Observations of the ABL Structures over a Heterogeneous ...opensky.ucar.edu/islandora/object/articles:6962/datastream/PDF/download... · from the Honeywell Laseref SM Inertial Reference

Observations of the ABL Structures over a Heterogeneous Land Surfaceduring IHOP_2002

SONG-LAK KANG AND KENNETH J. DAVIS

Department of Meteorology, The Pennsylvania State University, University Park, Pennsylvania

MARGARET LEMONE

Mesoscale and Microscale Meteorology Division, National Center for Atmospheric Research, Boulder, Colorado

(Manuscript received 2 November 2005, in final form 31 July 2006)

ABSTRACT

This study analyzes data collected by aircraft and surface flux sites over a 60-km north–south-orientedaircraft track for five fair-weather days during the International H2O Project (IHOP_2002) to investigatethe atmospheric boundary layer (ABL) structures over a heterogeneous land surface under different back-ground weather conditions. The surface skin temperature distribution over the aircraft track in this case ismostly explained by the soil thermal properties and soil moisture, and corresponds to the observed ABLdepths except one day having a weak surface temperature gradient and a weak capping inversion. For theother four days, the blending height of the surface heterogeneity likely exceeds the ABL depth and thus theABL establishes equilibrium with local surface conditions.

Among the four days, two days having relatively small Obukhov lengths are evaluated to show thebackground weather conditions under which small-scale surface heterogeneity can influence the entireABL. In fact, on one of these two days, relatively small-scale features of the surface temperature distribu-tion can be seen in the ABL depth distribution. On the two small Obukhov length days multiresolutionspectra and joint probability distributions, which are applied to the data collected from repeated low-levelaircraft passes, both imply the existence of surface-heterogeneity-generated mesoscale circulations on scalesof 10 km or more. Also on these two small Obukhov length days, the vertical profiles of dimensionlessvariances of velocity, temperature, and moisture show large deviations from the similarity curves, which alsoimply the existence of mesoscale circulations.

1. Introduction

Predicting the timing and location of convectivecloud development is a fundamental challenge in thestudy of meteorology. Spatial variability in the atmo-spheric boundary layer (ABL) is closely related to thedevelopment of moist convection (e.g., Crook 1996;Rozoff et al. 2003; among many papers in the litera-ture). The heterogeneity of the ABL is known to becaused by multiple factors including surface forcing andabove-ABL conditions. The influence of land surfaceheterogeneity on the ABL structure and moist convec-tion has been studied by numerous investigators usingnumerical models and observational data (for a com-prehensive review, see Pielke 2001).

Mesoscale numerical model studies of the effects ofland surface heterogeneity on the ABL structure andmoist convection have focused on the generation ofwell-organized circulation with prescribed surface sen-sible and latent heat fluxes at the meso-� scale, whichranges from 20 to 200 km (e.g., Segal et al. 1989; Chenand Avissar 1994a,b to name a few). Much of the earlierresearch involved idealized conditions, while Zhongand Doran (1998) used a real-data case and found thatheterogeneous conditions result in a very similar situa-tion as that simulated with homogenous surface condi-tions. These results suggest that well-organized meso-�-scale circulations generated by land surface hetero-geneity may be difficult to find outside of areas ofextreme contrasts, such as on the boundary of irrigated/nonirrigated, urban/rural, or bare/vegetated ground.

To fully simulate the effect of surface heterogeneityon the ABL structure at the meso-� scale (2–20 km),large-eddy simulations (LES) have been used (e.g.,Shen and Leclerc 1995; Avissar and Schmidt 1998; Pat-

Corresponding author address: Song-Lak Kang, Department ofMeteorology, The Pennsylvania State University, 503 WalkerBuilding, University Park, PA 16802.E-mail: [email protected]

APRIL 2007 K A N G E T A L . 221

DOI: 10.1175/JHM567.1

© 2007 American Meteorological Society

JHM567

ton et al. 2005 to name a few). Many LES results haveindicated that meso-�-scale land surface heterogeneitycan significantly influence the entire ABL structure.Most of the LES results, however, are limited by theiruse of idealized conditions for surface heat and mois-ture fluxes, mean winds, temperature and moisturesoundings, lateral boundary conditions, and terrain.

There have been some observational studies that ad-dress the influence of land surface heterogeneity at themeso-� scale on the ABL structure including existenceof mesoscale circulations generated by land surface het-erogeneity (e.g., Mahrt et al. 1994a; Shaw and Doran2001; LeMone et al. 2002, 2007). However, compared tothe numerical studies, there have been fewer observa-tional studies. This is primarily due to the difficulty inobtaining appropriate observations that describe theABL heterogeneity in a statistically robust manner.However, improvement in technology is expanding ourability to observe the heterogeneous ABL structure.

The International H2O Project (IHOP_2002; Weck-werth et al. 2004), in which multiple research com-ponents utilized various enhanced observational in-struments, provides an opportunity to observe theheterogeneous ABL structure. Observational datasetscollected in this field experiment can be used to assistthe research toward a better understanding of the ABLstructure over a heterogeneous land surface.

The goal of this study is to document the influence ofmeso-�-scale land surface heterogeneity on the ABLstructure under different background weather condi-tions using observational datasets collected fromIHOP_2002. This research includes the investigation ofthe primary causes of surface heterogeneity over thestudy area. Datasets used for this research will be de-lineated in the next section, along with a description ofthe study area. Section 3 will discuss the causes of thesurface heterogeneity over the study area mainly usingthe measurements at the surface flux sites. Section 4will describe the influence of land surface heterogeneityon the low-level ABL structure under different back-ground weather conditions by applying the statisticaland spectral analysis methods to the data from the re-peated low-level aircraft passes. In section 5, the influ-ence of the land surface heterogeneity on the upper-level ABL structure is qualitatively assessed based onnormalized variance profiles and observed spatialvariations of the ABL depth. Section 6 summarizes theresults and presents conclusions.

2. Data

The IHOP_2002 field experiment was conductedfrom 13 May to 25 June 2002 in the southern GreatPlains (SGP) of the United States with the aim of im-

proving our understanding of convective precipitation,including the development of heterogeneous boundarylayer (BL) structure that might contribute to this un-derstanding (Weckwerth et al. 2004). Numerous mobileobserving systems, ground-based observation facilities,and six research aircraft were used in this project.

As a part of the IHOP_2002 field experiment, theUniversity of Wyoming King Air (UWKA) flew overthree tracks—the western track, the central track, andthe eastern track—to obtain data for investigating theinfluence of heterogeneous land surface on the daytimeABL structure. [In this study, we will focus on the west-ern track. For more information on the other twotracks, one can refer to LeMone et al. (2007) and/orWeckwerth et al. (2004)]. The UWKA was equippedwith sensors that measure the state and motion of theABL. Detailed information on the instrumentation ofthe UWKA is available online at http://flights.uwyo.edu/base. Thus, only the aircraft-borne instruments thatprovide the data used in this study are briefly intro-duced.

For the aircraft’s position and motion, the outputsfrom the Honeywell Laseref SM Inertial ReferenceSystem (IRS) were corrected based on the global posi-tioning system (GPS). For the measurements of staticpressure, temperature, and surface radiation tempera-ture, a Rosemount 1201, a reverse-flow thermometer,and a downward-looking Heiman KT-19.85 radiometerwere used, respectively. Potential temperature was de-rived from the measurement of static pressure by aRosemount 1201 and from the measurement of tem-perature by a reverse-flow thermometer. For the watervapor mixing ratio, a Lyman-alpha hygrometer, whoselong-term mean is adjusted to that of a chilled-mirrorhygrometer, was used. Since the measurement instru-ments on the aircraft were not collocated, the lags be-tween the sensors have been taken into account(LeMone et al. 2003, 2007). Thus, based on the peaks ofthe cross correlations between the measured variables,the observations of static pressure, temperature, poten-tial temperature, and water vapor mixing ratio are ad-justed to reconcile a 0.08-s lag as compared to the windobservations.

The western track, a 60-km north–south-orientedstrip from approximately 36.4° to 37.0°N along 100.6°W(Fig. 1), shows the most heterogeneous surface tem-perature distribution along with the most sparse veg-etation distribution as illustrated in Fig. 9 of Weck-werth et al. (2004). Over this track, the UWKA flew atotal of 61 legs on the five boundary layer heterogeneity(BLH) mission days: 19, 20, 25A, 25B, 29 May, and 7June 2002 (see Table 1 for detailed descriptions). Of the61 legs, 30 are low-level legs, where the aircraft altitude

222 J O U R N A L O F H Y D R O M E T E O R O L O G Y VOLUME 8

was approximately 65 m above ground level (AGL).Those repeated low-level passes were done to obtainstatistically reliable measurements (see Mann andLenschow 1994). The observations were recorded at 25Hz, which corresponds to a spatial resolution of about3 m, between approximately 1700 and 2000 UTC (1100and 1400 CST). General information regarding theUWKA flights and the ABL characteristics on the fivedays is summarized in Table 1. In addition, to charac-terize the ABL on each day, the vertical soundings ofpotential temperature and water vapor mixing ratiofrom the rawinsonde released over Homestead site

(Fig. 1; 36.55°N, 100.6°W) at about flight start time areplotted in Fig. 2.

The surface elevation of the track from U.S. Geo-logical Survey (USGS) global 30-s dataset varies withina range of 150 m (Fig. 1). The lowest elevation region,which is around 36.83°N, is occupied by the BeaverRiver Floodplain that is about 2 km wide. The vegeta-tion cover over the track is estimated by the normalizeddifference vegetation index (NDVI; LeMone et al.2007), which was measured from a downward-lookingradiometer during the low-level passes of the UWKAflights (Fig. 1). The NDVI indicates sparse vegetationthat became greener with time along the track exceptfor the relatively green vegetated area that covers ap-proximately 5 km of the track in the vicinity of theBeaver River. Thus, both terrain elevation and vegeta-tion cover likely make up a relatively homogenous sur-face condition. However, as shown in Fig. 1, the StateSoil Geographic (STATSGO) map exhibits varying soiltypes along the track.

Over the track on the five BLH days, it was fairweather. The direction of mean wind at 65 m AGL wasusually southerly except on 25 May and the mean windmagnitude varied from 2.5 to 13.2 m s�1 (see Table 1).For each low-level leg, the relative importance of buoy-ant production compared to mechanical production ofturbulence is compared through Obukhov length (Stull1988). Here, Obukhov length is obtained from leg-averaged values of momentum flux, buoyancy flux,and virtual potential temperature. The range of theObukhov length calculated for each low-level leg on agiven day is shown in Table 1. Despite their larger sur-face heat fluxes, 19 and 20 May have relatively largeObukhov lengths due to the strong mean winds. Con-versely, 25 and 29 May have relatively small Obukhovlengths due to the relatively weak mean winds althoughthe surface sensible heat fluxes are smaller than thoseon the previous two days.

In addition to UWKA, other research instrumentswere concentrated around the track during IHOP_2002(for the detailed experimental array, see Fig. 4 ofWeckwerth et al. 2004). Among them were the threeIntegrated Surface Flux Facility (ISFF; Chen et al.2003) sites, and the differential absorption lidar (DIAL;Poberaj et al. 2002) aboard the Deutsche Luft- undRaumfahrt (DLR) Falcon.

The ISFF sites 1, 2, and 3 were deployed along thetrack from south to north in ascending order (Fig. 1).Surface skin temperatures measured by the Everest4000.4GL infrared surface temperature sensors at thethree ISFF sites are used for the comparison with thoseobserved by the UWKA. The tower-based eddy-covariance flux measurements are also compared with

FIG. 1. (top) Topography along the western track of the UWKAduring IHOP, which comes from USGS global 30-s dataset.(middle) Composite structure of NDVI with 4-km means at every1 km from the repeated passes of the aircraft at 65 m AGL at agiven day: 0519 (solid), 0520 (dotted), 0525A (dashed), 0525B(dash–dotted), 0529 (dash–two-dotted), 0607 (long dashed). (bot-tom) STATSGO soil map around the western track. The locationsof the ISFF sites 1 (rectangle), 2 (circle), and 3 (triangle) aremarked. The symbol X indicates the location of the Homesteadprofiling site. The color bar indicates the soil categories: 1 (sand),2 (loamy sand), 3 (sandy loam), 4 (silt loam), 5 (silt), 6 (loam), 7(sandy clay loam), 8 (silty clay loam), 9 (clay loam), 10 (sandyclay).


the airborne eddy-covariance measurements. In addi-tion, the surface energy balance (SEB) components,surface skin temperature, and soil temperature were allmeasured at the three ISFF sites.

Some segments of the UWKA flights were flownconcurrently with north–south transects of the DLRFalcon. From the aerosol backscatter data of the DIALaboard the DLR Falcon, the ABL depths over the track

FIG. 2. Vertical profiles of (a) potential temperature and (b) water vapor mixing ratio fromthe rawinsonde released over the Homestead profiling site (Fig. 1; 36.55°N, 100.6°W) at 1740,1731, 1733, 1741, and 1856 UTC 19 May, 20 May, 25 May, 29 May, and 7 Jun 2002, respec-tively.

TABLE 1. Information regarding the UWKA flights and the ABL characteristics over the western track at the five days. Here Lshows the range of the Obukhov length calculated for each low-level leg.

Date

The UWKA flights The ABL characteristics

Flight time(HHMM:SS)

No. of low-levellegs/No. oftotal legs

Mean winda

direction (°)/speed (m s�1) �L (m)

Heat fluxb

(top/surface)(K m s�1)

Moisture fluxb

(top/surface)(g kg�1 m s�1)

0519 1703:09–1920:22 6/10 177/13.2 82–202 �0.14/0.29 �0.03/0.030520 1722:36–1937:58 2/10 158/13.1 101–129 �0.05/0.30 0.05/0.020525 1701:10–2021:23 10/13 137/2.2 0–23 �0.01/0.19 0.10/0.030525Ac 1701:10–1850:10 5/7 148/1.1 0–220525Bc 1852:42–2021:23 5/6 125/3.4 1–230529 1640:07–2024:05 5/15 191/4.9 1–82 �0.03/0.13 0.22/0.100607 1650:37–1958:59 7/13 199/10.6 53–148 �0.07/0.18 0.46/0.05

a This mean wind is computed from the low-level legs at each case.b These fluxes of heat and moisture at the surface and the ABL top are estimated based on linear fits to the composite vertical profiles

of the sensible and latent heat fluxes from the in situ measurements at surface flux sites and aircraft.c Note that 25 May is sometimes separated into 25A and 25B May due to the slightly eastward shifted flight track for the last five legs.


were objectively determined by using the Haar wavelettechnique of Davis et al. (2000) at a horizontal resolu-tion of 700 to 1000 m.

3. Surface heterogeneity

a. Surface skin temperature

The surface skin temperature at a given location, de-fined as the temperature at the air–soil interface, de-pends on the radiation balance, near-surface atmo-spheric exchange processes, presence of vegetationcover, and thermal properties of the subsurface me-dium (Arya 2001). Over surfaces with simple vegeta-tion or bare soil, surface skin temperature can be usedas an indicator for thermal heterogeneity of land sur-face (Mahrt 2000). The vegetation along the track wassparse, with the soil sometimes visible from the aircraft,consistent with the low NDVI values observed alongthe track (Fig. 1). Thus, for the five fair-weather days,this surface skin temperature distribution may repre-sent the surface thermal heterogeneity over the aircrafttrack. Figure 3 shows the surface temperature distribu-tion, which is the composite of overlapping 4-km means

at every 1 km from the radiometric surface tempera-tures measured by repeated low-level passes of theUWKA on a given day. Surface skin temperatures arealso measured at the three ISFF sites, and the dailyaverages covering the aircraft flight hours are shown inFig. 3.

In Fig. 3, except on 25 May, the surface temperaturedistributions, which are consistent with those of surfaceheat flux and potential temperature at 65 m AGL (seesection 4a), are generally characterized by warmer con-ditions in the north and cooler conditions in the southwith the largest horizontal gradient of 12°C (50 km)�1

on 29 May and the smallest gradient of 5°C (50 km)�1

on 7 June. 25 May differs from the other days by havingan area of higher temperatures around 36.64°N with asurface temperature gradient of �6°C (15 km)�1 to thenorth and �8°C (15 km)�1 to the south. The differentpattern on 25 May is caused by more heating over theregion around site 2 (36.62°N), which is likely associ-ated with relatively small soil heat flux and weak meanwinds (see sections 3d and 3e). The high temperatureover the entire leg on 25B May is due to the late ob-servation hours, from 1852 to 2021 UTC (Table 1). InFig. 3, one can also notice that the surface temperaturesat about 36.83°N where the Beaver River is located are0.5°–1.5°C lower than those in surrounding areas allfive BLH days.

b. Rainfall, soil moisture, surface skin temperature,and soil temperature

Figure 4 shows the daily mean values of rainfall,volumetric soil moisture, surface skin temperature, andsoil temperature at a depth of 5 cm measured at theISFF sites 1, 2, and 3 from 13 May to 7 June 2002. Therewere four rainfall events: on 17 May less than 25 mm,on 24 May 3 mm, on 27 May 20–30 mm toward thenorth and over 80 mm to the south, and on 5 June 15–20mm. Although the amount of rainfall differs widely, thefour rainfall events always noticeably decrease bothsurface skin temperature and soil temperature. In spiteof the variations in the surface environment, the surfaceand soil temperature at site 1 persistently gave the low-est values of any site, which is reflected also in theaircraft-measured surface radiation temperature (Fig.3). We attempt to explain these persistent cooler con-ditions at site 1 in section 3d.

The significance of soil moisture in the interactionbetween land surface and atmosphere has been empha-sized by numerous researchers (e.g., Chen and Avissar1994a,b among many papers in the literature). The ob-served volumetric soil moistures at the three sites indi-cate that soil hydraulic properties, along with the dif-ferent characteristics depending on surface cover (e.g.,

FIG. 3. Composite structure of surface radiation temperaturewith 4-km means at every 1 km from the repeated passes of theaircraft at 65 m AGL at a given day. The rectangles, circles, andtriangles indicate the averages of the surface skin temperaturesmeasured at ISFF sites 1, 2, and 3 over the aircraft flight hours atthe written day beside the symbols. There were no surface radia-tion temperatures available on 7 June from the UWKA due to themalfunction of the measurement instrument (downward-lookingHeiman KT-19.85 radiometer).


runoff and infiltration), need to be considered to ex-plain the relationship between the spatial variability inrainfall and soil moisture. At site 2, the rainfall of 36mm [maximum rainfall rate of 10 mm (5 min)�1] on 27May caused an increase of 29% in the 5-cm volumetricsoil moisture. However, at site 1 the 5-cm volumetricsoil moisture increased only 19% in spite of a totalrainfall of 84 mm [maximum rainfall rate of 4.2 mm (5min)�1]. The difference in soil hydraulic properties be-tween site 1 and site 2, a spatial difference of around 17km, was seen again on 5 June (Fig. 4). On this day, the12-mm rainfall at site 1 [maximum rainfall of 1.4 mm (5min)�1] caused almost no change, but the 14-mm rain-fall at site 2 [maximum rainfall of 1.3 mm (5 min)�1]increased the volumetric soil moisture by 6%.

c. Sensible and latent heat fluxes

The relationship between aircraft-measured andtower-measured sensible and latent heat fluxes isshown in Fig. 5. First, for the UWKA measurement,

leg-averaged sensible and latent heat fluxes are com-puted for each low-level leg,

SH � ��CP��w��, �1�

LE � ��L��w�q��, �2�

where � � indicates leg average, is air density derivedfrom the aircraft-measured variables using the ideal gaslaw, CP is specific heat of moist air, and L is latent heatof vaporization computed as a function of temperature.The perturbations of vertical velocity (w�), potentialtemperature (��), and water vapor mixing ratio (q�) aredefined from the leg average as

�� , �3�

where represents one of the variables. The dailymean values of sensible and latent heat flux are ob-tained by averaging the leg-averaged values over therepeated low-level aircraft passes on a given day. Sec-ond, the sensible and latent heat fluxes measured ateach site during the aircraft flight are averaged to esti-

FIG. 4. Daily mean values of precipitation (solid line and filled symbol), volumetric soilmoisture (dotted line and filled symbol), surface skin temperature (solid line and unfilledsymbol), and soil temperature at a depth of 5 cm (dotted line and unfilled symbol) measuredat the ISFF sites 1 (square), 2 (circle), and 3 (triangle) from 13 May to 7 Jun 2002.


mate the surface fluxes on a given day. The sensible andlatent heat fluxes averaged over the three sites are usedfor the comparison with those of the UWKA. Figure 5shows that aircraft measurements give smaller valuesthan tower measurements by about 10%–20% for bothsensible and latent heat fluxes, except for the cases of29 May and 7 June. On 29 May, the aircraft latent heatflux is 30% higher than the tower measurements whilethe aircraft sensible heat flux is 10%–20% lower thanthe tower measurement. The sensible heat fluxes mea-sured by an aircraft are expected to be smaller thanthose measured by a tower due to the aircraft flightheight and the linear decrease of sensible heat flux withheight to a negative value at the ABL top (Stull 1988).The latent heat fluxes measured by an aircraft are oftenhigher than the surface fluxes measured by a tower. On7 June, the aircraft-measured latent heat flux exceedstower-measured latent heat flux by as much as 85%while the tower-measured sensible heat flux exceedsaircraft-measured sensible heat flux by 30%. This sig-nificant overestimation of aircraft-measured latent heatflux is likely caused by the considerable height depen-dence of this flux on this day (see Table 1 and section6c).

d. Surface energy balance

For the three surface flux sites, the diurnal curves ofnet radiation (Rnet), sensible heat flux (H), latent heatflux (LE), and heat flux into the soil (Gs) from 1200 to0000 UTC (from 0600 to 1800 CST) averaged over thefive days are shown in Fig. 6. In Fig. 6a, Rnet at site 2is larger than those at sites 1 and 3. The smaller Rnet atsite 3, compared with the Rnet at site 2, is likely causedby more frequent cloudiness based on the observedsmaller incoming shortwave radiation during the day-time. However, the smaller Rnet at site 1 is likely as-sociated with a relatively large albedo. The lower sur-face temperature (Figs. 3 and 4) indicates smaller up-welling longwave radiation (Q↑

LW). Thus, based onRnet � Qs (1 � A) � Q↓

LW�Q↑LW (Pielke 2001; Arya

2001; Stull 1988) with the assumption of no significantspatial variability in downwelling longwave radiation(Q�

LW) and insolation (Qs) between the sites, thesmaller Rnet at site 1 implies a relatively large albedo(A) at site 1.

The daytime available energy (Rnet � Gs) for thesensible and latent heat fluxes (H � LE) is estimatedbased on the closure of surface energy balance, Rnet �Gs � H � LE, which is balanced to within 6% residualin this dataset. In addition to the most net radiation(Fig. 6a), the smallest soil heat flux (Fig. 6d) creates themost available energy at site 2. Here, we estimate soilthermal conductivity, which determines the soil heat

FIG. 5. (a) Relationship between averages of sensible heat fluxesderived from the data collected by the repeated low-level aircraftpasses vs average of the sensible heat fluxes measured at the threeISFF sites during the aircraft flights. (b) Same as in (a) but for latentheat fluxes. Unfilled star denotes the relationship between the air-craft values and the spatial averages of the values at the three sites.The error bar over the star indicates standard error defined as thestandard deviation between the low-level legs divided by the squareroot of the number of the low-level legs (Mahrt 1998). Here standarderror is computed by dividing std dev by the square root of thenumber of the repeated low-level passes. Square, circle, and trianglerepresent the relationship between the aircraft values and the valuesmeasured at sites 1, 2, and 3, respectively.


flux, with thermal diffusivity and heat capacity (Arya2001; Stull 1988). First, soil thermal diffusivity is esti-mated from the theoretical solution of thermal wavepropagation in soil (Arya 2001; Stull 1988) by using theamplitudes of the changes in soil temperature measuredat 5-cm depth and in surface temperature. Second, ateach site, soil heat capacity is estimated by using Cs �s /m � 1.9 � 106 � Qs � 4.2 � 106 (where Qs isvolumetric soil moisture) with the value of soil bulkdensity (s) at each site and the density of mineral par-ticles (m), which are obtained from the ISFF/IHOP_2002 Web page (www.atd.ucar.edu/rtf/projects/ihop_2002/isff/). In Table 2, the soil thermal conductiv-ity is usually the lowest at site 2, which is of courseconsistent with the smallest soil heat flux (Fig. 6d).Thus, although its soil moisture is always the highest,site 2 is warmer than site 1 (Figs. 3 and 4). Under rela-tively weak mean wind on 25 May, this local feature atsite 2 is quite manifest in surface temperature distribu-

tion (Fig. 3). The small thermal conductivity indicatesthat thermal exchanges are concentrated near the up-permost surface. This surface, with a shallow activethermal-exchange layer, experiences extreme diurnaltemperature fluctuations (Zhang and Huang 2004).

At site 3, where the soil moisture at 5 cm is the lowest(Fig. 4), the dry soil causes most of the available energyto be used for the surface sensible heat flux, whichimplies the lowest values of the surface latent heat flux

TABLE 2. Estimated soil thermal conductivity (W m�1 K�1).

19 May 20 May 25 May 29 May 7 June

Site 1 0.57 0.41 0.75 (6.3 � 104)* 1.34Site 2 0.23 0.21 0.30 1.29 0.63Site 3 0.39 0.34 0.61 0.77 0.60

* This abnormally high value comes from soil thermal diffusivity,which is likely associated with the increased volumetric soilmoisture measured at this time.

FIG. 6. The surface energy balance components measured at the ISFF sites 1 (solid), 2(dotted), and 3 (dashed) averaged for the five days. Rnet, H, LE, and Gs represent netradiation, sensible heat flux, latent heat flux, and soil heat flux at a depth of 5 cm. Here, thesoil heat flux was measured with a heat flux plate at a depth of 5 cm. For Rnet and Gs, apositive sign is used for downward flux. However, for H and LE, a negative sign is used forupward flux.


(Fig. 6c). Thus, with the available energy similar to thatat site 1, site 3 shows the highest values of the surfacesensible heat flux (Fig. 6b) and the surface temperatureamong the three ISFF sites (Figs. 3 and 4). On thecontrary, site 1, which distributes a significant amountof the available energy to the latent heat flux (Fig. 6c),shows the lowest surface heat flux (Fig. 6b) and thesurface temperature (Figs. 3 and 4).

4. Stationary spatial variability

a. Decomposition

The stationary spatial variability of the flow and ver-tical fluxes at 65 m AGL is described by applying thedecomposition method of Mahrt et al. (1994b) to thedata collected from repeated aircraft passes. First, for aparticular observed variable, , the ith segment averageover a fixed-length window in the jth leg, [ ]i, j, is com-puted. From the segment averages over the jth leg, theleg average, � �j, is obtained. The spatial deviation of

the ith segment in the jth leg is defined from the legaverage

�i, jS � ��i, j � ��j. �4�

Here, time averaging the spatial deviations over all therepeated legs at a fixed location yields the stationarypart of the spatial deviation

��S�i �1J �

j�1

J

�i, jS , �5�

where {} indicates a time average and J is the number ofthe repeated aircraft passes. From this stationary partof the spatial deviation, the transient part of the spatialdeviation is defined as

�i, jST � �i, j

S � ��S�i. �6�

Figure 7 shows the stationary spatial deviations, Eq.(5), of potential temperature and water vapor mixingratio at 65 m AGL over the track. Consistent with the

FIG. 7. Composite structures of potential temperature, {�S}, and water vapor mixing ratio,{qS}, using 4-km means at every 1 km from the repeated passes of the aircraft at 65 m AGLover the western track on (a) 19 May, (b) 20 May, (c) 25A May, (d) 25B May, (e) 29 May, and(f) 7 Jun 2002. The error bar indicates standard error defined as the standard deviationbetween the low-level legs divided by the square root of the number of the low-level legs(Mahrt 1998).


spatial variability of surface skin temperature (Fig. 3),the potential temperature at 65 m has the strongestgradient on 25A May, 1.2 K (22 km)�1 between 36.57°and 36.77°N, and the weakest gradient on 7 June, 0.2 K(42 km)�1 between 36.50° and 36.88°N. On 25A May,the water vapor mixing ratio has a gradient of 0.45 gkg�1 (24 km)�1 between 36.60°N around the warmerregion, and 36.82°N around the cooler region, whichcreates warm–dry and cool–moist conditions. Similarlyon 29 May, the potential temperature increases at a rateof 1.1 K (48 km)�1 from south to north whereas thewater vapor mixing ratio decreases at a rate of 2.0 gkg�1 (48 km)�1, which also creates warm–dry and cool–moist conditions.

Figure 8 exhibits the stationary spatial deviations of and u winds. Because the mean wind is within 20° oftrue south (see Table 1), is approximately the along-wind component and u is approximately the crosswindcomponent for all days except 25 May. Excluding 25B

May and 7 June, the magnitude of the horizontal windis lower over more heated area (see Fig. 3). The sta-tionary spatial deviations of vertical heat and moisturefluxes based on the leg averages at 65 m AGL areshown in Fig. 9. The vertical heat flux shows a corre-spondence with surface temperature, although less ob-vious than the potential temperature at 65 m.

b. Variance decomposition

To quantify the relative importance of stationary spa-tial variance, which is closely associated with surfaceheterogeneity to temporal and transient variances, weemployed the variance decomposition of Mahrt et al.(1994b):

��[�]2�� 2� � ��S�2� � ��ST�2��. �7�

Here, the global average, {�[ ]�}, was previously re-moved from the segment averages when a formula forvariance was created. For example, the leg average in

FIG. 8. Composite structures of along-track wind, {S}, and cross-track wind, {uS}, using 4-kmmeans at every 1 km from the repeated passes of the aircraft at 65 m AGL over the westerntrack on (a) 19 May, (b) 20 May, (c) 25A May, (d) 25B May, (e) 29 May, and (f) 7 Jun 2002.The error bar indicates standard error defined as the standard deviation between the low-levellegs divided by the square root of the number of the low-level legs (Mahrt 1998).


the first term on the right-hand side, � �, is obtainedfrom the segment averages by removing the global av-erage. Thus, the terms on the right hand side in Eq. (7)were named as temporal, spatial, and transient vari-ances, respectively (Mahrt et al. 1994b). We used thisvariance decomposition with the five different segmentlengths of 1, 4, 6, 8, and 12 km to evaluate the relativesignificance of spatial variance as a function of spatialscale. Thus it can be determined at which scales thespatially stationary patterns are the most significant ona given day. This variance decomposition is applied topotential temperature, water vapor mixing ratio, windcomponents, and vertical heat and moisture fluxes.Here, the deviations for vertical flux of a scalar havebeen computed relative to the leg averages:

w�� w � �w�� , �8�

where is potential temperature or water vapor mixingratio. To avoid the complication in the interpretation

that arises by introducing a different observation periodand a different number of repeated flight passes, thefive low-level legs between 1700 and 1900 UTC (1100and 1300 CST) on 19 May, 25A May, and 7 June areused.

The variance decomposition results are shown in Fig.10. For the potential temperature and the along-windcomponent, the spatial variance normalized by totalvariance reaches values greater than 0.5 on 19 and 25AMay, implying significant horizontal variability on bothdays. Although temporal and transient variances arenot insignificant for the vertical heat and moisturefluxes, stationary spatial variances exceed temporal andtransient variances at 12 km on 19 May and at 8 km on25A May. However, on 7 June, stationary spatial vari-ance for all the variables is smaller than temporal andtransient variances at all the segment lengths, whichimplies that the influence of land surface variability isrelatively insignificant.

FIG. 9. Composite structures of vertical heat fluxes, {(w��)S}, and moisture fluxes, {(w�q�)S},using 4-km means at every 1 km from the repeated passes of the aircraft at 65 m AGL overthe western track on (a) 19 May, (b) 20 May, (c) 25A May, (d) 25B May, (e) 29 May, and (f)7 Jun 2002. The error bar indicates standard error defined as the standard deviation betweenthe low-level legs divided by the square root of the number of the low-level legs (Mahrt 1998).


5. Low-level ABL properties

a. MR spectra

Multiresolution (MR) spectrum is an alternativemethod to Fourier spectrum and a direct link to thedefinition of Reynolds averaging because it uses awavelet basis set with a constant basis function. Onecan refer to Vickers and Mahrt (2003) for a detaileddescription of the MR spectrum. We use these MRspectra to estimate the relative significance of the con-tribution of variance (and covariance) at a segmentlength in comparison to the total values, and to identifythe physical processes of the vertical heat and moisturefluxes at each segment length in the low-level ABLover the heterogeneous land surface.

Figures 11, 12, and 13 show the composites of the MRspectra (or cospectra) normalized by the total variance(or covariance) of each variable, namely, {Cw (2m)},where {} implies averaging over all the low-level legs ona given day and 2m indicates a segment length of 2m

points. In Figs. 11, 12, and 13, the segment lengths of 2m

points are converted into meters by using Lm � 2m �uac /f, where uac is true airspeed, which was about 80m s�1 and f is the sampling frequency of 25 Hz.

The composites of the normalized MR spectra of po-tential temperature (C��) and water vapor mixing ratio(Cqq) are shown in Fig. 11. For potential temperatureand water vapor mixing ratio, turbulence and meso-scale variances can be separated based on the spectralgap defined approximately between 3 and 15 km. On 25and 29 May, the peaks of mesoscale variances weremore significant than those of turbulence variancesboth for potential temperature and water vapor mixingratio. However, the composites of the normalized MRspectra of wind components (Cuu, C , Cww) shows thespectral gap being about 10 km only in the along-trackwind spectra (Fig. 12). That is to say, the spectra ofcross-track wind and vertical wind have no spectral gapand mesoscale variance peak. Even for the along-trackwind spectra, only on 25A May is the mesoscale vari-ance peak relatively significant compared to the turbu-lent variance peak. Thus only on 25A May the higherpeaks of mesoscale variances of potential temperatureand mixing ratio can be linked to the more significantpeak of mesoscale along-track wind variance.

The composites of the cospectra of w� and ��, w� andq�, and q� and �� normalized by the total covariance,{Cw�}, {Cwq}, and {Cq�}, respectively, are shown in Fig.13. In Table 3, the combinations of total covarianceswith differing signs are summarized into four modeswith accompanying physical interpretations. On 19, 20,and 25B May, heating and moistening from the surface,mode I, accounts for 97%, 95%, and 91% (95%, 98%,and 94%) of the total vertical heat (moisture) flux, re-spectively. However, the contributions of mode I to thetotal vertical heat (moisture) flux drop to 84%, 86%,and 82% (80%, 88%, and 83%) on 25A, 29 May, and 7June. Interestingly on 25A May, the negative {Cq�} hasa peak at 25-km segment length with positive {Cw�} anda negative {Cwq}. In Fig. 11a, the variance of potentialtemperature at 65 m AGL shows the mesoscale vari-ance peaks at a 25-km segment length on 25A May.Similarly on 29 May the mesoscale variance peaks at a50-km segment length in Fig. 11a can be linked to thepeak of the negative {Cq�} with positive {Cw�} and nega-tive {Cwq}. Considered the combination of warm anddry surface conditions with cool and moist surface con-ditions over the track as shown in previous figures (seeFigs. 4, 5, 6, 7, and 9), this updraft of warm and dry airand downdraft of cool and moist air can be interpretedas a mesoscale circulation (mode III). The entrain-ment–drying processes (mode II) were also involvedboth on 25A and 29 May. On 25B May and 7 June,some portions of the vertical heat and moisture fluxeswere described by the combinations of the negative{Cq�} with positive {Cw�} and positive {Cwq} (mode IV).For the case on 25B May, we hypothesized that modeIV is the consequence of the mixed signatures of heat-ing and moistening from the surface and from meso-scale circulations. The {Cw�} and {Cwq} are positive dueto heating and moistening by thermals from the surfacewhereas {Cq�} is negative because it is more influencedby warm and dry, and cool and moist, air associatedwith mesoscale circulations. For the case on 7 June,mode IV seems to be the consequence of the heatingand moistening from a surface mixed with dry entrain-ment. The strong entrainment of dry air from the freeatmosphere cause {Cq�} to be highly negative even inthe low-level ABL, while {Cw�} and {Cwq} are still posi-tive due to heating and moistening from the surface.

→

FIG. 10. The ratio of spatial variance to total variance in the variance decompositions of (a) potential temperature and water vapormixing ratio, of (b) along-track and cross-track winds, and of (c) vertical heat and moisture fluxes; th, q, , u, wth, and wq representpotential temperature, mixing ratio, along-track wind, cross-track wind, vertical heat flux, and vertical moisture flux, respectively. Here,1-, 4-, 6-, 8-, and 12-km nonoverlapping segments are used for the data from the five low-level legs between 1700 and 1900 UTC 19 May(square), 25A May (diamond), and 7 June (triangle).



b. Joint probability distributions

To identify similar regimes based on the distributionsof the perturbations of potential temperature, watervapor mixing ratio, and vertical velocity, which werecollected from the repeated aircraft passes at 65 mAGL, the joint probability distribution (JPD) tech-nique (Holland 1973; Mahrt and Paumier 1984; Dear-dorff and Willis 1985) is employed. For the total per-turbations defined from each leg average, the occur-rences of q�/qML

* (qML

* is the mixed-layer humidity scale;Stull 1988) and ��/�ML

* (�ML

* is the mixed-layer tempera-ture scale; Stull 1988), within the grid of �q/qML

* and��/�ML

* , were counted. The counted occurrences at eachgrid were normalized by the total number of points. Thus,

��Pq�dq�q*MLd��*

ML � 1, �9�

FIG. 11. The composite of the normalized MR spectra of (a)potential temperature (�) and (b) water vapor mixing ratio (q)over all the low-level legs on 19, 20, 25A, 25B, 29 May, and 7 June.The error bar indicates standard error defined as the standarddeviation between the low-level legs divided by the square root ofthe number of the low-level legs.

FIG. 12. The composite of the MR spectra of horizontal (u, )and vertical velocity (w) components over all the low-level legs on19, 20, 25A, 25B, 29 May, and 7 June. The error bar indicatesstandard error.


where Pq� is the probability of the occurrence withinthe grid of �q/qML

* and ��/�ML

* . Similarly, the jointprobability distribution was obtained for both w� andq�, as well as w� and ��. For the scaling parameter, thevalue is first computed for each low-level leg. Next theaverage of the scales over all the low-level legs on agiven day is obtained (see Table 4).

The JPDs categorize the six cases into two groups: 1)19 May, 20 May, and 7 June, and 2) 25A May, 25B May,and 29 May. Figure 14 shows that on the days in thesecond group, the dry–warm and moist–cold quadrants

become more significant than those in the first group.In addition, the vertical velocity–water vapor distribu-tion looks more elliptical in the second group (Fig. 15).Finally, in Fig. 16, the JPD extends slightly more intothe warm downdraft and cold updraft quadrants in thesecond group. These warm downdraft and cold updraftquadrants are associated with entrained free atmo-sphere and penetrative convection into the entrainmentzone, respectively, while the warm updraft and colddowndraft quadrants are related to thermals (Mahrtand Paumier 1984; Deardorff and Willis 1985). These

FIG. 13. The composites of the normalized MR cospectra of w�� (solid), w�q� (dotted), andq�� (dashed) over all the low-level legs on (a) 19 May, (b) 20 May, (c) 25A May, (d) 25B May,(e) 29 May, and (f) 7 June. The error bar represents standard error. The vertical solid lineindicates boundary between mode I and other modes. Modes I, II, III, and IV are explainedin Table 3.


JPDs imply the existence of different processes associ-ated with vertical heat and moisture fluxes in the low-levels of the ABL over the track, between those twogroups.

6. Vertical extension

a. Minimum length scale

Previous studies (Pielke 2001; Avissar and Schmidt1998; Shen and Leclerc 1995; to name a few) haveconcluded that the effect of large-scale surface het-erogeneity can extend upward farther into the ABL;the height in the atmosphere at which surface hetero-geneity of a given spatial scale and amplitude can nolonger be detected is called the blending height(Mahrt 2000). Based on the approximate balance be-tween the temperature advection and vertical diver-gence of the perturbation heat flux caused by surfaceheterogeneity, which is later canceled out during linear-ization, Wood and Mason (1991) built a thermal blend-ing height formulation. This can be inversely used toestimate the horizontal length scale of surface hetero-geneity (LTH) that starts to influence the flow at a cer-tain level, z:

LTH � CTH

U��

�w��sfc

z, �10�

where CTH � 3.1 � 10�3 is the nondimensional coeffi-cient (Mahrt 2000), U is the mean horizontal wind ve-locity, � is the mean virtual potential temperature, and(w��)sfc is the surface buoyancy flux (Mahrt 2000).With respect to the distance the flow can travel duringone large-eddy turnover time scale, the convectiveboundary layer (CBL) scaling formulation suggested byRaupach and Finnigan (1995) can be used to estimatethe minimum surface heterogeneity scale that can sig-nificantly influence the whole ABL:

LCO � CCO

Uzi

w*, �11�

where CCO � 0.8 is the nondimensional coefficient(Mahrt 2000), w* is the convective velocity scale, and zi

is the ABL depth.

Using (10) the minimum surface heterogeneity scalethat significantly influences the flow at 65 m AGL iscomputed for each low-level leg. Also for each low-level leg, the minimum surface heterogeneity scale forthe flow at the ABL top is estimated by using (11). Thecomposites of the length scales over all of the low-levellegs on a given day are summarized in Table 5. On 25AMay, the estimated values are one order smaller thanthose on the other days. This implies that surface het-erogeneity on a much smaller scale can significantlyinfluence the whole ABL structure on 25A May.

b. Deviations from mixed-layer similarity

Quasi-steady and horizontally homogenous ABLconditions are assumed to obtain the vertical profiles ofdimensionless variances of velocity, potential tempera-ture, and moisture, which can be expressed by themixed-layer similarity relationships (Figs. 4 and 5 ofLenschow et al. 1980; Stull 1988, 370–371). The dimen-sionless variance profiles are estimated from the air-craft measurements over the track for the five days. Forvertical velocity, temperature, and moisture, the vari-ances based on leg averages are normalized by theirappropriate scaling parameters, � �2�/ 2

*, where * isconvective velocity scale (w*), mixed-layer tempera-ture scale (�ML

* ), or mixed-layer moisture scale (qML

* ).For horizontal and vertical wind, the convective stressvelocity scale (uML

* � u2

*/w*; Stull 1988) is used to nor-malize the variances. The scaling parameters obtainedfrom the data collected by the repeated low-level passesare summarized in Table 4. For each leg, the flight

TABLE 4. The mixed-layer scaling parameters on the five casedays. For the definitions of these scaling parameters, refer to Stull(1988).

u*

(m s�1)w

*(m s�1)

uML

*(m s�1)

�ML

*(K)

qML

*(g kg�1)

19 May 0.76 2.23 0.259 0.139 0.01320 May 0.76 2.00 0.285 0.145 0.01225 May 0.27 1.66 0.044 0.119 0.02629 May 0.39 1.61 0.093 0.088 0.081

7 June 0.54 1.65 0.176 0.104 0.068

TABLE 3. Combinations of the composited MR cospectra of w�and ��, {Cw�}, w�and q�, {Cwq}, and q� and ��, {Cq�}({} is omitted in the table).

Mode Combinations Description Physical interpretation

I Cw� � 0, Cwq � 0, Cq� � 0 Updraft: Warm, moist air; downdraft: Cool, dry air Heating and moistening from surfaceII Cw� � 0, Cwq � 0, Cq� � 0 Downdraft: Warm, dry air Entrainment from free atmosphereIII Cw� � 0, Cwq � 0, Cq� � 0 Updraft: Warm, dry air; downdraft: Cool, moist air Mesoscale circulationIV Cw� � 0, Cwq � 0, Cq� � 0 — Mixed signatures


height is normalized by the ABL depth, which is esti-mated based on the UWKA soundings and the spatialaverages of the ABL depths from the DLR Falcon overthe track.

As shown in Fig. 17, except below the height of 0.1 zi

(zi is the ABL depth), the normalized variances of hori-zontal wind velocity are well fitted by �u�2�/uML

*2 � ��2�/

uML

*2 � const. (Stull 1988). Normalized variances of

horizontal wind on 25 and 29 May are aligned with theconstant values that are 19 times and 4 times larger thanthat on the other three days, respectively. Also, thevertical velocity variances normalized by convective

stress velocity, �w�2�/uML

*2, can be fitted by constant val-

ues, which are 18 times larger on 25 May and 3 timeslarger on 29 May than that on the other days (Fig. 18a).However, the normalized variances of wind velocity atlevels higher than 0.6 zi on 7 June deviate from thesimilarity curves, which may be associated with theweak capping inversion of 2 K (200 m)�1 measured atthe Homestead profiling site at 1856 UTC (Fig. 2).

In Fig. 19a, the variances of potential temperature on19 May, 20 May, and 7 June are well fitted by the simi-larity curve of ��2�/�2

* � 1.8 (z/zi)�2/3 (Lenschow et al.

1980; Stull 1988) except for the values near the ABL

FIG. 14. Joint probability distributions of the perturbations of water vapor mixing ratio and potential temperature on (a) 19 May, (b)20 May, (c) 25A May, (d) 25B May, (e) 29 May, and (f) 7 June. The color bar indicates the range of the probability in percent withina grid of �q�/qML

*and ��/�ML

*. Here �q� is 12 g kg�1, and �� is 6 K. For each day, the values of qML

*and �ML

*are given in Table 4.


top where the effect of entrainment is significant (Wyn-gaard and LeMone 1980). The deviations of the poten-tial temperature variances from the similarity curve on25 and 29 May are likely associated with the significantmesoscale peaks in the spectra of potential temperatureat 65 m AGL (Fig. 11a). Moisture variances are gener-ally larger than the expected values from the similaritycurve of �q�2�/q2

* � 1.8 (z/zi)�2/3 (Lenschow et al. 1980;

Stull 1988). In general these large deviations can belinked to more sensitive response of moisture to en-trainment or mesoscale motions than temperature andvertical wind (Mahrt et al. 1994a,b; Mahrt 1991). Also

consistent with the large mesoscale peaks in the spectraof water vapor mixing ratio at 65 m AGL (Fig. 11b), thedeviations of the variances on 25 and 29 May are muchlarger than those on the other days.

Assuming the departures from similarity are due tomesoscale motions, we recalculate the variances rela-tive to segment averages. As shown in Fig. 20, the ver-tical profiles of the variances based on segment aver-ages are well fitted to the similarity curve whereas somedepartures are still shown due to the negligence of thestrong wind effect on 19 and 20 May and the deepentrainment on 7 June. This implies that filtering out

FIG. 15. Joint probability distributions of the perturbations of vertical velocity and water vapor mixing ratio on (a) 19 May, (b) 20May, (c) 25A May, (d) 25B May, (e) 29 May, and (f) 7 June. The color bar indicates the range of the probability in percent within agrid of �w�/wML

*and �q�/qML

*. Here �w� is 0.8 m s�1, and �q� is 12 g kg�1. For each day, the values of wML

*and qML



the mesoscale motions makes the similarity more ap-plicable.1 Here it is to be noted that—different thanvertical velocity and moisture, which use a 1-km high-pass spatial filter—potential temperature uses a 4-kmspatial filter to fit the observations onto the similaritycurves.

1 In the absence of a significant spectral gap (between f/3 to5f/3, where f is the cutoff frequency for a perfect filter separatingthe mesoscale from turbulence; LeMone 1976), one would stillexpect nonlinear transfer of variance from the mesoscale motionsto higher frequencies, so similarity theory would probably under-estimate variances slightly.

FIG. 16. Joint probability distributions of the perturbations of vertical velocity and potential temperature on (a) 19 May, (b) 20 May,(c) 25A May, (d) 25B May, (e) 29 May, and (f) 7 June. The color bar indicates the range of the probability in percent within a grid of�w�/wML

*and ��/�ML

*. Here �w� is 0.8 m s�1, and �� is 6 K. For each day, the values of wML

*and �ML


TABLE 5. The minimum horizontal length scale* of surface het-erogeneity, which can significantly influence the flow at 65 mAGL, estimated from (10), and at the ABL top, estimated from(11) by using the data from the low-level aircraft passes.

The minimum horizontal length scale (km)

19 May 20 May 25A May 25B May 29 May 7 June

LTH 5.21 4.59 0.33 1.33 2.28 4.19LCO 2.85 2.92 0.35 1.02 2.40 3.63

* These values are scaling variables, not cutoff values due to thedeficiency in the two formulas, (10) and (11), which is caused byrough estimation of the coefficients and lack of information onthe surface heterogeneity amplitude.


c. Observed spatial variations of the ABL depth

The spatial distributions of the ABL depths over thetrack for the five BLH days, which are estimated fromthe aerosol backscatter measurements of the DIAL

aboard the DLR Falcon, are shown in Fig. 21. On 19,20, and 29 May, the ABL was deeper to the north andshallower to the south, consistent with the surface skintemperature distribution (Fig. 3). On 25 May, a differ-ent ABL depth distribution, elevated around 36.60°N

FIG. 17. The vertical profiles of the normalized horizontal ve-locity variances. The dotted lines are the curves of (a) �u�2�/uML

*2

� const. and (b) ��2�/uML

*2

� const. Here the convectivestress velocity scale, uML

*, is defined as u2

*/w*

, where u*

is frictionvelocity and is convective velocity scale (Stull 1988). The con-stants are 30, 110, and 580 for the (a) along-track wind varianceand (b) the cross-track wind variances.

FIG. 18. The vertical profiles of the normalized vertical velocityvariances normalized by (a) convective stress velocity scale, uML

*,

and (b) convective velocity scale, w*

. The dotted lines are thecurve of (a) �w�2�/uML

*2

� const. and (b) �w�2�/w2* � 1.8 (z/zi)

�2/3

(1 � 0.8z/zi)2 (from Lenschow et al. 1980). The constants are 30,

90, and 550 for the vertical velocity variances normalized by con-vective stress velocity.


and depressed around 36.80°N, was also consistent withthe surface temperature distribution on that day (Fig.3). Smaller-scale surface heterogeneity was expected tosignificantly influence the whole ABL structure on 25May based on the minimum length scale argument insection 6a. For the four days, these ABL depth distri-butions are consistent also with the potential tempera-ture distributions measured at 65 m AGL. On 7 June,only the first leg, flown 1651 and 1704 UTC, showed the

deeper-north pattern. On subsequent legs, the ABLgrew rapidly across the entire track, associated with aweak capping inversion of 2 K (200 m)�1 measured atHomestead profiling site at 1856 UTC (Fig. 2); ABLheights were randomly distributed along the track.

FIG. 19. The vertical profiles of the normalized potential tem-perature and water vapor mixing ratio variances. The dotted linesare the curve of (a) ��2/�2

* � 1.8 (z/zi)�2/3 and (b) q�2/q2

* � 1.8(z/zi)

�2/3 (from Lenschow et al. 1980).

FIG. 20. The vertical variance profiles of (a) vertical velocity, (b)potential temperature, and (c) water vapor mixing ratio normal-ized by convective velocity scale, mixed-layer temperature scale,and mixed-layer moisture scale, respectively. For vertical velocityand water vapor mixing ratio, 1-km spatial filters are used. Forpotential temperature, 4-km spatial filters are used. The dottedlines are the curve of (a) �w�2�/w2

* � 1.8 (z/zi)�2/3 (1 � 0.8z/zi)

2,(b) ��2/�2

* � 1.8 (z/zi)�2/3, and (c) q�2/q2

* � 1.8 (z/zi)�2/3 (from

Lenschow et al. 1980).


Thus one can conclude that the features of the surfaceheterogeneity over the track influenced the entire ABLstructures except on 7 June. In other words, over thetrack for the four days among the five days, the blend-ing height determined from the surface heterogeneityexceeded the ABL depths and the ABL establishedequilibrium with the local surface condition, a regimecalled macroscale heterogeneity (Mahrt 2000).

7. Summary and conclusions

This study analyzes data collected by aircraft on fivefair-weather days during IHOP_2002 to investigate theABL structures over a heterogeneous land surfacealong the track (36.4°–37.0°N, 100.6°W; Fig. 1) underdifferent background weather conditions. The tower-and aircraft- measured surface skin temperature distri-

butions both show the warm–north and cool–south pat-tern over the track. The surface skin temperature wasincreased by about 10°C or more from southern end tothe northern end of the aircraft track for four of the fivedays, whereas on 7 June the surface temperature at thenorthern end of the track was slightly less than 5°Chigher than at the southern end. On 25 May, the warmregion around 36.64°N was superimposed on the warm–north and cool–south pattern.

The measurements at the surface flux sites suggestthat the surface temperature distribution is mainlycaused by the combination of soil thermal propertiesand soil moisture. With a similar amount of availablesurface energy (Rn-G) along the track, site 3, locatedaround the northern end of the track (36.86°N), usedmost of the available energy to produce sensible heat

FIG. 21. The estimated ABL depths based on the aerosol backscatter measurements of theDIAL aboard the DLR Falcon flown over the western track on (a) 19 May, (b) 20 May, (c)25 May, (d) 29 May, and (e) 7 Jun 2002. The circles, filled circles, squares, and filled squaresindicate first, second, third, and fourth legs. The first leg on 19 May is flown at 1650–1704UTC. The first and second legs on 20 May are flown at 1920–1932 and 1941–1953 UTC,respectively. The first, second, and third legs on 25 May are flown at 1745–1758, 1807–1818,and 1844–1857 UTC. The first, second, third, and fourth legs on 29 May are flown at 1806–1820, 1829–1839, 1904–1918, and 1927–1938 UTC. The first, second, third, and fourth legs on7 Jun are flown at 1651–1704, 1754–1803, 1829–1835, and 1932–1944 UTC.


flux due to dry soil, as shown in Fig. 6. At site 2(36.62°N), the heat flux into ground (G) is less due torelatively small soil thermal conductivity (Table 2).Thus, even with similar amount of latent heat flux, site2 could produce more sensible heat flux than site 1 dueto more available surface energy. On 25 May, surfacetemperature around site 2 can more rapidly rise due tothe shallower active thermal-exchange layer that is as-sociated with the smaller soil thermal conductivity, es-pecially under relatively calm winds. This additionalheating around site 2 causes the small-scale heteroge-neity superimposed on the large-scale pattern.

The stationary spatial variability described by thedata collected from the repeated aircraft passes at 65 mAGL shows a slowdown of the along-wind componentand warmer air temperatures over warmer surfaces.High vertical flux is also associated with warmer sur-faces, but the association is less robust due to consid-erable temporal and transient variability.

Mesoscale circulations (warm, dry updrafts and cool,moist downdrafts) generated by the land surface het-erogeneity can be seen through MR cospectra andJPDs on 25 and 29 May. On these two days, the ABL islikely dominated by buoyancy-generated turbulence,given the smallness of the Obukhov lengths, and therelatively small-scale surface heterogeneity needed toinfluence the whole ABL. On 25A May, MR spectrashow a significant peak of mesoscale along-track windvariance (Fig. 12b), which is linked to the significantpeaks of mesoscale variances of potential temperatureand water vapor mixing ratio. Also on these two days,the vertical profiles of dimensionless variances of windcomponents, potential temperature, and moisture(Figs. 17, 18, and 19) exhibit significant deviations fromthe similarity curves, which are based on the assump-tion of homogeneous surface conditions. However,when we used high-pass spatial filters, these dimension-less variances fit the similarity curves better (Fig. 20),which also implies the existence of mesoscale circula-tion on 25 and 29 May.

On 25 and 29 May, however, the lack of a high-amplitude mesoscale peak in the wind spectra (Fig. 12),except for the along-track wind component on 25AMay, makes the existence of well-organized mesoscalecirculations doubtful. Considering the scales of the sur-face heterogeneity feature, which can be seen in theABL depth distribution on these two days, the ob-served ABL seems to be in the mixed-scale ABL re-gime of turbulence and less-organized mesoscale circu-lations. With the similar horizontal scale of surface het-erogeneity (20–40 km), LES studies (Patton et al. 2005;Avissar and Schmidt 1998) have shown that persistentturbulent thermals coexisting with less-organized me-

soscale circulations over a sinusoidal-shaped heteroge-neous land surface. However, they have mostly focusedon the well-organized mesoscale circulations. In futureresearch, we plan to simulate the ABL under themixed-scale regime, which was likely observed on 25and 29 May by using LES with the background weatherconditions measured at the Homestead profiling site(Fig. 1) and the observed surface heat fluxes at thethree ISFF sites. These simulations will allow us to in-vestigate the ABL structure under the mixed-scale re-gime in detail and its potential link to convective clouddevelopment.

Acknowledgments. This research was supported bythe National Science Foundation through Grant ATM-0130349. The third author’s participation is sponsoredthrough NCAR. This work could not have been pos-sible without the observational dataset collected by thegreat efforts of the staffs of UWKA, NCAR ISFF, andDLR Falcon. The authors thank Ken Craig for provid-ing the ABL depth data estimated from the aerosolbackscatter data of the DIAL aboard the DLR Falcon,Dean Vickers and L. Mahrt for providing MR spectrasoftware and commenting on the spectra results, andBrian Reen and Robert B. Seigel for helping to preparefor this manuscript.

REFERENCES

Arya, S. P., 2001: Introduction to Micrometeorology. AcademicPress, 420 pp.

Avissar, R., and T. Schmidt, 1998: An evaluation of the scale atwhich ground-surface heat flux patchiness affects the convec-tive boundary layer using large-eddy simulations. J. Atmos.Sci., 55, 2666–2689.

Chen, F., and R. Avissar, 1994a: The impact of land-surface wet-ness heterogeneity on mesoscale heat fluxes. J. Appl. Meteor.,33, 1323–1340.

——, and ——, 1994b: Impact of land-surface moisture variabilityon local shallow convective cumulus and precipitation inlarge-scale models. J. Appl. Meteor., 33, 1382–1401.

——, and Coauthors, 2003: Utilizing the IHOP 2002 data to studythe variability in surface evaporation and water vapor in theboundary layer. Preprints, Symp. on Observing and Under-standing the Variability of Water in Weather and Climate,Long Beach, CA, Amer. Meteor. Soc., CD-ROM, P1.2.

Crook, N. A., 1996: Sensitivity of moist convection forced byboundary layer processes to low-level thermodynamic fields.Mon. Wea. Rev., 124, 1767–1785.

Davis, K. J., N. Gamage, C. R. Hagelberg, C. Kiemle, D. H. Len-schow, and P. P. Sullivan, 2000: An objective method for de-riving atmospheric structure from airborne lidar observa-tions. J. Atmos. Oceanic Technol., 17, 1455–1468.

Deardorff, J. W., and G. E. Willis, 1985: Further results from alaboratory model of the convective planetary boundary layer.Bound.-Layer Meteor., 32, 205–236.

Holland, J. Z., 1973: A statistical method for analyzing wave


shapes and phase relationships of fluctuating geophysicalvariables. J. Phys. Oceanogr., 3, 139–155.

LeMone, M. A., 1976: Modulation of turbulent energy by longi-tudinal rolls in an unstable planetary boundary layer. J. At-mos. Sci., 33, 1308–1320.

——, and Coauthors, 2002: CASES-97: Late-morning warmingand moistening of the convective boundary layer over theWalnut River watershed. Bound.-Layer Meteor., 104, 1–52.

——, R. L. Grossman, F. Chen, K. Ikeda, and D. Yates, 2003:Choosing the averaging interval for comparison of observedand modeled fluxes along aircraft transects over a heteroge-neous surface. J. Hydrometeor., 4, 179–195.

——, and Coauthors, 2007: Influence of land cover and soil mois-ture on the horizontal distribution of sensible and latent heatfluxes in southeast Kansas during IHOP_2002 and CASES-97. J. Hydrometeor., 8, 68–87.

Lenschow, D. H., J. C. Wyngaard, and W. T. Pennell, 1980: Mean-field and second- moment budgets in a baroclinic, convectiveboundary layer. J. Atmos. Sci., 37, 1313–1326.

Mahrt, L., 1991: Boundary-layer moisture regimes. Quart. J. Roy.Meteor. Soc., 117, 151–176.

——, 1998: Flux sampling errors for aircraft and towers. J. Atmos.Oceanic Technol., 15, 416–429.

——, 2000: Surface heterogeneity and vertical structure of theboundary layer. Bound.-Layer Meteor., 96, 33–62.

——, and J. Paumier, 1984: Heat transport in the atmosphericboundary layer. J. Atmos. Sci., 41, 3061–3075.

——, J. Sun, D. Vickers, J. I. Macpherson, J. R. Pederson, andR. L. Desjardins, 1994a: Observations of fluxes and inlandbreezes over a heterogeneous surface. J. Atmos. Sci., 51,2484–2499.

——, J. I. Macpherson, and R. L. Desjardins, 1994b: Observationsof fluxes over heterogeneous surfaces. Bound.-Layer Meteor.,67, 345–367.

Mann, J., and D. H. Lenschow, 1994: Errors in airborne flux mea-surements. J. Geophys. Res., 99, 14 519–14 526.

Patton, E. G., P. P. Sullivan, and C.-H. Moeng, 2005: The influ-ence of idealized heterogeneity on wet and dry planetaryboundary layers coupled to the land surface. J. Atmos. Sci.,62, 2078–2097.

Pielke, R. A., 2001: Influence of the spatial distribution of veg-etation and soils on the prediction and soils on the cumulusconvective rainfall. Rev. Geophys., 39, 151–177.

Poberaj, G., A. Fix, A. Assion, M. Wirth, C. Kiemle, and G.Ehret, 2002: Airborne all-solid-state DIAL for water vapormeasurements in the tropopause region: System descriptionand assessment of accuracy. Appl. Phys., B75, 165–172.

Raupach, R. R., and J. J. Finnigan, 1995: Scale issues in boundary-layer meteorology: Surface energy balances in heterogeneousterrain. Hydrol. Processes, 9, 589–612.

Rozoff, C. M., W. R. Cotton, and J. O. Adegoke, 2003: Simulationof St. Louis, Missouri, land use impacts on thunderstorms. J.Appl. Meteor., 42, 716–738.

Segal, M., J. R. Garratt, R. A. Pielke, W. E. Schreiber, A. Rodi,G. Kallos, and J. Weaver 1989: The impact of crop areas innortheast Colorado on midsummer mesoscale thermal circu-lations. Mon. Wea. Rev., 117, 809–825.

Shaw, W. J., and J. C. Doran, 2001: Observations of systematicboundary layer divergence patterns and their relationship toland use and topography. J. Climate, 14, 1753–1764.

Shen, S., and M. Y. Leclerc, 1995: How large must surface inho-mogeneities be before they influence the convective bound-ary layer? A case study. Quart. J. Roy. Meteor. Soc., 121,1209–1228.

Stull, R. B., 1988: An Introduction to Boundary Layer Meteorol-ogy. Kluwer Academic, 666 pp.

Vickers, D., and L. Mahrt, 2003: The cospectral gap and turbulentflux calculations. J. Atmos. Oceanic Technol., 20, 660–672.

Weckwerth, T. M., and Coauthors, 2004: An overview of the In-ternational H2O Project (IHOP_2002) and some preliminaryhighlights. Bull. Amer. Meteor. Soc., 85, 253–277.

Wood, N., and P. J. Mason, 1991: The influence of stability on theeffective roughness lengths for momentum and heat flux.Quart. J. Roy. Meteor. Soc., 117, 1025–1056.

Wyngaard, J. C., and M. A. LeMone, 1980: Behavior of the re-fractive index structure parameter in the entraining convec-tive boundary layer. J. Atmos. Sci., 37, 1573–1585.

Zhang, Q., and R. Huang, 2004: Water vapor exchange betweensoil and atmosphere over a Gobi surface near an oasis in thesummer. J. Appl. Meteor., 43, 1917–1928.

Zhong, S., and J. C. Doran, 1998: An evaluation of the importanceof surface flux variability on GCM-scale boundary-layercharacteristics using realistic meteorological and surface forc-ing. J. Climate, 11, 2774–2788.


Documents

Observations of the ABL Structures over a Heterogeneous ...opensky.ucar.edu/islandora/object/articles:6962/datastream/PDF/download... · from the Honeywell Laseref SM Inertial Reference