sensitivity of wrf-simulated planetary boundary layer...

BMeteorologische Zeitschrift, Vol. 23, No. 3, 279–293 (published online July 11, 2014) Open Access Article© 2014 The authors

Sensitivity of WRF-simulated planetary boundary layerheight to land cover and soil changes

Ferenc Ács1∗, András Zénó Gyöngyösi1, Hajnalka Breuer1, Ákos Horváth2, Tamás Mona1 andKálmán Rajkai3

1Eötvös Loránd University, Department of Meteorology, Budapest, Hungary2Hungarian Meteorological Service, Siófok, Hungary3Centre for Agricultural Research, Institute for Soil Science and Agricultural Chemistry, Budapest, Hungary

(Manuscript received October 30, 2013; in revised form April 23, 2014; accepted May 15, 2014)

AbstractPlanetary boundary layer (PBL) height sensitivity to both so-called single and accumulated land cover andsoil changes is investigated in shallow convection under cloud-free conditions to compare the effects. Singleland cover type and soil changes are carried out to be able to unequivocally separate the cause and effectrelationships. The Yonsei University scheme in the framework of the Weather Research Forecasting (WRF)mesoscale modeling system is used as a research tool. The area investigated lies in the Carpathian Basin,where anticyclonic weather type influence dominated on the five summer days chosen for simulations.Observation-based methods applied for validating diurnal PBL height courses manifest great deviationsreaching 500–1300 m. The obtained deviations are somewhat smaller around midday and greater at night.They can originate either from the differences in the measuring principles or from the differences in theatmospheric profiles used. Concerning sensitivity analyses, we showed that PBL height differences causedby soil change are comparable with the PBL height differences caused by land cover change. The differencesare much greater in the single than in the accumulated tests. Space averaged diurnal course differencearound midday reaching a few tens of meters can be presumably treated as strongly significant. PBL heightdifferences obtained in the sensitivity analyses are, at least in our case, smaller than those obtained by applyingdifferent observation based methods. The results may be utilized in PBL height diurnal course analyses.

Keywords: planetary boundary layer height, land cover, soil, shallow convection, Carpathian Basin, WRFmodeling system

1 Introduction

Hungary has a continental climate, which may be pre-sented as being cool and dry with extreme seasonalchanges of temperature (Feddema, 2005). This climateis also characterized by hot summers with long-lastingdry periods. During these periods, the prevailing pro-cess is shallow convection often generating shallow con-vective clouds, known as “fair weather cumuli”, whichare typical in the afternoon. Since Hungary’s relief isquite uniform, the planetary boundary layer (PBL) statein such cases is mostly determined by the state of theland-surface, that is, by soil and land cover characteris-tics.

Land use induced land cover change (LULCC) canhave a profound effect on shallow convection. Vege-tation differences can cause “vegetation breeze”, thatis, locally induced thermally driven winds (e.g. Segalet al., 1988, 1989; Mahfouf et al., 1987; Hong et al.,1995), cloud formation (e.g. Garrett, 1982; Rabinet al., 1990; Chen and Avissar, 1994; Adegoke et al.,2007), changes in the thermodynamic state of the PBL

∗Corresponding author: Ferenc Ács, Eötvös Loránd University, Departmentof Meteorology, Pázmány Péter sétány 1/a., 1117 Budapest, Hungary, e-mail:[email protected]

(e.g. McPherson et al., 2004; McPherson and Sten-srud, 2005; Garcia-Carreras et al., 2010). The impactof soil characteristics on PBL is also well documented.Among soil properties, soil texture is the most impor-tant and this is analyzed thoroughly e.g. by Alapatyet al. (1997) and Niyogi et al. (2002). The role of soilhydraulic functions (e.g. Cuenca et al., 1996; Shao andIrannejad, 1999) and the variation of soil hydraulic pa-rameters (e.g. Ek and Cuenca, 1994; Mölders, 2005)are also well documented. There are also investigationson the effect of the use of diffent soil databases (e.g. Ácset al., 2010; Breuer et al., 2012).

In these studies, the effect of soil or land coverchanges on PBL is analyzed separately without anycomparison of the two effects. To bridge this gap, we an-alyzed comparatively the soil and land cover change ef-fects on PBL height investigating its diurnal course dur-ing five hot summer days in the Carpathian Basin. In thecomparisons, we used both so-called single and accumu-lated (Mölders, 1999) land cover and soil change tests.The Weather Research Forecasting model (WRF) wasused as a research tool. The model setup together withPBL and land-surface parameterizations is described insection 2.1. Measurement techniques, PBL height deter-mination methods used, and the days chosen for simula-

© 2014 The authorsDOI 10.1127/0941-2948/2014/0544 Gebrüder Borntraeger Science Publishers, Stuttgart, www.borntraeger-cramer.com

Downloaded from www.schweizerbart.deUnauthorized distribution of this copyrighted material is strictly forbidden!

http://www.borntraeger-cramer.de/journals/metz

http://www.borntraeger-cramer.de/journals/metz

http://www.borntraeger-cramer.com

280 F. Ács et al.: Sensitivity of WRF-simulated planetary boundary layer height Meteorol. Z., 23, 2014

tion are presented in section 2.2, 2.3 and 2.4. Data usedin the simulations, experimental design and the signifi-cance test method are briefly discussed in sections 2.5,2.6, 2.7 and 2.8. Validation and sensitivity test resultsare presented in section 3, the conclusions drawn are de-scribed in section 4.

2 Methods and data

2.1 WRF model

The Weather Research and Forecasting model has beendeveloped and used by a wide community of researchersand forecasters from its early release in 2002. This soft-ware is applicable for both atmospheric research andweather forecasting purposes ranging from micro toglobal scales.

2.1.1 Model setup

Simulations were performed using the WRF-ARW(Weather Research Forecasting-Advanced ResearchWeather) v3.3.1 model (Skamarock et al., 2008). Thenumber of vertical levels in the atmosphere was 44, fromthese 24 levels were under a height of 2 km. The do-main centered in the Carpathian Basin (N47.10; E19.30)included 97 by 97 grid points covering the CarpathianBasin and some parts of surrounding mountains andother neighboring area. The horizontal resolution wasset to 10 by 10 km.

The physical parameterizations were as follows: theRRTM (Rapid Radiative Transfer Model) for longwave radiation (Mlawer et al., 1997), Dudhia’s (1989)scheme for shortwave radiation, the YSU (YonseiUniversity) scheme (Hong et al., 2006) for PBL, theWSM 3-class (WRF-Single Moment 3-class) simple icescheme (Hong et al., 2004) for microphysics, the modi-fied version of the Kain-Fritsch scheme (Kain, 2004) forcumulus convection and the Noah (National Centers forEnvironmental Prediction-Oregon State University-AirForce-Hydrologic Research Lab) scheme (Chen andDudhia, 2001) for land-surface processes. The packagewas extensively tested by Gyöngyösi et al. (2013) forsimulating shallow convection in typical summer condi-tions. According to the results obtained, the parameteri-zation package seemed to be cost-efficient and suitable.

2.1.2 Planetary boundary layer parameterization

There are several options for representing planetaryboundary layer treatment in the new WRF release. Weused the Yonsei University (YSU) scheme (Hong et al.,2006), which is an advanced PBL scheme based on thenonlocal mixing theory taking into account the contri-bution of large-scale eddies to the total flux. This isachieved by an explicite treatment of the entrainmentprocess at the inversion layer. Boundary layer height is

estimated by calculating the bulk Richardson numberRib

h = Ribcr ·θva|U(hu,d)|2

g · [θv(hu,d)−θs], (2.1)

where Ribcr is the critical bulk Richardson number, θvais the virtual potential temperature at the lowest modellevel, U(hu,d) is the horizontal wind speed at hu,d modellevels around h, g is the acceleration of gravity, θv(hu,d)is the virtual potential temperature at hu,d around h andθs is the near surface virtual potential temperature. θs isexpressed in terms of the so-called virtual temperatureexcess (θT ), which depends on the virtual heat flux fromthe surface and on the scaling velocity term, ws, that isthe function of so-called friction velocity and dimen-sionless wind shear function evaluated at the inversionlayer height (see eq. 2 in Hong et al., 2006). The effectof θT on h is achieved by an iteration procedure calcu-lating Rib(z) values starting with θT = 0,

Rib(z) =g · [θv(z)−θs] · z

θva ·U(z)2 . (2.2)

h is estimated comparing the Rib(z) at height h and theRibcr. h is that height where Rib(z) = Ribcr. In mostcases, this can be achieved only by interpolation ofRib(z) using its values from the adjacent model levels.Ribcr for unstable stratification (Ribun

cr ) is equal to 0,while for stable stratification (Ribst

cr) it is equal to 0.25.

2.1.3 Land-surface parameterization

The Noah LSM (Chen and Dudhia, 2001) is appliedfor representing soil and vegetation processes. Noah isa well known land-surface parameterization scheme, thedevelopment of which began in the 1980s (Mahrt andPan, 1984; Pan and Mahrt, 1987). It possesses a mul-tilayer soil and a single layer snow and canopy model.More about the soil and the vegetation modules can befound e.g. in Horváth et al. (2009). These modules arepresented from the point of view of water transport,which is actually the core of the scheme. Other aspects,e.g. turbulent exchange processes, are not consideredsince they are relatively less important.

2.2 Measurements

The measurements of the PBL profile were performedat the Observatory of Szeged (46.25 °N, 20.10 °E) ofthe Hungarian Meteorological Service. The site is out-side the city in an agricultural environment. A radiome-ter and a wind profiler are in operation at the sta-tion. The MP-3000A ground-base microwave radiome-ter measures temperature, humidity and liquid watercontent profiles at every 50 meters in the 0–500 m layer,at every 100 meters in the 500–2000 m layer, and at ev-ery 250 meters in the 2000–10000 m layer. The instru-ment has 21 calibrated channels in 22–30 GHz (K-band)and 14 channels in 51–59 GHz (V-band). The instrument


Meteorol. Z., 23, 2014 F. Ács et al.: Sensitivity of WRF-simulated planetary boundary layer height 281

Table 1: Main weather characteristics of the simulated days

Number Days Sunshineduration

Tmin[°C]

Tmax[°C]

p [hPa]

1 5 July 2012 13.25 18 38 10112 7 August 2012 13.00 21 32 10203 20 August 2012 13.25 13 35 10214 28 August 2012 12.75 9 27 10225 2 August 2013 13.50 15 34 1019

is also equipped with sensors for registering surface tem-perature, relative humidity and pressure. The VaisalaLAP-3000 lower atmosphere wind profiler with a radioacoustic sounding system (RASS) serves for collectingwind speed and direction as well as vertical wind ve-locity data at every 220 meters in the 150–4000 m layer.The operating frequency of the instrument is 1290 MHz.The pulses are emitted at least in three directions (onetowards the zenith, one slightly tilted towards the northand one to the south), in our case four directional (north,west, south, east) beams with a tilt of 15.1 ° were used.Data are collected every 15 minutes.

2.3 Planetary boundary layer heightdetermination methods

The maximum Signal-to-Noise-Ratio (SNR) method(Angevine et al., 1994) is applied to wind profiler data.White (1993) argued that the maximum value of theSNR profile together with the maximum value of re-fractive index structure-function parameter profile, C2

n ,are located around the level of PBL height. These factsare also documented by numerous observational and/ormodeling studies (e.g. Wyngaard and LeMone, 1980;Cohn and Angevine, 2000). In this study, first the me-dian of the five SNR values is determined, afterwardsthe level with the highest SNR is assigned as the PBLheight.

PBL height is also estimated by applying the YSUscheme using wind profiler and radiometer data. Theapplied procedure slightly differs from the one presentedin section 2.1.2, namely, when calculating Rib(z) theeffect of θT is not taken into account at all because ofthe missing near surface data. In most cases, the firstlevel of available measurement results was at the heightof 405 m (second measurement level).

2.4 Simulated days

The days chosen for simulations were 5 July 2012;7 August 2012; 20 August 2012; 28 August 2012 and2 August 2013. Some basic weather elements for thesedays from the point of view of the subject consideredare presented in Table 1. Sunshine duration was about13 hours on all days. Daily temperature amplitude fluc-tuated between 18–22 °C except on 7 August 2012 andthe surface atmospheric pressure was close to 1020 hPa

or higher except on 5 July 2012. On all days the anti-cyclonic weather type influence was dominant, this in-fluence was somewhat disturbed by an oncoming coldfront from the south-west on 5 July 2012 and by a ratherdry cold front passing over in the morning hours on7 August 2012. The impact of the cold fronts was com-pletely diverse. The cooling effect was negligible on5 July, but it was much stronger on 7 August. Hence, thethermal and dynamic characteristics of local air masseswere rather diverse on these two days. The anticyclonicweather influence ensured mostly cloud-free conditions(somewhat more clouds beginning from midday wereformed only on 5 July 2012) enabling shallow convec-tion induced by strong incoming solar radiation.

2.5 Initializations

Initial and boundary conditions for the calculation do-main were generated from GFS (Global ForecastingSystem) model data by the WPS (WRF PreprocessingSystem) preprocessor of the WRF system. Globalweather data were available every 3 hours in a horizon-tal resolution of half a degree. The vertical grid spac-ing of the global model was identical to that of the lo-cal model integration, the top pressure level was set to50 hPa. Initialization of soil temperature and moisturecontent is performed using interpolated final analysisfields of GFS. These state variables represent layer av-eraged values in sub-layers 0–10, 10–40, 40–100 and100–200 cm below the ground. In the generation of ini-tial fields data assimilation was not applied, instead ofthis, each run is performed using a 12-hour spin-up timein order to obtain variables that are as steady-state aspossible.

2.6 Land-surface data

Noah uses a dozen (for each land cover and soil tex-ture combination at least 10–15 parameters; the possi-ble number of the combinations is a few hundred) land-surface parameters. Many of them are highly dependenton the land cover and soil texture type.

Two land cover type categorizations are used, theUSGS (United State Geological Survey) and the Corine2000 (Coordinate Information on the Environment). InUSGS, 21 land cover types are distinguished, amongthem only a few (e.g. crop, crop/grass, forest, shrub-land, bare ground, water, urban) are used in Hungary(Fig. 1 left). Corine 2000 land cover type classifica-tion (European Environmental Agency, 2002) isadapted for WRF applications in Hungary by Drüsz-ler (2011) (Fig. 1 right). Note that the spatial dis-tributions of the land cover types obtained by thetwo methods are quite similar. The main differenceis in the specification of land cover type “shrubland”(Göndöcs, 2013). Soil texture categorization is givenafter USDA (United State Department of Agriculture),this means 11 soil texture classes in total. The areadistribution of classes is specified according to both



Figure 1: Land cover types in Hungary on the simulated days according to USGS (left) and Corine 2000 (right). The legends (color withcorresponding name) are the same for both maps.

Figure 2: Spatial distribution of soil textural classes in Hungary according to FAO (left) and DKSIS (right). The legends (color withcorresponding name) are the same for both maps.

FAO (Food and Agriculture Organization) (Fig. 2 left)and DKSIS (Digital Kreybig Soil Information System)(Fig. 2 right). The latter database is the product of theHAS CAR (Hungarian Academy of Sciences Centrefor Agricultural Research), its description can be foundin Pásztor et al. (2010). Among the 11 soil texturalclasses there are only 7 classes in Hungary. Soil hy-draulic parameters for Hungarian and USA soils aregiven in Table 2 and Table 3, respectively.

2.7 Experimental designIn total eleven run types are performed for each day (Ta-ble 4). In each run a specific land cover type/soil tex-ture/soil parameter value combination is used. In run 0,the actual land cover type specified by USGS (Fig. 1,left), the actual spatial distribution of soil texture repre-sented by FAO (Fig. 2, left) and Hungarian soil param-eter values (Table 2) are used. This is denoted as run00-00-HU (column 1 is for land cover type, column 2



Table 2: Hungarian soil hydraulic parameters as used in the Noah LSM. Θs = soil moisture content at saturation, Θ f = field capacity, Θw =wilting point, Ψs = soil moisture potential at saturation, b = pore size distribution index and Ks = hydraulic conductivity at saturation.

Soil texture Θs (m3 m−3) Θ f (m3 m−3) Θw (m3 m−3) Ψs (m) b Ks (m s−1)

Sand 0.540 0.319 0.033 0.075 2.59 3.26E−05Loamy sand 0.625 0.481 0.080 0.165 3.37 2.52E−05Sandy loam 0.486 0.361 0.062 0.156 3.74 1.14E−05Loam 0.469 0.376 0.077 0.225 3.78 4.58E−06Sandy clay loam 0.484 0.366 0.083 0.160 4.07 7.98E−06Clay loam 0.486 0.411 0.104 0.281 4.05 3.05E−06Clay 0.547 0.490 0.152 0.271 6.00 8.00E−07Silty loam 0.488 0.377 0.076 0.188 3.93 2.73E−06Silty clay loam 0.492 0.418 0.114 0.254 4.60 6.20E−07Silt 0.511 0.426 0.070 0.231 3.51 2.00E−06

Table 3: USA soil hydraulic parameters as used in the Noah LSM. Symbols are as in Table 2.

Soil texture Θs (m3 m−3) Θ f (m3 m−3) Θw (m3 m−3) Ψs (m) b Ks (m s−-1)

Sand 0.339 0.236 0.01 0.069 2.79 4.60E−05Loamy sand 0.421 0.283 0.028 0.036 4.26 1.41E−05Sandy loam 0.434 0.312 0.047 0.141 4.74 5.23E−06Loam 0.476 0.36 0.084 0.759 5.33 2.81E−06Sandy clay loam 0.439 0.329 0.066 0.355 5.25 3.38E−06Clay loam 0.404 0.314 0.067 0.135 6.66 4.45E−06Clay 0.464 0.387 0.12 0.617 8.72 2.04E−06Silty loam 0.406 0.338 0.1 0.098 10.73 7.22E−06Silty clay loam 0.468 0.404 0.126 0.324 10.39 1.34E−06Silt 0.468 0.412 0.138 0.468 11.55 9.74E−07

Table 4: A brief specification of the main land-surface conditions used in the numerical experiments.

Run number Run type Main land-surface conditions as used in HungaryLand cover type Soil texture Soil parameter values

0 00-00-HU actual from USGS actual from FAO HU1 00-00-US actual from USGS actual from FAO US2 00-01-HU actual from USGS sand (01) HU3 00-12-HU actual from USGS clay (12) HU4 05-00-HU crop and grassland mosaic (05) actual from FAO HU5 06-00-HU crop and woodland mosaic (06) actual from FAO HU6 11-00-HU deciduous broadleaf forest (11) actual from FAO HU7 12-00-HU deciduous needleleaf forest (12) actual from FAO HU8 14-00-HU evergreen needleleaf forest (14) actual from FAO HU9 00-00-HU-dksis actual from USGS actual from DKSIS HU

10 00-00-HU-corine actual from Corine actual from FAO HU

for soil texture and column 3 for soil parameter values)and will be referred to as the “reference” run. Run 1, run00-00-US, differs from run 0 only in terms of soil pa-rameter values. Instead of Hungarian the USA soil pa-rameter values are chosen. In run 2, run 00-01-HU, theonly difference with respect to run 0 is in the soil texture.Instead of actual spatial distribution of soil texture givenby FAO, the soil texture “sand” covered the whole coun-try. Run 3, run 00-12-HU, is similar to the former case.The actual spatial distribution of soil texture is replacedby the soil texture “clay”. In run 4, run 05-00-HU, theonly difference with respect to run 0 is in the land covertype. Instead of actual land cover distribution given byUSGS the land cover type “cropland/grassland mosaic”

covered the whole country. In runs 5, 6, 7 and 8, sim-ilarly to run 4, the only difference with respect to thereference run is in the land cover type. Run 9, run 00-00-HU-dksis, differs from the “reference” run only inthe spatial distribution of soil texture; instead of FAO,DKSIS specification is used. Lastly, run 10 differs fromthe “reference” run only in the representation of actualland cover type distribution. The USGS is replaced byCorine representation.

Sensitivity tests are performed for so-called singleand accumulated land use (e.g. Mölders, 1999) andsoil texture changes. In the so-called single test, theactual spatial distribution is simply replaced by a ho-mogeneous distribution. In the so-called accumulated



Figure 3: Diurnal courses of the PBL height estimated by using the SNR method and the YSU scheme applied to measurement (YSU-MEA)and WRF simulated (YSU-WRF) results on four summer days of 2012 in Szeged.

Figure 4: Profile of the square of horizontal wind speed difference ([ΔU(z)]2) (left) and of the bulk Richardson number (right) estimated byusing wind profiler measurements and WRF simulation at 0900 UTC on 20 August 2012 in Szeged.

test, one actual spatial distribution is replaced by an an-other actual spatial distribution. In total seven single andthree accumulated tests are performed for each day. Sin-gle tests referring to land-surface effects are performedby comparing the results of run 0 and runs 4, 5, 6, 7and 8. So, we can get an insight into how important the

land cover changes are. An accumulated test referring toland-surface effects is performed by comparing the re-sults of run 0 and 10. Single tests referring to soil effectsare performed by comparing the results of run 0 andruns 2 and 3. With these comparisons, we can analyzethe effects caused by changing soil hydraulic properties.



Figure 5: Spatial distribution of planetary boundary layer height in the Carpathian Basin at 1200 UTC on 5 July 2012. The simulationsare performed with the WRF using the Yonsei University PBL parameterization scheme for reference run conditions. The investigatedsub-region is denoted by a square.

The same, but for accumulated tests, can be achieved bycomparing the results of run 0 and runs 1 and 9. In allruns, the model was started at 12 UTC the previous day.The model was running for 36 hours to ensure a 12-hourspin-up time.

2.8 Significance test

The diurnal change of PBL height is an autoregressivestochastic periodic process in a statistical sense and de-pends mainly on incoming radiation. To separate landcover and soil change effects in the diurnal course, thenatural diurnal course of PBL height was overcome withFourier-series analysis. It has to be noted that Fourier-series application puts limitations to the applicability ofthe method, for instance, in such cases, when the vari-ance of PBL height is less than zero. The method is alsonot applicable when the available number of timesteps isless than 6, or the autocorrelation coefficient is greaterthan 1. The method’s applicability could also be lim-ited by some adjustments (e.g. the standard deviation ofnormalized PBL heights has to be greater than 0.05 orthe average difference of PBL height has to be greaterthan 1), which have to be introduced in order to avoidgreat ratios when there is no actual difference betweenthe datasets. If the method is applicable, the obtainedstatistical indicator (Breuer et al., 2012; Eq. 5) is testedby the Student t-test to P < 0.10, P < 0.05, P < 0.01and P < 0.001 probabilities. The null hypothesis is thatthere is no sensitivity (no systematic difference) of PBLheight to land cover and soil changes.

3 Results and discussion

3.1 Validation resultsValidation tests are performed on 5 July and on 7, 20 and28 August 2012 using the measurement results obtainedat the Observatory of Szeged of the Hungarian Meteoro-logical Service.

3.1.1 Comparison of the PBL heightsDiurnal courses of the PBL height estimated by SNRas well as by YSU using measurements and WRF sim-ulated results are presented in Fig. 3. The YSU-MEA–YSU-WRF differences change between about −1000 mand 1000 m. They are variable in the midday and ratherconstant at night. In most cases, they are caused bydifferences existing between the observed and simu-lated wind profiles. This is illustrated in a morningcase (9 UTC) on August 20, 2012 (Fig. 4), when thePBL height difference is about 800 m. The shape ofthe [ΔU(z)]2 (ΔU(z)2 = (Δu)2 +(Δv)2, where Δ is thedifference between consecutive levels) profile is simi-lar in both cases, though [ΔU(z)]2 obtained by WRF isgreater than [ΔU(z)]2 obtained by measurements. Fur-ther, in the lower levels (below 500 m) the [ΔU(z)]2 istoo high in the simulations causing high gradients. Be-cause of the high gradient, the Rib number increases inthe lower levels over 0 resulting in a PBL height of about500 m (YSU-WRF). In the measurements, the Ribcr = 0is reached only at around 1300 m. Note that without highvalues of [ΔU(z)]2 on the lower levels, the model would



Figure 6: Area-averaged diurnal courses of PBL height for different soil textural classes in the sub-region within the Great Hungarian Plainon 5 July 2012 and 7 August 2012. The run types are specified in the figures. D = space/time averaged PBL height difference between thereference and the actual run in the sub-region and in the time period 1100–1500 UTC. STD = time averaged standard deviation of the PBLheight obtained for the actual run in the time period 1100–1500 UTC.

estimate about the same height for the PBL as is givenin the measurements.

YSU-WRF overestimates YSU-MEA on 5 July 2012,on 7 August 2012 and on 28 August 2012, but itis smaller than YSU-MEA on 20 August 2012. Atnight, YSU-WRF is always smaller than YSU-MEA. Itis noticeable that the SNR–YSU-MEA differences areat least as great as the YSU-WRF–YSU-MEA differ-ences. In extreme cases they reach about 1300 m. Atnight, SNR usually overestimates YSU-MEA. At mid-day, SNR approaches YSU-WRF somewhat better thanYSU-MEA. Summarizing, the PBL height courses ob-tained by the methods considered are considerably scat-tered, and they show somewhat better agreement aroundmidday than at night, which facilitates the investigation.In the case of YSU-WRF and YSU-MEA, the differ-ences originate from different wind speed profiles.

3.2 Spatial distribution of PBL height

Among the chosen days, 5 July 2012 is the most ex-treme. Hence, the spatial distribution of PBL height isillustrated for this day, which we present in Fig. 5 forreference run (run 00-00-HU) conditions at 1200 UTC.

The highest PBL height values reach 2400–2600 meters,they can be found in the Great Hungarian Plain. Muchlower PBL height values (500–1000 m) are above moun-tain and water surface regions as well as in the cloudyregions of western Transdanubia. Diurnal courses ofPBL height are analyzed in the sub-region denoted (seeFig. 5), where the sky was completely or almost com-pletely cloud-free during the daytime period.

3.3 Sensitivity tests

Space/time averaged PBL height differences betweenthe reference and the actual run in the sub-region and inthe time period 1100–1500 UTC on all five days are pre-sented in Table 5. The smallest PBL height differencesare obtained in comparisons 00-00-HU/00-00-HU-dksis(except 5 July 2012) and 00-00-HU/00-00-HU-corine.Note that PBL height differences are close to zero. Thesomewhat greater differences obtained on 5 July 2012are caused by differences in the simulated cloud coverin spite their small values. Somewhat larger PBL heightdifferences refer to comparisons 00-00-HU/00-00-US.The differences obtained in comparison 00-00-HU/00-01-HU are greater than the former and they are compara-



Table 5: Mean PBL height differences between the reference andthe actual run. Day numbers are as in Table 1.

Comparison Mean PBL height difference (D)[m] (reference− actual)

Days1 2 3 4 5

00-00-HU/00-00-US 46 34 26 11 4700-00-HU/00-01-HU 106 62 39 29 9900-00-HU/00-12-HU −494 −487 −400 −158 −47500-00-HU/05-00-HU −41 −22 −20 −19 −3200-00-HU/06-00-HU −224 −143 −139 −68 −17100-00-HU/11-00-HU −411 −289 −262 −122 −14500-00-HU/12-00-HU −437 −336 −305 −147 −17500-00-HU/14-00-HU −349 −274 −264 −134 −13200-00-HU/00-00-HU-dksis 141 −10 −17 2 −2600-00-HU/00-00-HU-corine 53 −10 −3 4 −4

Table 6: Significance test results characterizing the difference be-tween the diurnal courses of PBL height caused by soil changes.Day numbers are as in Table 1, Symbols: na–not applicable.

Comparison of the runs Probability levels(p-values)

Significantlydifferent Days

1 2 3 4 5

00-00-HU/00-00-US 0.10 yes na na na na0.05 yes na na na na0.01 no na na na na

0.001 no na na na na

00-00-HU/00-01-HU 0.10 na yes na yes yes0.05 na yes na yes yes0.01 na yes na yes yes

0.001 na yes na yes yes

00-00-HU/00-12-HU 0.10 na yes yes na yes0.05 na yes yes na yes0.01 na yes yes na yes

0.001 na yes yes na yes

00-00-HU/00-00-HU-dksis 0.10 no na na na na0.05 no na na na na0.01 no na na na na


ble in terms of absolute value to the differences obtainedin comparison 00-00-HU/05-00-HU. The averaged PBLheight difference from comparisons 00-00-HU/06-00-HU is −149 m, which in terms of absolute value ismuch higher than the former differences. The greatestPBL height differences can be found in comparisons 00-00-HU/00-12-HU, 00-00-HU/11-00-HU, 00-00-HU/14-00-HU and 00-00-HU/12-00-HU. On average they aregreater in terms of absolute value than 200 m. Here-inafter, the soil and land cover change effects will betreated separately.

3.3.1 Soil change

The results of significance testing regarding the differ-ence between the diurnal courses of PBL height caused

by soil changes are presented in Table 6. In compar-ison 00-00-HU/00-00-US, the method was applicableonly on 5 July 2012 (day 1). Then, the differencesare significant at significance levels of 0.05 and 0.10.In comparisons 00-00-HU/00-01-HU, 00-00-HU/00-12-HU the applicability of the method is somewhat better,namely, the method was applicable on three of the fivedays. The applicability is also small in comparison 00-00-HU/00-00-HU-dksis. In this case, the method is ap-plicable only on 5 July 2012, nevertheless, the differ-ences are insignificant at all significance levels thoughthey seem to be high (Table 5). Summarizing, the signifi-cance test method was not applicable in most cases (onlyin thirty-two cases from eighty). When it was applicable,the differences caused by soil changes were significantin most cases.

Diurnal courses are also analyzed from the physi-cal point of view. Among the five days, two such dayswere chosen which mostly differed from each other.These two days were 5 July 2012 and 7 August 2012.Area-averaged diurnal courses of PBL height obtainedin the sub-region for comparisons 00-00-HU/00-01-HUand 00-00-HU/00-12-HU on 5 July 2012 and on 7 Au-gust 2012 are presented in Fig. 6. Each plot contains in-formation on mean differences and standard deviations.“D” denotes the space/time averaged PBL height dif-ference between the reference and the actual run in thesub-region and in the time period 1100–1500 UTC (seealso Table 5), when PBL height values are the high-est. “STD” denotes the time averaged standard devia-tion of the PBL height obtained for the actual run inthe time period 1100–1500 UTC. The curves are pre-sented with their standard deviations. In these singletests, the cause and effect relationships are unequivo-cal. On changing soil texture, the plant available soilmoisture content (ASMC), as one of the most impor-tant soil parameters (Reen et al., 2006; Teuling et al.,2009) also changes. According to FAO, the sub-regionis covered by loam, clay loam and sandy loam textu-ral classes. These soil textures possess greater ASMCthan sand (Table 2), consequently, the ratio betweenthe actual and the plant available soil moisture content(RASMC) will be lower, that is, dryness will be greaterthan in the areas covered by sand. Of course, areas withlower RASMC will produce more intense turbulence,which is also reflected in the higher PBL height (com-parison 00-00-HU/00-01-HU). An analogous consider-ation could also be made for comparison 00-00-HU/00-12-HU. The clay textural class possesses greater ASMCthan loam, clay loam and sandy loam textural classes(Table 2), consequently, PBL height above clayey areas(run 00-12-HU) is significantly higher than the PBLheight above loamy areas (run 00-00-HU). With respectto the effect of changing soil map (Fig. 7), the devia-tions are negligable on 7 August, but somewhat greateron 5 July. These deviations are caused by the deviationsobtained in cloud cover prediction beginning from about1200 UTC. In this so-called accumulated test, the abovementioned cause-effect approach is less applicable.



Figure 7: Area-averaged diurnal courses of PBL height for differentsoil maps in the sub-region within the Great Hungarian Plain on5 July 2012 and 7 August 2012. The run types are specified in thefigures. D = space/time averaged PBL height difference betweenthe reference and the actual run in the sub-region and in the timeperiod 1100–1500 UTC. STD=time averaged standard deviation ofthe PBL height obtained for the actual run in the time period 1100–1500 UTC.

It is noticeable that PBL height courses on the twodays considerably differ from each other. The morningincrease rate of PBL height is much greater on 5 Julythan on 7 August. Considerable differences can also befound in the night time period. Calculated PBL heightvalues between 300–600 m in the pre-dawn period on7 August seem to be rather high, these could be at-tributed to strong mechanical forcing caused by an up-coming and crossing cold front. At the same time, thecold front’s cooling effect moderated the increasing rateof PBL height in the morning. The abrupt-like increaseof PBL height in the morning hours on 5 July is mostly

Table 7: Significance test results characterizing the difference be-tween the daily courses of PBL height caused by land cover typechanges. Day numbers are as in Table 1, Symbols: na–not applica-ble.

Comparison of the runs Probability levels(p-values)

Significantlydifferent Days

1 2 3 4 500-00-HU/05-00-US 0.10 na na yes yes yes

0.05 na na yes yes yes0.01 na na yes yes yes

0.001 na na yes yes yes

00-00-HU/06-00-HU 0.10 yes yes yes yes yes0.05 yes yes yes yes yes0.01 yes yes yes yes yes

0.001 yes yes yes yes yes

00-00-HU/11-00-HU 0.10 yes yes yes yes yes0.05 yes yes yes yes yes0.01 yes yes yes no yes

0.001 yes yes yes no yes

00-00-HU/12-00-HU 0.10 yes yes yes yes yes0.05 yes yes yes yes yes0.01 yes yes yes no yes

0.001 yes yes yes no yes

00-00-HU/14-00-HU 0.10 na yes yes yes yes0.05 na yes yes yes yes0.01 na yes yes no yes

0.001 na yes yes no yes

00-00-HU/00-00-HU-corine 0.10 yes na na na na0.05 no na na na na0.01 no na na na na


thermally driven, namely, the former day was similarlyhot and the cold front did not cause cooling in the sub-region.

3.3.2 Land cover change

The results of significance testing regarding the differ-ence between the diurnal courses of PBL height causedby land cover changes are presented in Table 7. In mostcases, the PBL height differences are significant at allsignificance levels investigated except in comparison00-00-HU/00-00-HU-corine, when the method was notapplicable on four days. On 5 July 2012, there is onlyone significant difference case at significance level 0.10.

Area-averaged diurnal courses of PBL height ob-tained in the sub-region for different land cover typeson 5 July and on 7 August, 2012 are presented in Fig. 8.Of course, the diurnal courses obtained on 5 July and on7 August are unequivocally different as in the previoustests discussed above. The results refer to three so-calledsingle tests (comparisons 00-00-HU/06-00-HU, 00-00-HU/12-00-HU and 00-00-HU/14-00-HU) and one so-called accumulated (comparison 00-00-HU/00-00-HU-corine) test. As mentioned, the single tests facilitate theunderstanding of cause/effect relationships, while accu-



Figure 8: Area-averaged diurnal courses of PBL height for different land cover types in the sub-region within the Great Hungarian Plainon 5 July 2012 and 7 August 2012. The run types are specified in the figures. D = space/time averaged PBL height difference between thereference and the actual run in the sub-region and in the time period 1100–1500 UTC. STD = time averaged standard deviation of the PBLheight obtained for the actual run in the time period 1100-=1500 UTC.

mulated tests do not enable this. According to USGS,the sub-region is covered by land cover type “drylandcropland and pasture” (vegetation type 2). Replacingthis land cover type by the land cover types denoted inTable 4, the albedo (α) decreases, while the minimumstomatal resistance (rstmin) and the roughness length (z0)increase in most cases. The considered land cover typechanges will produce higher sensible heat flux and, ac-cordingly, higher PBL height, namely, the decrease of αincreases the available surface energy flux, so the turbu-

lent transport, and the increase of rstmin and z0 contributeto the increase of sensible heat flux. These changes pro-duced by single tests are remarkably well representedin Table 5 and in Fig. 8. This straightforward think-ing could not be applied when performing accumulatedtests. This is demonstrated in comparison 00-00-HU/00-00-HU-corine (Fig. 9), when instead of the USGS landcover map, the land cover map Corine 2000 is used. Notethat PBL height differences are quite small and they canbe both positive and negative (Table 5).



Figure 9: Area-averaged diurnal courses of PBL height for differentland cover type maps in the sub-region within the Great HungarianPlain on 5 July 2012 and 7 August 2012. The run types are speci-fied in the figures. D = space/time averaged PBL height differencebetween the reference and the actual run in the sub-region and in thetime period 1100–1500 UTC. STD=time averaged standard devia-tion of the PBL height obtained for the actual run in the time period1100–1500 UTC.

3.3.3 Comparison of land cover and soil changes

Comparing D values obtained in the comparisons (Ta-ble 5) we can get an insight into the sensitivity of PBLheight to land cover and soil changes. It is unequivocalthat D values referring to the so-called single and ac-cumulated tests have to be separately considered. Thesmallest D = 11m value from the single soil change testis comparable to the smallest |D| = 19m value got inthe single land cover change test. Similarly, the greatest|D|= 494m value obtained in the single soil change testis comparable with the greatest |D| = 437m value ob-tained in the single land cover change test. In case of soil

change, the significance test method, unfortunately, wasnot even applicable (Table 6), while in case of land coverchange the difference was significant at all significancelevels (Table 7). In the so-called accumulated tests, theD values obtained were either positive or negative andamounted to a few meters or a few tens of meters ex-cept on 5 July 2012. In these cases, the differences wereinsignificant about 50 per cent of the time. Summariz-ing, the produced PBL height differences caused by soiland land cover changes are comparable to each other, atleast in the tests used, and they were much greater in theso-called single than in the so-called accumulated tests.

3.3.4 Sensitivity to Ribcr

As it was mentioned, the basic setting in WRF regard-ing Ribcr values is Ribun

cr = 0 and Ribstcr = 0.25. The

question arises, to what extent is the diurnal course ofPBL height sensitive to the changes of Ribun

cr and Ribstcr.

To try to answer the question, we used two settings inthe reference run: Ribun

cr = 0.25 and Ribstcr = 1 denoted

as run 00-00-HU-r1 and Ribuncr = 0.7 and Ribst

cr = 1 de-noted as run 00-00-HU-r2. The PBL height courses re-ferring to comparisons 00-00-HU/00-00-HU-r1 and 00-00-HU/00-00-HU-r2 on 5 July and 7 August 2012 arepresented in Fig. 10. D, as defined above, is negativeon all five days (not presented here), which is to be ex-pected since PBL height is proportional to Ribcr. Aroundmidday (11–15 UTC), the differences are small, at nightthey are somewhat greater. The greatest |D|= 130m ap-peared in comparison 00-00-HU/00-00-HU-r2 on 7 Au-gust 2012. Summarizing, PBL height course, especiallyaround midday, seems not to be sensitive to the changesof Ribcr. It is to be mentiod that a definite conclusioncould not be drawn, namely, some investigations (e.g.Jericevic and Grisogono, 2006) showed that the sen-sitivity depends on the intensity of the turbulence.

4 Conclusion

The sensitivity of WRF-simulated PBL height to landcover and soil changes is investigated in the CarpathianBasin on five summer days when shallow convectionwas prevailing because of anticyclone influence. So-called single and accumulated land cover type and soilchange tests are made. In total, seven single and threeaccumulated tests are performed for each day. Singletests are performed to be able to unequivocally separatethe cause and effect relationships, which is hardly possi-ble when performing accumulated tests. Land cover andsoil changes affect the atmosphere in completely differ-ent ways. By changing soil hydraulic properties, plantavailable soil moisture content changes causing shifts inthe partitioning of latent and sensible heat fluxes. Landcover type changes induce not only physical (albedoand roughness length) but also eco-physiological (e.g.minimum stomatal resistance) changes in land-surfacecharacteristics modifying both available energy flux andthe partitioning between latent and sensible heat fluxes.



Figure 10: Area-averaged diurnal courses of PBL height for different Ribcr value combinations in the sub-region within the Great HungarianPlain on 5 July 2012 and 7 August 2012. The run types are specified in the figures. D = space/time averaged PBL height difference betweenthe reference and the actual run in the sub-region and in the time period 1100–1500 UTC. STD=time averaged standard deviation of thePBL height obtained for the actual run in the time period 1100–1500 UTC.

PBL height is analyzed because it could serve as an in-tegral measure for characterizing turbulent mixing in thePBL. PBL is parameterized after the Yonsei Universityscheme (Hong et al., 2006). The analysis is performedfor the most simple case, that is, for cloud-free condi-tions.

The most important results are as follows. Themethods applied for validating PBL height diurnalcourses (SNR and YSU-MEA) show substantial devi-ations which can vary between −1000 m and 1500 m.The YSU-MEA–YSU-WRF differences are, in gen-eral, somewhat smaller than the SNR–YSU-MEA dif-ferences. These facts suggest that PBL height diur-nal course estimations are highly uncertain not onlybecause the measuring principles (SNR versus YSU-MEA) differ but also because the atmospheric pro-files determined using measurements or models (YSU-WRF–YSU-MEA) can also substantially differ. Con-cerning sensitivity analysis, the space/time averagedPBL height differences in the sub-region around mid-day (11–15 UTC) obtained in the single tests are muchgreater than those obtained in the accumulated tests. Thegreatest |D| = 494m value obtained in the single soilchange test 00-00-HU/00-12-HU highlights the impor-

tance of the soil texture clay. This is in accordance withthe findings of other researchers (e.g. Mölders, 2001,2005; Li et al., 1994). Similarly greater differences arederived from the single land cover change tests, forinstance, in comparison 00-00-HU/12-00-HU. In mostcases, these diurnal course PBL height differences aresignificant at all significance levels. |D| reaching a fewtens of meters can be treated as strongly significant.|D| values obtained in the accumulated tests seem to besmall, many times around zero. In these cases, the sig-nificance test method was frequently not applicable. Itis to be mentioned that the PBL height deviations ob-tained in validation tests (Fig. 3), at least in our case, aremuch greater than the deviations obtained in sensitivityanalyses (Figs. 6 and 8).

Further investigations are needed. Detailed diurnalcourse PBL height analyses are quite rare (e.g. Hernán-dez-Ceballos et al., 2012), irrespective of whethersimulated or observation-based. The courses are highlyuncertain even in such extremely simple weather situa-tions as ours. Of course, we had to choose as simple con-ditions as possible to be able to isolate the land-surfaceeffects, namely, in less ideal weather and terrain condi-tions this would not be possible.



5 Acknowledgements

The work was supported by OTKA (Hungarian Sci-entific Research Foundation) under contract numberK81432.

References

Ács, F., Á. Horváth, H. Breuer, F. Rubel, 2010: Effect of soilhydraulic parameters on the local convective precipitation. –Meteorol. Z. 19, 143–153.

Adegoke, J.O., R. Sr. Pielke, A.M. Carleton, 2007: Obser-vational and modelling studies of the impacts of agriculture-related land use change on planetary boundary layer processesin the central U.S.. – Agric. Forest Meteor. 142, 203–215.

Alapaty, K., S. Raman, D. Niyogi, 1997: Uncertainty in thespecification of surface characteristics: A Study of PredictionErrors in the Boundary Layer. – Bound.-Layer Meteor. 82,475–502.

Angevine, W.M., A.B. White, S.K. Avery, 1994: Boundary-layer depth and entrainment zone characterization with aboundary-layer profiler. – Bound.-Layer Meteor. 68, 375–385.

Breuer, H., F. Ács, B. Laza, Á. Horváth, I. Matyasovszky,K. Rajkai, 2012: Sensitivity of MM5-simulated planetaryboundary layer height to soil dataset: Comparison of soil andatmospheric effects. – Theor. Appl. Climatol. 109, 577–590.

Chen, F., R. Avissar, 1994: Impact of land-surface moisturevariability on local shallow convective cumulus and precipita-tion in large-scale models. – J. Appl. Meteor. 33, 1382–1401.

Chen, F., J. Dudhia, 2001: Coupling and advanced land surface-hydrology model with the Penn State-NCAR MM5 ModelingSystem. Part I. Model Implementation and Sensitivity. – Mon.Wea. Rev. 129, 569–585.

Cohn, S.A., W.M. Angevine, 2000: Boundary layer height andentrainment zone thickness measured by lidars and wind-profiling radars. – J. Appl. Meteor. 39, 1233–1247.

Cuenca R., M. Ek, L. Mahrt, 1996: Impact of soil water prop-erty parameterization on atmospheric boundary layer simula-tion. – J. Geophys. Res. 101, 7269–7277.

Drüszler, 2011: Meteorological effects of the land cover typeschanges during the twentieth century in Hungary. – PhD The-sis, University of West Hungary, Sopron, 137 pp, (in Hungar-ian).

Dudhia, J., 1989: Numerical study of convection observed dur-ing the winter monsoon experiment using a mesoscale two-dimensional model. – J. Atmos. Sci. 46, 3077–3107.

Ek, M., R.H. Cuenca, 1994: Variation in Soil Parameters:Implications for Modeling Surface Fluxes and AtmosphericBoundary-Layer Development. – Bound.-Layer Meteor. 70,369–383.

European Environment Agency, 2002: Corine LandCover 2000 (CLC2000) seamless vector data. – Avail-able at: http://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2000-clc2000-seamless-vector-database.

Feddema, J.J., 2005: A revised Thornthwaite-type global climateclassification. – Phys. Geography, 26, 442–466.

Garcia-Carreras, L., D.J. Parker, C.M. Taylor,C.E. Reeves, J.G. Murphy, 2010: Impact of mesoscalevegetation heterogeneities on the dynamical and termo-dynamic properties of the planetary boundary layer. –J. Geophys. Res. 115, D03102, DOI:10.1029/2009JD012811.

Garrett, A.J., 1982: A parameter study of interactions betweenconvective clouds, the convective boundary layer, and forestedsurface. – Mon. Wea. Rev. 110, 1041–1059.

Göndöcs, J., 2013: The impact of the spatial distribution ofsoil texture and land cover type on convective precipitation. –BSc Thesis, University Eötvös Loránd, Budapest, 37 pp. (inHungarian).

Gyöngyösi, A.Z., P. Kardos, R. Kurunczi, Z. Bottyán,2013: Development of a complex dynamical modeling systemfor the meteorological support of unmanned aerial operationin Hungary. – Conference Proceeding, International Confer-ence on Unmanned Aircraft Systems (ICUAS), 28-31 May2013, Atlanta, GA, IEEE. ISBN: 978-1-4799-0815-8, pp. 8-16, DOI:10.1109/ICUAS.2013.6564668

Hernández-Ceballos, M.A., J.A. Adame, J.P. Bolivar,B.A. De la Morena, 2012: The Performance of DifferentBoundary-Layer Parameterisations in Meteorological Mod-elling in a Southwestern Coastal Area of the Iberian Penin-sula. – ISRN Meteor. 2012, 983080.

Hong, X., M.J. Leach, S. Raman, 1995: A sensitivity study ofconvective cloud formation by vegetation forcing with differ-ent atmospheric conditions. – J. Appl. Meteor. 34, 2008–2028.

Hong, S.-Y., J. Dudhia, S.-H. Chen, 2004: A Revised Approachto Ice Microphysical Processes for the Bulk Parameterizationof Clouds and Precipitation. – Mon. Wea. Rev. 132, 103–120.

Hong, S.-Y., Y. Noh, J. Dudhia, 2006: A new vertical diffusionpackage with an explicit treatment of entrainment processes. –Mon. Wea. Rev. 134, 2318–2341.

Horváth, Á., F. Ács, H. Breuer, 2009: On the relationship be-tween soil, vegetation and severe convective storms: Hungar-ian case studies. – Atmos. Res. 93, 66–81.

Jericevic, A., B. Grisogono, 2006: The critical bulk Richard-son number in urban areas: verification and application in anumerical weather prediction model. – Tellus, 58A, 19–27.

Kain, J.S., 2004: The Kain-Fritsch convective parameterization:An update. J. Appl. Meteor. 43, 170–181.

Li, C., S. Frolking, R. Harriss, 1994: Modeling carbon bio-geochemistry in agricultural soils. – Global Biogeochem. Cy.8, 237–254.

Mahfouf, J.-F., E. Richard, P. Mascart, 1987: The influenceof soil and vegetation on the developement of mesoscale cir-culations. – J. Climate Appl. Meteor. 26, 1483–1495.

Mahrt, L., H.L. Pan, 1984: A two-layer model of soil hydrol-ogy. – Bound.-Layer Meteor. 29, 1–20.

McPherson, R.A., D.J. Stensrud, 2005: Influences of a WinterWeat Belt ont he Evolution of the Boundary Layer. – Mon.Wea. Rev. 133, 2178–2199.

McPherson, R.A., D.J. Stensrud, K.C. Crawford, 2004: Theimpact of Oklahoma’s winter wheat belt on the mesoscaleenvironment. – Mon. Wea. Rev. 132, 405–421.

Mlawer, E.J., S.J. Taubman, P.D. Brown, M.J. Iacono,S.A. Clough, 1997: Radiative Transfer for inhomogeneousatmospheres: RRTM, a validated correlated-k model for thelongwave. – J. Geophys. Res. 102, 16,663–16,682.

Mölders, N., 1999: Einfache und akkumulierte Land-nutzungsänderungen und ihre Auswirkungen auf Evapo-transpiration, Wolken- und Niederschlagsbildung. – Wis-senschaftliche Mitteilungen 15, Institut für Meteorologie derUniversität Leipzig, ISBN 3-9806117-4-4.

Mölders, N., 2001: On the uncertainty in mesoscale modelingcaused by surface parameters. – Meteor. Atmos. Phys. 76,119–141.

Mölders, N., 2005: Plant- and Soil-Parameter-Caused Uncer-tainty of Predicted Surface Fluxes. – Mon. Wea. Rev. 133,3498–3516.

Niyogi, D., Y. Xue, S. Raman, 2002: Hydrological Feedbackin Land Atmosphere Coupling: Comparison of a Tropical andmidlatitudinal Regime. – J. Hydrometeorol. 3, 39–56.

Pan, H.L., L. Mahrt, 1987: Interaction between soil hydrologyand boundary layer development. – Bound.-Layer Meteor. 38,185–202.


http://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2000-clc2000-seamless-vector-database

http://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2000-clc2000-seamless-vector-database

http://dx.doi.org/10.1029/2009JD012811

http://dx.doi.org/10.1109/ICUAS.2013.6564668


Pásztor, L., J. Szabó, Zs. Bakacsi, 2010: Digital processingand upgrading of legacy data collected during the 1:25.000scale Kreybig soil survey. – Acta Geodaetica et GeophysicaHungarica, 45, 127-136.

Rabin, R.M., S. Stadler, P.J. Wetzel, D.J. Stensrud,M. Gregory, 1990: Observed effects of landscape variabilityon convective clouds. – Bull. Amer. Meteor. Soc. 71, 272–280.

Reen, B.P., D.R. Stauffer, K.J. Davis, A.R. Desai, 2006:A case study on the effects of heterogeneous soil moistureon mesoscale boundary-layer structure in the Southern GreatPlains, U.S.A. Part II: mesoscale modelling. – Bound.-LayerMeteor. 120, 275–314.

Segal, M., R. Avissar, M.C. McCumber, R.A. Pielke, 1988:Evaluation of vegetation effects on the generation and modi-fication on mesoscale circulations. – J. Atmos. Sci. 45, 2268–2292.

Segal, M., W.E. Schreiber, G. Kallos, J.R. Garratt,A. Rodi, J. Weawer, R.A. Pielke, 1989: The impact of cropareas in nordeast Colorado on midsummer mesoscale thermalcirculations. – Mon. Wea. Rev. 117, 809–825.

Shao, Y., P. Irannejad, 1999: On the choice of soil hydraulicmodels in land-surface schemes. – Bound.-Layer Meteor. 90,83–115.

Skamarock, W.C., J.B. Klemp, J. Dudhia, D.O. Gill,D.M. Barker, M.G. Duda, X.-Y. Huang, W. Wang,J.G. Powers, 2008: A Description of the Advanced ResearchWRF Version 3 NCAR/TN–475+STR, June 2008. – NCARTechical Note.

Teuling, A.J., R. Uijlenhoet, B. van den Hurk, S.I. Senevi-ratne, 2009: Parameter sensitivity in LSMs: an analysis usingstochastic soil moisture models and ELDAS soil parameters. –J. Hydrometeorol. 10, 751–765.

White, A.B., 1993: Mixing Depth Detection Using 915-MHzRadar Reflectivity Data. – 8th Symposium on Meteorologi-cal Observations and Instrumentation, Anaheim, California,Jan. 17–22. – Amer. Meteor. Soc. 248–250.

Wyngaard, J.C., M.A. LeMone, 1980: Behavior of the refrac-tive index structure parameter in the entraining convectiveboundary layer. – J. Atmos. Sci. 37, 1573–1585.


sensitivity of wrf-simulated planetary boundary layer...

Documents