and recur-rence [21]. Here we explore a new approach tounderstand the 2011 Texas drought by identifyingremote land surface temperature (LST) and land-atmospheric interactions as notable contributors.

2.Methods

The present study combines both climate observationsand model simulations to investigate the remoteeffects of spring LST and subsurface soil temperature(SUBT) anomalies in the northwest US (NWUS, 117 °W–125 °W and 33 °N–49.8 °N (land only))8 as con-tributing factors to drought in the US Southern GreatPlains and its adjacent areas, including some areas tothe east of the Southern Great Plains (SGP, 88 °W–

103 °W and 29 °N–38 °N)9. Ground measurementsfrom the Climate Prediction Center’s global gauge-based analysis of precipitation and surface temper-ature (GTS) [22] and surface temperature station datafrom the Climate Anomaly Monitoring System(CAMS) [23], as well as the Rutgers University GlobalSnow Lab snow cover extent data (all of which coverthe period from 1980 through 2011) were used tostudy the relationship between LST in the NWUS andprecipitation in the SGP, as well as the contribution ofsnow cover in western US coastal area to this relation-ship. A recent study found that since 1980, theprecipitation decrease in Central US has intensi-fied [17].

Furthermore, to test these observed relationshipsand the hypothesis that abnormally cold NWUSspring conditions may have contributed to the 2011June–July SGP record drought and heat anomalies,and to compare LST and SUBT effects with SST for-cing, the WRF-NMM regional climate model (RCM)with 50 km horizontal resolution and the NCEP Glo-bal Forecast System (GFS)with T62 horizontal resolu-tion were employed for numerical sensitivity studies.Both were coupled with the Simplified Simple Bio-sphere model (SSiB) [24], which is a biophysicallybased model that simulates land-atmosphere interac-tions by calculating the surface energy budget and sur-face water balance for three soil layers and onevegetation layer. A force–restoremethod [25]was usedto calculate the heat transfer between the surface andsubsurface soil layers.

3. Results

3.1.Observational evidenceAlthough the SST remote effects on the drought havebeen extensively investigated, paradoxically, less atten-tion has been devoted to the effects of large-scale LST,over shorter but still considerable distances, on USregional droughts, despite the fact that the areas withLST anomalies are geographically closer to the droughtarea than the areas with SST anomalies.

While there have been studies that suggest aninverse relationship between spring snowmass (whichis closely linked to LST) in the western US and sub-sequent summer precipitation over the US southwestassociated with the North American monsoon system[26, 27], this snow–monsoon relationship has beenunstable over time and space, showing a peculiar on-and-off characteristic in the last century [28], and itseffect on monsoon through soil moisture has beenchallenged in another study [29]. Using the GTS andCAMS surface temperature station data from 1980through 2011, we examine the observed LST–pre-cipitation relationship and find that conditions withcold spring LST in the NWUS have a high probabilityof being associated with drier and warmer summerconditions in the SGP as displayed in figures 1(a) and(c), which show scatter plots of NWUSMay LST com-pared with June precipitation and with surface temp-erature over the SGP for 1980–2011, respectively. Forcomparison, the scatter plots ofMay SST inNorth EastPacific (NEP) (124 °W–135 °W and 30 °N–50 °N)10

versus June precipitation and temperature in the SGPare also shown in figures 1(b) and (d), respectively.The domain selections for LST and SST are based onfigures 2 and S1, which show maximum anomalies ofLST over the NWUS and SST over the NEP, respec-tively, to be discussed later. The generally positive cor-relations with precipitation for these 32 years areapparent for both LST and SST, with correlation coef-ficients of 0.33 and 0.35, respectively, and statisticalsignificance atα<0.1 for the two-sided T-test, whichis also used for other statistical tests in this paper. TheSST-precipitation correlations are consistent with pre-viously published correlations between the Pacific SSTand precipitation in the Great Plains [18–20], whichwere based on different data sets.

If we select the years with large LST anomalies outof the 32-years record spanning 1980 to 2011, highercorrelations amongst NWUS spring temperatures andsummer time SGP rainfall and temperature are found.For instance, there were 19 years, 10 cold (1980, 1988,1990, 1991, 1996, 1998, 1999, 2002, 2010, 2011) and 9warm (1987, 1992, 1993, 1994, 1997, 2001, 2006, 2007,2009) (see figure 1(a)), when absolute LST anomaliesover the NWUS were larger than 0.6 °C (about half ofone standard deviation). Among the 10 cold years(blue diamonds plus red triangle in figure 1(a)), seven

8See red box infigure 2(a) for reference.

9See box infigure 2(b) for reference.

10See black box infigure 2(a) andfigure S1 for reference.

2

Environ. Res. Lett. 11 (2016) 044018

showed negative June precipitation anomaly over theSGP (with five of those seven years having negativeprecipitation anomalies exceeding one standard

deviation), two were normal, and another had a slightpositive anomaly for the area as a whole but a strongnegative anomaly near the south Texas coast.

Figure 1.May temperature and June precipitation scatter plots. Left panels: surface temperature; right panels: SST. Study regions are124–135 °W/30–50 °N for SST (K), 117–125 °W/33–49.8 °N for LST (K), and 88–103 °W/29–38 °N for precipitation (mm/day).Bright red triangles denote year 2011 values. Black dashed lines indicatemeans; blue dashed lines denote+/−1 standard deviationboundaries for June P (a), (b) and June LST (c), (d) over SGP; orange dashed lines denote+/−0.6 °C forMay LST (a), (c) overNWUSand+/−0.4 °C forMay SST (b), (d) overNEP. The points are colors coded (blue, green, orange) to represent cool, neutral andwarmtemperature anomalies.

Figure 2. (a)ObservedMay LST and SST difference (°C) between 9 coldest years and 9warmest years (based onNWUSLST), (b)observed June precipitation difference (mmday-1) between the same years as infigures 2(a), and (c) observed June LST difference (°C)between the same years as infigure 2(a). The dotted areas denote statistical significance at theα=0.01 level of t-test values.

3

Environ. Res. Lett. 11 (2016) 044018

Removing one cold year to have same number ofwarm and cold years, the correlation between MayLST in the NWUS and June precipitation in the SGPfor the 18 large anomaly years is 0.53, with statisticalsignificance atα< 0.05.

Meanwhile, May LSTs in the NWUS are also nega-tively correlated with June LSTs in the SGP(figure 1(c)). The correlation coefficients betweenthem for all 32 years and for the 18 large anomaly yearsare−0.46 and−0.68, respectively, with significance atthe 0.01 (table 1). Among the 10 coldest NWUS Mays(blue diamonds plus red triangle in figure 1(c)), fivewere associated with high positive June temperatureanomalies in the SGP exceeding one standard devia-tion. There were only two other such hot SGP anomalywithin the 32-year record not belonging to the tencoldest NWUS Mays (figure 1(c)). The results infigures 1(a) and (c) suggest an association betweencold NWUS springs anomalies and a substantiallyincreased probability of summer drought and heatevents in the SGP.

To delineate the spatial characteristics of this rela-tionship, figure 2 shows temperature and precipitationdifferences between the nine coldest and nine warmestyears previously discussed. Figure 2(a) shows the MayCAMS-observed LST difference. The significant nega-tive temperature anomaly in NWUS is obvious. ColdLST anomalies in NWUS typically emerge in Marchand are well developed by April and May, ending inJune (not shown). Figure 2(b) shows observed negativeJune rainfall anomalies over the SGP and surroundingareas, from New Mexico to Texas, Oklahoma, Louisi-ana, and Mississippi, and a slightly positive rainfallband to the north. Warm temperature anomalies inJune over the Southern Plains are displayed infigure 2(c). The warm anomaly covers a much largerarea compared with the precipitation anomaly, withthe heat core further west. Note that all differences infigure 2 are statistically significant atα<0.1 level.

Figure 2(a) also shows SST differences based onReanalysis data. There is a positive correlation of 0.44(significant at α < 0.01) between May NWUS LSTanomalies and May NEP SST anomalies throughoutthe 32 study years. However, the larger LST and SSTanomaly years are not always associated with eachother. For example, figure 2(a) shows that although

the SST anomaly in NEP is consistent with a NWUSLST anomaly, it is much weaker for those years. Therewere 18 years (nine very cold and nine very warmyears) with absolute NEP SST anomalies larger than0.4 °C (about half of one standard deviation)(figures 1(b) and (d)) (which again, do not coincideentirely with the 18 years based on May LST discussedearlier). The correlation coefficient betweenMay NEPSST and June SGP precipitation for these 18 years is0.5, with statistical significance at the 0.05, similar tothe LST correlation discussed earlier. For reference,the spatial patterns of the difference between thesenine coldest and nine warmest SST years are shown infigure S1; the magnitudes of SST anomaly in theseyears are much larger than those shown in figure 2(a).The SST spatial anomaly patterns over the Pacific infigure 2(a) and S1 are consistent with the cold phase ofthe PDO pattern [30]. The warm SSTs over the Gulf ofMexico and nearby Atlantic are also apparent. Theseocean patterns are in agreement with the aforemen-tioned previous studies linking SSTs and US drought.Our empirical analysis of correlations between SST/SGP precipitation anomalies is consistent with pre-vious studies [18–20].

The concurrent correlation between May LST inNWUS and SST in NEP, as previously mentioned, hasa coefficient of 0.44.We have also computed lag corre-lation between SST in NEP and LST in NWUS. Thecorrelation coefficients between Feb SST, March SST,and April SST with May LST are only 0.12, 0.21, and0.28, respectively. The correlation coefficients dis-cussed earlier (and summarized in table 1) suggest thatLST and SST may explain a comparable fraction ofvariance in the SGP precipitation variability. More-over, previous modeling studies with specified SSTshave also consistently showed that the SST cannot pro-duce the full scope of the droughts [16], and the sever-ity and recurrence of US droughts cannot be explainedentirely by the SST fluctuations [21]. These lead us toconjecture that LST anomalies may also significantlycontribute to precipitation and heat anomalies in SGP,as it is further tested in this paper.

There are few studies linking the LST anomaly inNWUS to other variables. In addition to the associa-tion between LST and SST anomalies as discussedabove, investigations based on observational data

Table 1.Correlation coefficients between LST (K), SST (K), snow extent (%), and precipitation (mmday-1).

32 years 19 (LST)/18 SST years

May LST inNWUS vs June Precipitation in SGP 0.33* 0.54***

MAYSST inNEP vs June Precipitation in SGP 0.35** 0.50**

MayLST inNWUS vs June LST in SGP −0.46*** −0.68***

MayLST inNWUS vsMay SST inNEP 0.44***

MayLST inNWUS vsApril Snow cover inNWUS −0.36**

MayLST inNWUS vsMay precipitation inNWUS −0.44***

See text for areas over which averages were done.*, **, *** indicate statistics at 0.10, 0.05, and 0.02 significance levels, respectively.

4

Environ. Res. Lett. 11 (2016) 044018

from snow course and precipitation sites have estab-lished a close relationship between prior winter snow-fall/snow cover and spring LST in the NWUS: greater(less) snow water equivalent (SWE) and/or positive(negative) snowfall anomalies [31, 32] are associatedwith cooler (warmer) surface temperatures. Based onthe Rutgers University Global Snow Lab snow coverextent data, we find the correlation between MayNWUS LST and April snow cover extent over the areawith 118 °Was an eastern boundary, 50 °N as a north-ern boundary and the coastline as the western andsouthern boundaries, to be−0.36 at a 0.05 significancelevel. Meanwhile, the correlation coefficient betweenNWUS May LST and NWUS May precipitation is−0.44 at 0.02 significance level. A statistically sig-nificant correlation between winter precipitation andspring snow cover extent for the 18 years with largestanomalies is also found. Table 1 summarizes the cor-relations that we discuss above. A better under-standing of relationship between spring NWUS LSTanomalies and different variables, such as snow, SST,and geothermal heat flow as proposed in the early stu-dies on the subsurface soil temperature anomalies,requires a more comprehensive investigation [31–33],and is of importance in connecting winter/spring pre-cipitation to summer drought from a seasonal forecastperspective.

3.2. Approach of numerical experimentsTo further confirm whether the correlations discussedin the last section are more pronounced than expectedfrom random sampling variability, and to understandthe causal relationship between these variables, mod-eling studies are necessary. An exploratory modelingstudy using the Eta regional climate model (RCM) andthe NCEP general circulation model, which studiedthe effect of warm SUBT over the western US onsouthern US June wet condition in 1992, suggestedthat imposed initial warm spring LST and SUBTanomalies in the western US could persist and have animpact on North American June circulation andprecipitation [34]. The anomalous perturbationinduced by surface heating due to a warm springwestern LST and SUBT anomaly propagated eastwardthrough atmospheric Rossby waves within the meanwesterly flow. In addition, the steering flow contrib-uted to the dissipation of this perturbation in thenorthern US and its enhancement in the southern US.These processes led to a positive 1992 June precipita-tion anomaly in the southernUS.

The observational evidence discussed in the lastsection and our previous modeling results [34] com-pelled us to evaluate whether this relationship existedduring 2011when severe SGP drought and heat occur-red. The 2011 SGP drought intensified rapidly in latespring, and in the summer of 2011 Texas and parts ofseveral surrounding states suffered one of the worstdroughts on record in this region [3, 35, 36]. Overall,

Texas had its driest year on record (October 2010 toSeptember 2011) at 54% of normal rainfall, and thedriest summer on record, with just 62 mm of pre-cipitation. Several climate divisions in Texas andnearby states had record low Palmer HydrologicalDrought Index values within the 117-year database[37]. Meanwhile, the 2011 summer positive temper-ature anomaly of 2.98 °C above the 1981–2010 meanacross Texas was larger than the previous record dat-ing back to 1895 [3, 38]. The causes of this greatdrought are still largely unknown. It has been sug-gested that summer rainfall deficit over Texas was rela-ted to increased convective inhibition due to local soilmoisture feedback [11]. Another study links the heatin 2011 to greenhouse gas emissions but does not linkto the precipitation deficit [39]. The 2011 extreme SGPdrought and heat are highlighted in figure 1. Spring2011 had high NWUS snowfall [40] and record lowLST (−2.16 °C, figure 1(a)). The discussion above sug-gests such an extreme spring situation could con-tribute substantially to an increased probability forsevere drought in the Southern Plains.

In this study, the NCEP GFS [41] andWRF-NMMRCM [42] are employed. The GFS and WRF have thesame SST conditions and initial atmospheric and landsurface conditions, but the SST anomaly in the GFScovered the globe based on observational data. TheRCM lateral boundary conditions (LBC) wereobtained from the corresponding NCEP GFS cases.For example, the WRF LBC for the April 29, 2011initial condition was obtained from the correspondingNCEP GFS run that was also integrated starting fromApril 29, 2011. Due to known deficiencies in the globalmodel simulation, we use WRF to downscale the GFSresults, which produces better regional simulations.This issue has been discussed in [34] and will beaddressed further in the discussion section. This paperonly presents theWRFRCMresults.

Three numerical experiments (referred to as Case2011, Case SUBT, and Case SST) with the NCEP GFSand WRF are carried out to integrate the models forthree months with five different initial conditions(April 28, 29, 30, May 1, and 2) through July 31, 2011.The sample means from these five integrations withslightly different initial conditions for each Case arepresented in this paper. The 2011 drought and heatanomalies are simulated in Case 2011, which uses the2011 Reanalysis II [43] atmospheric and land data thathave abnormal cold LST and SUBTover theNWUS, asinitial conditions, as well as 2011 Reanalysis II sea iceand SST that have abnormal cold NEP SST, as bound-ary conditions for both GFS and WRF. Figure S2demonstrates that Case 2011 from the RCM run pro-duces a reasonable 2011 June drought and hot condi-tions over SGP, albeit with some biases. Spatialcorrelations over US between observations and simu-lations are 0.91 and 0.67 for temperature and pre-cipitation, respectively. The simulated Juneprecipitation and LST over SGP were 1.07 mm day−1

5

Environ. Res. Lett. 11 (2016) 044018

and 304.9 K, respectively, while the observed valueswere 1.55 mm day−1 and 301.9 K, respectively. The2011 SST, LST, SUBT, and atmospheric internal varia-bility all play a role in the simulation of Case 2011.

The other two experiments are designed to havethe same initial atmospheric and surface conditions asCase 2011, but one has warm initial LST and SUBT(Case SUBT) over NWUS, and the other has warm/

cold monthly mean SST anomalies over NEP/USCoastal-Atlantic (Case SST), anomalies which werebased on the nine warmest NWUS spring LST for CaseSUBT and nine warmest NEP spring and summer SSTfor Case SST, respectively. According to our hypoth-esis, Case SUBT and Case SST should produce wetsummer conditions in SGP due to SUBT and SSTeffect, respectively. These experiments isolate thepotential impact on SGP drought and heat due to thesefactors.

Based on the high correlation between SUBT andLST [28], observed LST was used as a reference to gen-erate the initial SUBT anomaly in Case SUBT (inabsence of large-scale observations for SUBT) as wedid in an earlier study [34].Without an imposed initialSUBT anomaly, any imposed initial LST anomalywould disappear in a couple of days due to thin surfacesoil layer in themodel. The imposed anomalous SUBTwas only used as an initial condition in Case SUBT,with SSiB land model updating the SUBT throughoutthe integrations, helping maintain SUBT anomaliesfor longer than a couple of days. Figure 3(a) shows thedifferences in initial SUBT between Case 2011 andCase SUBT. The initial SUBT difference between Case2011 and Case SUBT in figure 3(a) shows cold SUBTover NWUS. This is consistent with what is shown infigure 2(a) but with larger anomalousmagnitudes. Thenegative anomalous areas are shifted to the north. Theinitially imposed anomaly can last a little more than amonth (fromMay to early June), similar to the resultsdiscussed in [34]. Figure 3(b) shows the spatial pat-terns of May SST differences between Case 2011 and

Case SST inWRF, which are consistent with the differ-ence shown in figure S1 but of larger magnitude. TheSST in NEP in May 2011 is cool, as shown in figure 1.The warm SSTs over the Gulf of Mexico and nearbyAtlantic are also apparent infigure 3(b).

3.3. Impact of SUBT and SST anomaliesFigures 4(a), 5(a), and 6(a) show the observed differ-ences of June and July LST and June precipitationbetween 2011 and the means of nine years with thewarmest spring LST in the NWUS since 1980 (asdiscussed in previous section), respectively. They serveas observational benchmarks for comparison to themodel-simulated differences between Case 2011 andCase SUBT and between Case 2011 and Case SST,since imposed initial SUBT anomalies in Case SUBTand SST anomaly in Case SST are based on thedifference between 2011 and these years. Precipitationand LST differences over the SGP between the variouscases presented here are used to assess SGP droughtimpact due to LST and SST. It is expected that thedifference between Case 2011 and Case SUBT wouldbe comparable with the benchmark if SUBT has asubstantial impact on drought.

Due to the internal variability in model simulationsand in the realworld, plus deficiencies in current climatemodels, the model normally needs to have strong pre-scribed forcing to produce statistically significantresults. In this study, by imposing the larger spring LSTperturbations that are comparable to the differencebetween an extreme cold year (here 2011) and very hotyears over NWUS, we aim to obtain a significant Juneprecipitation difference that is comparable to the differ-ence between 2011 and wet years. Use of a relativelylarge forcing to test a newhypothesis at the preliminarilyexperimental stage and to understand mechanisms iscommon [8, 44, 45], while, in fact, our imposed LSTanomalies canbe compared to those observed.

To facilitate the comparison with the observa-tional data and to emphasize the SUBT effect, Case

Figure 3. (a) Initial SUBTdifference (°C) betweenCase 2011 andCase SUBT; (b)May SST difference (°C) betweenCase 2011 andCaseSST.

6

Environ. Res. Lett. 11 (2016) 044018

SUBT and the sum of Case SUBT+Case SST (refer-red to as Case SUBTSST) are emphasized. CaseSUBTSST does not include the interactions between

SST and SUBT, which are assumed to be secondaryeffects in this study; further tests would be needed toquantify this effect. The difference between Case 2011

Figure 4.Observed/WRF simulated anomaly/difference of surface temperature (°C) for June. (a)Observed, (b) SUBT effect, (c) SSTeffect, and (d) SST effect plus SUBT effect. The dotted areas denote statistical significance at theα=0.01 level of t-test values.

Figure 5.Observed/WRF simulated anomaly/difference of surface temperature (°C) for July. (a)Observed, (b) SUBT effect, (c) SSTeffect, and (d) SST effect plus SUBT effect. The dotted areas denote statistical significance at theα=0.01 level of t-test values.

7

Environ. Res. Lett. 11 (2016) 044018

and Case SUBTSST is expected to be closer toobserved differences shown in figures 4(a), 5(a), and6(a) if both SUBT and SST play roles in the drought.The observed anomaly and simulated June and JulyLST difference between Case 2011 and Case SUBT(referred to as SUBT effect), between Case 2011 andCase SST (referred to as SST effect), and between Case2011 and Case SUBTSST (referred to as SUBT effectplus SST effect) are shown in Figures 4–5. The areaaverages of the differences over the SGP are summar-ized in table 2.

In summer 2011, the area of very warm tempera-tures was centered over Texas and Oklahoma, andincluded western portions of Louisiana and Arkansas,southern Kansas, and eastern New Mexico [3](figures 4(a), 5(a)). Warm anomalies over the SGP inJune are generally simulated with either SUBT effect orSST effect but the warm areas shift to the west with theSUBT effect (figures 4(b) and (c)). These two forcingsseem to produce the warm anomalies over differentareas. The LST anomaly in NWUS produced the

statistically significant positive LST anomalies over westand southwest US (figure 4(b)). However, this anomalyis counteracted by negative LST anomalies caused bySST (figure 4(c)) such that only SST plus LST anomaliesproduce an adequate surface temperature anomaly pat-tern over SGP (figure 4(d)). At this point, we are unableto determine whether anomalies over west and south-west US are characteristics of NWUS LST anomalyeffects or due to model deficiencies. Further modelingand case studies are needed to investigate this issue.

The SST effect (figure 4(c)) shows statistically sig-nificant impact over the southwest US and the Mex-ican Highlands. Heating over the Mexican Highlandscoupled with the large-scale dynamical circulation canlead to heat advection over Texas and the southernGreat Plains as reported in Myoung and Nielsen-Gammon [11]. Only SUBT effect plus SST effect com-bined produce better (statistically significant) spatialdistribution of June heat anomalies over SGP(figure 4(d)) with comparable magnitude as observed(figure 4(a)). The spatial anomaly patterns in

Figure 6.Observed/WRF simulated anomaly/difference of precipitation (mm/day) for June. (a)Observed, (b) SUBT effect, (c) SSTeffect, and (d) SST effect plus SUBT effect. The dotted areas denote statistical significance at theα=0.01 level of t-test values.

Table 2.Observed differences between years 2011 and the benchmarks andWRF-NMMsimulated effects for different scenarios.

Surface Temperature (K) Precipitation (mmday-1)

May 0.57 −0.06 −0.81 −0.99 −0.48 −0.04

June 3.45 1.18 2.40* −2.31 −0.71** −1.75**

July 3.15 1.22* 2.33** −1.46 −1.05 −0.62

Averages over 88 °W–103 °Wand 29 °N–38 °N.*, **, and *** indicate statistics at 0.10, 0.05, and 0.02 significant levels, respectively.

8

Environ. Res. Lett. 11 (2016) 044018

figure 4(d) seem to be closer to those in figure 2(c) thanthe spatial patterns in 4(b) or 4(c) do. Table 2 showsthat SUBT effect and SUBT effect plus SST effect pro-duced about 34% and 70% of the observed June SGPheat anomaly, respectively. The observed July positivetemperature anomaly extends to the north, with aslight southwest-northeast tilt of the anomaly area(figure 5(a)). Case SUBTSST (figure 5(d)) properlyproduces the spatial pattern of July hot anomaly overSGP because both SUBT and SST simulate better hotanomalies during this month than during June(figure 4). The SUBT effect produces about 40% of thehot anomaly, while the SUBT effect plus SST effectproduces 74%of the anomaly.

The observed dry anomaly started to develop inMay 2011 [37]. The SUBT effect properly simulatedthe dry summer conditions over the SGP (table 2). InJune, observed dry anomaly conditions covered thesouthern US from Texas to Florida, with the droughtcentered over Texas (figure 6(a)). To the north, there

was a slight positive rainfall anomaly over the Mid-western states. The SUBT effect properly produces thedry June conditions over the SGP but with smallerextent and lower intensity (figure 6(b)), showing about31% of the actual rainfall anomaly (table 2). The com-bined effects of SUBT and SST produce both moreaccurate drought area and intensity (figure 6(d)),about 71%of the actual anomalies (table 2). Both casesalso produce wet conditions to the north. In July, theextreme dry conditions relaxed (figure S3(a)).Although Case SUBT is still able to produce thedrought conditions (table 2, figure S3(b)), it is notstatistically significant. Case SST fails to producedrought, suggesting the difficulty in accurately simu-lating the dry area when the drought has weakened(figure S3), i.e., when the signal/noise ratio is small, itis difficult to properly simulate the anomaly with sta-tistical significance.

To delineate how the hot and dry conditions devel-oped over the SGP, figure 7 shows the daily time series

Figure 7.Time series of difference betweenCase 2011 andCase SUBT averaged over the SGP for (a) vertical integratedmoisturefluxdivergence [mmday-1], (b) precipitation [mmday-1], (c) cloud cover [%], (d)net radiation [Wm−2], (e) soil wetness, and (f) surfacetemperature [°C].

9

Environ. Res. Lett. 11 (2016) 044018

of differences for several variables between Case 2011and Case SUBT averaged over the SGP—i.e., the SUBTeffect. The cold temperature anomaly in the NWUSproduces an anomalous planetary wave pattern acrossNorth America (figure S4). The negative vorticityanomaly (106 s−1) at 500 hPa along thewest coastal areaaveraged over 35 °N and 50 °Ndue to the SUBT propa-gates to the east (figure S5). Similar processes—butwithopposite anomalies—have been comprehensively dis-cussed in [34]. The vertically integrated moisture fluxdivergence (MFD) associated with negative vorticityover the SGP starts in the second half of May(figure 7(a)). Strong MFD persists in June and July,associated with an anticyclonic circulation pattern(positive geopotential anomaly) in the central US, andlow precipitation (figure 7(b)). Geopotential highs havebeen associated with previous heatwave cases in theMidwest [46] and the European heatwave of 2003 [47],tending to block the zonal circulation and producingstationary wave warm advection. Figures 7(a) and (b)also show that theMFD almost equals the precipitationreduction during late May. As progressing into Juneand July, however, the contribution of MFD to pre-cipitation reduction is reduced from 96% in May toabout 65% in June and July. Negative soil moistureanomalies (figure 7(e)) and corresponding low evapora-tion start to develop in response to the low precipita-tion. Studies have shown that a lack of rainfall in themain rainfall season extending from April–June can setup a dry soil-induced temperature positive feedbackover this region. By July local land surface temperaturefeedbacks tend to dominate over large-scale circulationacross the SGP [48].

During the same three-month time period, cloudcover is reduced accordingly (figure 7(c)), leading tomore short wave radiation reaching the ground andcontributing to positive net radiation (figure 7(d)),which combined with low evapotranspiration con-tribute to a high downward ground heat flux, leadingto high surface temperature (figure 7(f)).

4.Discussion and conclusions

While numerous studies have focused on the telecon-nection between SST anomalies and US drought, theremote effect of large-scale land temperature anoma-lies on regional droughts has been largely ignored. Ourprevious case study [34] established the theoreticalbase for this teleconnection. The observational evi-dence and modeling study presented here suggest thatthe associated strength in observations between ante-cedent early May SUBT anomaly in the NWUS andsummer precipitation anomalies over the SGP is likelylarger than random internal atmospheric variationsarising from the chaotic variability of the atmosphere;table 1 has shown that the observed correlationcoefficients for SST and for NWUS LST are compar-able. Our model study shows that the SST effect

produces 45% and LST produces 31% of the observedrainfall anomaly, respectively. The latter is smallerthan SST effects, but it is still a first order effect.Specifically, it is shown that cold spring NWUS landtemperature anomalies are a factor in increasing theprobability of dry and hot conditions in the SGP, whileatmospheric internal variability and other factors alsocontribute substantially to year-to-year variance.More case studies are needed to fully understand therole of LST effect.

Furthermore, since the May SST anomalies duringthe years with cold NWUS LST anomalies are ratherweaker compared with the LST anomalies (figure 2(a)),and since LST and SST effects show different geo-graphical characteristics [figures 4(b) and (c), 5(b) and(c), 6(b) and (c), S2(b) and (c)], the relationship betweenLST and SST and other variables needs to be furtherinvestigated. It should also be pointed out that despitethe imposed large initial LST and SUBT anomaly[figure 3(a)], the simulated May LST surface temper-ature difference between Case 2011 and Case SUBT isstill weaker than the observational benchmark (figureS6(b)), which imply that this model simulated LSTeffect could be underestimated. Since the relationshipbetween May SST in NEP and LST is not that strong(figures 2(a) and S6(c)), the cause of the deficiency inholding the May LST anomaly in simulations due toother causes, such as the shortcoming in land surfacemodel or failure in producing proper snow memoryeffect, etc, needs to be further investigated.

In this study, we find that although GCM simula-tions produced the SGP June drought and large anom-alous heat, the drought spatial patterns were not veryconsistent with the observations, and the July hot con-ditions were not reproduced (table S1). We thereforefocused on the use and analysis of RCM simulations inthis study. The improvement of regional climate simu-lations using RCM downscaling has been reviewed inother papers [49, 50].

To more comprehensively understand the role ofantecedent spring land surface temperature anomalyin the NWUS on the remote summer drought in theGreat Plains, more case studies with different modelsand external forcings from different SST, SUBT, andsoil moisture sources are necessary. To further evalu-ate the strength of this relationship and associatedmechanisms, other high profile North Americanextreme events such as the North Great Plain droughtsince 2012 (listed in the NOAA MAPP Drought TaskForce [51]) and the 2015 Texas Flood should be usedas additional case studies in future work.

Climate change and variability cause extremeregional responses at different spatial and temporalscales. Over the past 30 years, both winter and springin the NWUS have seen an increase in average pre-cipitation [52]. The NWUS spring LST anomalies aresignificantly negatively correlated to the antecedentsnow extent and precipitation (cf section 3.1), whichseems to be consistent with the intensified

10

Environ. Res. Lett. 11 (2016) 044018

precipitation decrease in Central US since 1979 [17].The analysis based on the Coupled Model Inter-comparison Project Phase 5 (CMIP5) simulations sug-gests there will be more winter precipitation in thenorthwest US with high agreement [53]. According toour and others’ analyses, this would cause a cold win-ter and spring. Another study also based on CMIP5suggests unprecedented 21st century drought risk inthe Central Plains [54]. All these results seem to sup-port our hypothesis that cold spring in NWUS has apotentially high possibility of causing a dry summer inSGP. The results of our study suggest the potential formore skillful seasonal predictions of US droughts, par-ticularly in theGreat Plains.

Acknowledgement

This workwas supported byNational Science Founda-tion grants AGS-1346813, AGS-1115506, and AGS-1540518. We thank Mr John Mioduszewski and MrThomas Estilow of Rutgers University for the snowdata. The authors also thank Professor Jared Diamondand Professor David Rigby of UCLA for their com-ments and contributions to the manuscript revisions,as well as two anonymous reviewers’ very constructivecomments/suggestions to help improve the paper.The model runs were conducted at the NationalCenter for Atmospheric Research Bluefire Supercom-puter and the Texas Advanced Computer CenterStampede Supercomputer.

References

[1] Andreadis KM,Clark EA,WoodAW,Hamlet A F andLettenmaierDP 2005Twentieth-century drought in theconterminousUnited States J. Hydromet. 6 985–1001

[2] BarlowM,Nigam S andBerbery EH2001 ENSO, pacificdecadal variability, andUS summer time precipitation,drought, and streamflow J. Clim. 14 2105–28

[3] HoerlingM,Kumar A,Dole R,Nielsen-Gammon JW,Eischeid J, Perlwitz J, QuanX, Zhang T, Pegion P andChenM2013Anatomy of an extreme event J. Clim. 26 2811–32

[4] Groisman PY andKnight RW2008 Prolonged dry episodesover the conterminousUnited States: new tendenciesemerging during the last 40 years J. Clim. 21 1850–62

[5] WoodhouseCA andOverpeck J T 1998 2000 years of droughtvariability in the central United StatesBull. Am.Meteorol. Soc.79 2693–714

[6] HoerlingM et al 2014Causes and predictability of the 2012Great Plains droughtBull. Am.Meteorol. Soc. 95 269–82

[7] Karl TR andQuayle RG 1981The 1980 summer heat wave anddrought in historical perspectiveMon.Weather Rev. 1092055–73

[8] Hong S-Y andKalnay E 2002 The 1998Oklahoma–Texasdrought:mechanistic experiments withNCEP global andregionalmodels J. Clim. 15 945–63

[9] Oglesby R J and EricksonD J III 1989 Soilmoisture andthe persistence ofNorth American drought J. Clim. 2 1362–80

[10] ZhuCM, Leung LR,GochisD,QianY and LettenmaierDP2009 Evaluating the influence of antecedent soilmoisture onvariability of theNorthAmericanMonsoon precipitation inthe coupledMM5/VICmodeling system J. Adv.Model. EarthSyst. 1 13

[11] Myoung B andNielsen-Gammon JW2010The convectiveinstability pathway towarm season drought in Texas: I. Therole of convective inhibition and itsmodulation by soilmoisture J. Clim. 23 4461–73

[12] XueY, FennessyM J and Sellers P J 1996 Impact of vegetationproperties onUS summerweather prediction J. Geophys. Res.101 7419–30

[13] TrenberthKE, Branstator GWandArkin PA 1988Origins ofthe 1988North American drought Science 242 1640–5

[14] TingMandWangH1997 SummertimeUS precipitationvariability and its relation to Pacific sea surface temperatureJ. Clim. 10 1853–73

[15] NigamS,GuanB andRuiz-Barradas A 2011Key role of theAtlanticmultidecadal oscillation in 20th century drought andwet periods over theGreat PlainsGeophys. Res. Lett. 38 L16713

[16] Schubert S et al 2009AUSCLIVARproject to assess andcompare the responses of global climatemodels to drought-related SST forcing patterns: overview and results J. Clim. 225251–72

[17] Wang S Y S et al 2015An intensified seasonal transition in theCentral US that enhances summer drought J. Geophys. Res.Atmos. 120 8804–16

[18] Rajagopalan B, Cook E, Lall U andRay BK2000Spatiotemporal variability of ENSO and SST teleconnectionsto summer drought over theUnited States during thetwentieth century J. Clim. 13 4244–55

[19] MoKC, Schemm J-K E andYoo S-H 2009 Influence of ENSOand theAtlanticmultidecadal oscillation on drought over theUnited States J. Clim. 22 5962–82

[20] Rui R andWangG 2011 Impact of sea surface temperature andsoilmoisture on summer precipitation in theUnited Statesbased on observational data J. Hydromet. 12 1086–99

[21] Seager R et al 2014Dynamical causes of the 2010/11Texas–northernMexico drought J. Hydromet. 15 39–68

[22] ChenM,Xie P, Janowiak J E andArkin PA 2002Global landprecipitation: a 50-yrmonthly analysis based on gaugeobservations J. Hydromet. 3 249–66

[23] Ropelewski C F, Jonowiak J E andHalperME1985Theanalysis and display of real time surface climate dataMon.Weather Rev. 113 1101–7

[24] XueY, Sellers P J, Kinter J L III and Shukla J 1991A simplifiedbiospheremodel for global climate studies J. Clim. 4 345–64

[25] DickinsonRE 1988The force-restoremodel for surfacetemperature and its generalization J. of Clim. 1 1086–97

[26] Higgins RWand ShiW2000Dominant factors responsible forinterannual variability of the summermonsoon in thesouthwesternUnited States J. Clim. 13 759–76

[27] GutzlerDS 2000Covariability of spring snowpack and summerrainfall across the southwestUnited States J.Clim.134018–27

[28] HuQand Feng S 2004ARole of the soil enthalpy in landmemory J. Clim. 17 3633–43

[29] RobockA,MuM,VinnikovK andRobinsonD2003 Landsurface conditions over Eurasia and Indian summermonsoonrainfall J. Geophys. Res. 108 4131

[30] Barnett T P, PierceDW, LatifM,Dommenget D andSaravananR 1999 Interdecadal interactions between thetropics andmidlatitudes in the Pacific basinGeophys. Res. Lett.26 615–8

[31] LeathersD J andRobinsonDA1983The association betweenextremes inNorth American snow cover andUnited Statestemperature J. Clim. 6 1345–55

[32] CayanDR1966 Interannual climate variability and snowpackin thewesternUnited States J. Clim. 9 928–48

[33] GuoW, Sun S andQian Y 2002Case analyses andNumericalSimulation of soil thermal impacts on land surface energybudget based on an off-line land surfacemodelAdv. Atmos. Sci.19 500–12

[34] XueY, Vasic R, Janjic Z, Liu YMandChuPC2012The impactof spring subsurface soil temperature anomaly in thewesternUS onNorthAmerican summer precipitation—a case studyusing regional climatemodel downscaling J. Geophys. Res. 117D11103

11

Environ. Res. Lett. 11 (2016) 044018

[35] PetersonTC, Stott PA andHerring S 2012 Explaining extremeevents of 2011 from a climate perspective Bull. Am.Meteorol.Soc. 93 1041–67

[36] Tadesse T,WardlowBD, Brown J F, SvobodaMD,HayesM J,Fuchs B andGutzmerD2015Assessing the vegetationcondition impacts of the 2011 drought across theUS SouthernGreat Plains using the vegetation drought response index(VegDRI) J. Appl.Meteo. andClim. 54 153–69

[37] Blunden J andDerek A S 2012 State of the climate in 2011Bull.Am.Meteorol. Soc. 93 1–282

[38] Luo L andZhangY 2012Didwe see the 2011 summer heatwave coming?Geophys. Res. Lett. 39 L09708

[39] RuppDE, Li S,MasseyN, Sparrow SN,Mote PWandAllenM2015Anthropogenic influence on the changing likelihood ofan exceptionally warm summer in Texas, 2011Geophys. Res.Lett. 42 2392–400

[40] RobinsonDA2012 Snowblitz, theUS 2010–2011 snow reportWeatherwise March/April 65 28–37 (www.weatherwise.org)

[41] KanamitsuM, Alpert J C, CampanaKA, Caplan PM,DeavenDG, IredellM,Katz B, PanH-L, Sela J andWhiteGH1991Recent changes implemented into the global forecastsystem atNMCWea. and Forecasting 6 425–35

[42] Janjic Z et al 2014User’s Guide for theNMMCore of theWeather Research and Forecast (WRF)Modeling SystemVersion 3 (http://dtcenter.org/wrf-nmm/users/)Develop-mental TestbedCenter/National Centers for EnvironmentalPrediction, Bounlder p 213

[43] KanamitsuM, EbisuzakiW,Woollen J, Yang S-K,Hnilo J J,FiorinoMandPotter GL 2002NCEP–DOEAMIP-IIreanalysis (R-2)Bull. Amer.Meteor. Soc. 83 1631–43

[44] Shukla J and Mintz Y 1982 Influence of land-surfaceevapotranspiration on the Earth’s climate Science 191498–501

[45] DickinsonRE andHenderson-Sellers AA 1988Modelingtropical deforestation: a study ofGCM land-surfaceparameterizationsQ. J. R.Meteorol. Soc. 114 439–62

[46] Kunkel K E,Changnon S S, Reike B andArritt RW1996TheJuly 1995 heat wave in theMidwest: a climatic perspective andcritical weather factorsBull. Amer.Meteor. Soc. 77 1507–18

[47] Fischer EM, Seneviratne S I, Vidale P L, LuthiD and Schar C2007 Soilmoisture—atmosphere interactions during the 2003European summer heat wave J. Clim. 20 5081–99

[48] TrenberthKE and SheaD J 2005Relationships betweenprecipitation and surface temperatureGeophys. Res. Lett. 32L14703

[49] XueY, Jajnic Z, Dudhia J, Vasic R andDe Sales F 2014A reviewon regional dynamical downscaling in intra-seasonal toseasonal simulation/prediction andmajor factors that affectdownscaling abilityAtmos. Res. 147-148 68–85

[50] Leung LR,Mearns LO,Giorgi F andWilby R L 2003Regionalclimate research: needs and opportunitiesBull. Am.Meteorol.Soc. 84 89–95

[51] Mariotti A, Schubert S,MoK, Peters-Lidard C,WoodA,Pulwarty R,Huang J andBarrieD 2013Advancing droughtunderstanding,monitoring, and predictionBull. Amer.Meteor. Soc. 94ES186–8

[52] Higgins RWandKouskyVE 2013Changes in observed dailyprecipitation over theUnited States between 1950–79 and1980–2009 J. Hydromet. 14 105–21

[53] Neelin JD, Langenbrunner B,Meyerson J E,Hall A andBergN2013California winter precipitation changes under globalwarming in the coupledmodel intercomparison project phase5 ensemble J. Clim. 26 6238–56

[54] CookB I, Ault T R and Smerdon J E 2015Unprecedented 21stcentury drought risk in theAmerican Southwest andCentralPlains Science Advances 1 e1400082

12

Environ. Res. Lett. 11 (2016) 044018

This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 76.89.166.69

This content was downloaded on 22/05/2016 at 02:03

Please note that terms and conditions apply.

Corrigendum: Spring land temperature anomalies in northwestern US and the summer

drought over Southern Plains and adjacent areas (2016 Environ. Res. Lett. 11 044018)

View the table of contents for this issue, or go to the journal homepage for more

2016 Environ. Res. Lett. 11 059502

(http://iopscience.iop.org/1748-9326/11/5/059502)

Home Search Collections Journals About Contact us My IOPscience

Environ. Res. Lett. 11 (2016) 059502 doi:10.1088/1748-9326/11/5/059502

CORRIGENDUM

Corrigendum: Spring land temperature anomalies in northwesternUS and the summer drought over Southern Plains and adjacentareas (2016Environ. Res. Lett.11 044018)

YongkangXue1,2, CatalinaMOaida2, IsmailaDiallo1, J DavidNeelin2, Suosuo Li1,3, FernandoDe Sales1,4,YuGu2, DavidARobinson5, RatkoVasic6 and LanYi7

1 Department ofGeography, University of California, Los Angeles, CA 90095,USA2 Department of Atmospheric &Oceanic Sciences, University of California, Los Angeles, CA 90095,USA3 Cold andArid Regions Environmental andEngineering Research Institute, Chinese Academy of Sciences, Lanzhou, 730000, People’s

Republic of China4 SanDiego StateUniversity, SanDiego, CA 92182,USA5 Department ofGeography, RutgersUniversity, Piscataway,NJ 08854,USA6 National Center for Environmental Prediction, College Park,MD20740,USA7 Chinese Academy ofMeteorological sciences, Beijing, 10081, People’s Republic of China

E-mail: yxue@geog.ucla.edu

Table 2 in the original version of this paper accidentlyomits row two, which explains what the variousstatistics represent (i.e. observations, SUBT effect,SUBT+SST effects). The complete version of table 2should read as below. The statistics in this table have

not changed. The table simply includes additionalheaders for clarity. This correction to table 2 does notaffect understanding the results, discussion, or conclu-sions found in the originalmanuscript.

OPEN ACCESS

PUBLISHED

20May 2016

Original content from thisworkmay be used underthe terms of the CreativeCommonsAttribution 3.0licence.

Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.

Table 2.Observed differences between year 2011 and the benchmark aswell as differences between variousWRF-NMMsimulatedcases, highlighting the SUBT and SUBT+SST effects.

Surface temperature (K) Precipitation (mm d−1)

Observation SUBT effect SUBT+SST effects Observation SUBT effect SUBT+SST effects

May 0.57 −0.06 −0.81 −0.99 −0.48 −0.04

June 3.45 1.18 2.40* −2.31 −0.71** −1.75**

July 3.15 1.22* 2.33** −1.46 −1.05 −0.62

Average area: 88 W–103 Wand 29 N–38 N.*, **, and ***: statistics at 0.10, 0.05, and 0.02 significant levels, respectively.

© 2016 IOPPublishing Ltd

Figure S1: Observed May SST difference (ºC) between 9 coldest years and the 9 warmest years based on SST Anomalies. The dotted areas denote the statistical significance at the α=0.01 level of t-test values.

CAMS Surface Temperature WRF-NMM surface Temperature

Figure S2. Mean surface temperature (ºC) and precipitation (mm day-1) for June 2011 from CAMs and GTS observations and WRF-NMM simulation.

GTS Precipitation WRF-NMM Precipitation

Figure S3. Observed/WRF simulated anomaly/difference of precipitation (mm/day) for July. (a) Observed; (b) SUBT effect; (c) SSTEffect; (d) SST effect plus SUBT effect. The dotted areas denote statistical significance at the α=0.01 level of t-test values.

a. b.

c. d.

Figure S4. Simulated 16-31 May geopotential height (GPM) difference at 500 hPa between Case 2011 and Case SUBT.

Figure S5. Perturbation pathway: temporal-zonal cross section of vorticity anomaly (106 s-1) at 500 hPa averaged over 35N and 50N.

Figure S6. Observed/WRF simulated anomaly/difference of surface temperature (ºC) for May. (a) Observed; (b) SUBT effect; (c) SST Effect. The dotted areas denote statistical significance at the α=0.01 level of t-test values.

a. b.

c.

Table S1. Observed differences between year 2011 and the benchmarks, and GCM-simulated differences for different scenarios