Research Article Simulating the Energy and Water Fluxes...
Transcript of Research Article Simulating the Energy and Water Fluxes...
Research ArticleSimulating the Energy and Water Fluxes from Two AlkalineDesert Ecosystems over Central Asia
Chang-Qing Jing12 and Long-Hui Li2
1College of Grassland and Environment Sciences Xinjiang Agricultural University Urumqi 830052 China2State Key Laboratory of Desert and Oasis Ecology Xinjiang Institute of Ecology and GeographyChinese Academy of Sciences Urumqi 830011 China
Correspondence should be addressed to Long-Hui Li lhlimsxjbaccn
Received 15 September 2015 Revised 13 November 2015 Accepted 16 November 2015
Academic Editor Sergio M Vicente-Serrano
Copyright copy 2016 C-Q Jing and L-H Li This is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work is properlycited
The Central Asia region is covered by vast desert ecosystems where the characteristic of energy and water fluxes is different fromother humid ecosystems The application of land surface models (LSMs) in arid and semiarid ecosystems was largely limitedThis paper presents a detailed evaluation of Common Land Model (CoLM) at two eddy covariance (EC) sites in alkaline desertecosystems over Central Asia Simulations of the net radiation (119877net) latent heat flux (119876le) sensible heat flux (119876ℎ) and soiltemperature showed that refined estimate of roughness length (1198850119898) significantly improved the performance of CoLM in simulatingturbulent heat fluxes 119876le was increased but 119876ℎ was decreased which were in better agreement with the observations from ECsystemThe results indicated that accurate parameterization of 1198850119898 is of crucial importance for predicting energy and water fluxesin LSM when applied in Central Asia desert ecosystems Sensitivity analysis regarding leaf area index (LAI) 1198850119898 and albedo (120572)showed that 119877net is very sensitive to 120572 but119876le119876ℎ and soil temperature (119879soil) are sensitively varying with the estimate of 1198850119898 at thetwo EC sites over Central Asia
1 Introduction
Terrestrial ecosystem is an important part of the earthsystem as it controls land-atmosphere interactions Landsurface model which described plant physiological behaviorin relation to soil and atmospheric processes was widely usedto quantify the land-atmosphere interactions for improvingthe predictability of Earth System Models (ESMs) [1ndash4]LSM consists of different biogeophysical and biogeochemicalprocesses which described energy momentum water andCO2 exchange between the atmosphere and the land sur-face LSMs became increasingly complicated since Manabeintroduced the first bucket model in 1969 [5] Increases inobservations obtained from in situ instruments and satellitesprovidedmore opportunities to test validate and evaluate theperformance of LSMs in specific terrestrial ecosystem [6 7]Despite the significant improvement of LSMs during the pastfew decades large errors and uncertainties still existed [8]Errors and uncertainties underlying the LSMs can result from
model structure model initial conditions model parametersand model forcing data and evaluating the performance ofLSMs among different ecosystems is helpful to identify anddiagnose the errors of the model in structure and parametersand hence to reduce the uncertainties [9]
Evaluating LSMs would provide a reliable indicator ofperformance of models under various climates and amongdifferent biomes Comparing themodel results withmeasure-ments is a routine approach in the assessment of a particularLSM [10] Application of LSMs was intensively implementedin humid and dense vegetation regions but LSM was rarelyinvestigated in dryland or sparse-vegetated areas (eg desertsecosystems in Central Asia) [11 12] Absence of the exquisitedescription on specially ecophysiological characteristics ofdryland ecosystems further enlarged the errors of the LSMsin simulating turbulent fluxes [13 14] Under humid areasLSMs were found to performwell [12] However around one-third of the worldrsquos area was belonging to dryland ecosystems[15] where the land surface fluxes such as 119876ℎ 119877net and
Hindawi Publishing CorporationAdvances in MeteorologyVolume 2016 Article ID 4849525 13 pageshttpdxdoiorg10115520164849525
2 Advances in Meteorology
Table 1 Location soil vegetation climate factors and flux time series at KZ-Ara site and KZ-Bal site
Site Lon Lat Soil Dominant vegetation Temperature (∘C) Precipitation (mm) Simulation periodKZ-Ara 61∘051015840E 45∘581015840N Alkaline Shrub 83 1405 MayndashAug 2012KZ-Bal 76∘391015840E 44∘341015840N Alkaline Grass 57 1402 MayndashSep 2012
ground surface temperature are less dependent on vegeta-tion parameters and soil hydraulic parameters but moreon energy-related soil parameters and surface parameters(surface albedo the ground surface emissivity aerodynamicroughness length and thermal roughness length) [16] Unfor-tunately application of LSMs in arid and semiarid ecosystemswas largely limited For example recent studies have foundthat CoLM extremely underpredicted latent heat fluxes indryland areas [13 17] Prediction errors can be derived fromthe forcing data physics processes and the parameterizationsof land characteristic (which mainly included vegetation andsoil effects) [2 18] Vegetation effects consist of differentplant types that differ in leaf areas root profile heightoptical properties stomatal conductance and roughnesslength Soil properties can be soil texture (percent of sandand clay) and soil thermal and hydraulic parameters Theerrors resulting from parameterizations of these processeswill cause large errors for LSMs in modelling the turbulentfluxes
TheCentral Asia dryland area is characterized by a typicalarid continental climate complex terrain sparse vegetationand high heterogeneity in land surface which inevitablyproposes high challenges on the application of LSMs [19]Investigating the energy and water exchange between landsurface and atmosphere was of great importance to hydro-logical and ecological research [15 20] Few recent studieshave evaluated the performance of the CoLM to reproduceenergy and water vapor fluxes in Chinarsquos desert ecosystem[13 21]They have found that root distribution and root wateruptake function have significant effects on the performanceof CoLM in estimating energy and water fluxes However120572 1198850119898 and LAI are also expected to significantly impactthe performance of LSMs which largely influence energybudget thermodynamic resistance and turbulence fluxesThe errors resulting from these critical parameterizationswill cause large error and uncertainties for LSMs in theestimates of sensible and latent heat fluxes Unfortunately acomprehensive evaluation of LSMs and their uncertaintiescaused by 120572 1198850119898 and LAI was never investigated in CentralAsia dryland ecosystems
Therefore the main objectives of this research are (1) toevaluate the performance of CoLM of two desert ecosystemsin Kazakhstan using EC observations during the growingseasons for the purpose of investigating how well the CoLMis able to simulate the energy and water fluxes over CentralAsia land surface and (2) to analyze model sensitivities toalbedo roughness length and leaf area index
2 Materials and Methods
21 Site Description Two sets of EC equipment were estab-lished in representative dryland ecosystem in Kazakhstan
in 2012 (Figure 1) One site is close to Aral Sea and theother is close to Balkhash Lake The Aral Sea site (KZ-Ara 6108∘E 4596∘N) is located northeast of the Aral Seaand at the edge of Aralkum Desert Within radius of 5 kmof the KZ-Ara site the dominant plant species are weed-grass and reed vegetation in combination with xerophyticand halophytic desert semishrubs and shrubs The averagefraction of vegetation coverage was about 30The dominantsoil type was solonchak Historical climatic records with longterm observations showed that mean annual precipitation is140mm and mean annual air temperature is 83∘C (Table 1)The Balkhash Lake site (KZ-Bal 7663∘E 4457∘N) is locatedbetween the Balkhash Lake and the Kapchagay Reservoirand between original deserts and oasis croplands Withinradius of 5 km of the KZ-Bal site the dominant plantspecies are irrigated crops grasses and desert shrubs Theaverage fraction of vegetation coverage was about 40 Thesoil in the KZ-Bal site is saline solonchak [22] Historicalclimatic records with long term observations showed thatmean annual precipitation is 140mm and mean annual airtemperature is 57∘C (Table 1) For detailed information aboutthe two sites refer to Li et al [23]
22 Eddy Covariance and Ancillary Measurements At eachsite a set of EC systems was used to measure energy watervapor and CO2 fluxes between the land surface and theatmosphere The eddy covariance consists of an open pathgas analyzer (LI-7500 LICOR) and a three-dimensionalsonic anemometer thermometer (Wind Master Pro GillInstruments Lymington UK) The EC system also measuresair temperature and humidity (HMP45C Campbell) pre-cipitation (TE525MM Texas Electronics Dallas TX USA)short-wave and long-wave radiation (CNR-1 Kipp amp ZonenDelft Netherlands) soil temperature (TCAV Campbell)soil moisture content (CS616 Campbell Sci) and soil heatflux (HFP01 Hukseflux Delft Netherlands) The open pathgas analyzer and the three-dimensional sonic anemometerthermometer are mounted at 20m above the ground Soiltemperature soilmoisture content and soil heat flux (119866) weremeasured at 20 40 60 and 80 cm depth below the groundTherefore this study chose the interpolated soil temperaturein CoLMon the corresponding depth for comparisonsThesedata were logged every 30min by the data logger at 10Hz andall variables were carried out with data processing and gapfillingThe data was used for the analysis in this study duringgrowing seasons at the two sites and covered the period from30April to 18 August 2012 at the KZ-Ara site and from 23Mayto 6 September 2012 at the KZ-Bal site
23 Common Land Model CoLM [18] is one of the widelyused land surface models which was originally proposed toprovide a framework for a truly community developed land
Advances in Meteorology 3
KZ-BalKZ-Ara
Figure 1 Study area and location of the two eddy covariance (EC) sites in Kazakhstan
component of the National Centre for Atmospheric Research(NCAR) Community Climate System Model (CCSM) [2425] and it was subsequently developed by an open collabora-tion of many scientists A variety of comprehensive multiyearpoint observational data over different regions of the worldhave been used [26ndash28] These data were included in theproject for the Intercomparison of Land ParameterizationSchemes [29] and the results from these extensive tests wereuseful for CoLM improvement
CoLM simulates the exchange of energy mass andmomentum between the atmosphere and terrestrial bio-sphere through a wide range of ground and canopy biogeo-physical processes and allows for the simulation of explicitbiophysical mechanisms including the representation ofthermodynamic hydrologic and physiological processes [1830] In CoLM soil temperatures are predicted using a heatdiffusion equation in 10 soil layers and the soil thermal con-ductivity depends on soil water density Surface evapotranspi-ration consists of evaporation of precipitation intercepted byleaves canopy transpiration and bare soil evaporation Thesensible heat and water vapor fluxes between the surface andthe reference height can be written in the following form
119876ℎ = minus120588119886119888119901
(120579119886 minus 120579119904)
119903119886ℎ
119864 = minus120588119886
(119902119886 minus 119902119904)
119903119886119908
(1)
where 119876ℎ is sensible heat flux and 119864 is water vaporflux 120588119886 is air density and 119888119901 is specific heat for dry air(100467 J kgminus1 Kminus1) 120579119886 and 120579119904 are air potential temperatureat reference height and surface potential temperature respec-tively 119902119886 and 119902119904 are water vapor specific humidity at referenceheight and surface air humidity respectively 119903119886ℎ and 119903119886119908
are the aerodynamic resistances for heat and water vaporrespectively which are crucial parameters in determining119876ℎ and 119864 calculated by Monin-Obukhov similarity theory[31] and mainly dependent on the thermal roughness length(1198850ℎ) and 1198850119898 [16] as well as the meteorological forcingvariables in CoLM such as air temperature wind speed andprecipitation
24 Model Simulations and Sensitivity Analysis The versionof CoLM which integrated an optimal root water uptakefunction for dryland ecosystem [21] was used (termed S0)in this study To investigate the effect of roughness lengthon the performance of CoLM a simulation driven by anempirical estimate of roughness length (S1) was conducted(Table 2) In S0 simulation momentum roughness lengththermal roughness length and the water vapor roughnesslength (1198850119908) were all defined the same in CoLM for baresoil (1198850119898 = 1198850ℎ = 1198850119908 = 005) However some researchin dryland ecosystems indicated that the abovementionedroughness length values have different magnitude [4 32ndash34]As in the previous studies Chen et al [16] found that the
4 Advances in Meteorology
Table 2 The parameterizations of the reference model (S0) and the model with refined roughness length (S1) in CLM at KZ-Ara site andKZ-Bal site RWUF is an optimal root water uptake function for dryland ecosystem (Jing et al 2014 [21]) and119882119888 119882119909 and 119898 are empiricalconstants in RWUF
Site Simulation Land cover type Vegetation fraction RWUF Roughness length for bare soil(119882119888119882119909119898) 1198850119898 1198850ℎ 1198850119908
KZ-Ara S0 Mixed shrub and grassland 30 (08054) 5119864 minus 2 5119864 minus 2 5119864 minus 2
S1 S0 S0 S0 2119864 minus 3 2119864 minus 4 2119864 minus 4
KZ-Bal S0(1) Irrigated cropland and pasture (35)(2) Herbaceous wetland (44)(3) Shrub (21)
40 (03074) 5119864 minus 2 5119864 minus 2 5119864 minus 2
S1 S0 S0 S0 2119864 minus 3 2119864 minus 4 2119864 minus 4
Table 3 A baseline of reference exprement (S0) and six different configurations of CLMat each site (S2ndashS7) as used in this study for sensitivitytesting Two sets of leaf area index (LAI) average roughness length (1198850119898) and average albedo (120572) were used in CLM ldquo119863rdquo refers to the modeldefault parameter setting
KZ-Ara KZ-BalSimulation LAI 1198850119898 120572 Simulation LAI 1198850119898 120572
S0 119863 119863(005) 119863 S0 119863 119863(01) 119863
S2 119863 lowast 07 119863 119863 S2 119863 lowast 07 119863 119863
S3 119863 lowast 13 119863 119863 S3 119863 lowast 13 119863 119863
S4 119863 0005 119863 S4 119863 001 119863
S5 119863 05 119863 S5 119863 1 119863
S6 119863 119863 119863 lowast 07 S6 119863 119863 119863 lowast 07
S7 119863 119863 119863 lowast 13 S7 119863 119863 119863 lowast 13
surface flux and temperature have different sensitivities to1198850ℎand 1198850119898 and the sensible heat flux is very sensitive to theparameterization schemes of 1198850ℎ in arid regions [16]
A common method to calculate 1198850119898 and 1198850ℎ is
1198850119898 = 119911119890(minus119896119880119906
lowast)minus120593119898(119911119871)
1198850ℎ = 119911119890(minus119896(119879minus119879
119904)119879119904)minus120593ℎ(119911119871)
(2)
where 119911119871 is a stability parameter (119871 is the Monin-Obukhovlength and 119911 is the observational height) and 120593119898(119911119871) isthe stability function of wind profile and becomes 0 at theneutral condition 119880 is the average wind speed and 119906lowast isthe surface friction velocity 120593ℎ(119911119871) is the stability functionof the temperature profile and becomes 0 at the neutralcondition 119896 is vonKarman constant and equals 04 generally119879 is the air temperature and119879119904 is the surface temperatureTherelationship between 1198850119898 and 1198850ℎ can be described as
ln(1198850119898
1198850ℎ
) = 119896119887minus1 (3)
where 119896119887minus1 can be obtained from the bulk transfer equationas
119896119887minus1=119896119906lowast (119879119904 minus 119879)
119867120588119888119901
minus [ln119911 minus 1198890
1198850119898
minus 120593ℎ (119911
119871)] (4)
where119867 is the observed sensible heat flux 120588 is the air density119888119901 is the specific heat for dry air and 1198890 is the zero plane
displacement Thus the relationship among 1198850119898 1198850ℎ and1198850119908 is described as
1198850ℎ = 1198850119908 = 1198850119898119890minus119896119887minus1
(5)
where the excess resistance to heat transfer 119896119887minus1 is importantto the sensible heat exchange between land surface andatmosphere and there are linear correlations between 119896119887
minus1
and surface temperature1198850119898 is physically related to the geometric roughness of
surface elements and can be derived from the wind speedand temperature profiles Bao et al [4] and Yang et al [34]argued that this scheme overestimated 1198850ℎ and would mis-estimate the energy and water fluxes Momentum transportis more efficient than heat transport due to the influence ofpressure fluctuation because individual roughness elementsmay enhance the momentum flux through form drag withlittle contribution to the area-averaged heat flux [35]
An appropriatemethod suggested by Zhang et al [36] wasemployed to estimate1198850119898 at desert ecosystems In desert1198850119898was estimated as 00019 plusmn 00071m and 1198850ℎ and 1198850119908 were atsame order but almost onemagnitude lower than1198850119898 [36] Inthis study1198850119898 was set as 0002m and1198850ℎ = 1198850119908 = 00002mfor bare soil at the two sites (Table 2)
This research further investigated the sensitivities of theCoLM to LAI 1198850119898 and 120572 Therefore other six simulationsby increasing or decreasing the value of each parameter wereconducted The specifications of all simulations are listedin Table 3 In order to avoid the cross-influence of eachparameter all sensitivity tests took the simulation S0 as areference
Advances in Meteorology 5
0 200100 300 500400 600 700minus100
minus100
0
100
200
300
400
500
600
700
Qle+Qh
(W m
minus2)
y = 076 lowast x + 2895
R2= 091
RMSE = 4645
Rnet minus G (W mminus2)
(a) KZ-Ara
minus100 5004003002001000 600 700
minus100
0
100
200
300
400
500
600
700
y = 095 lowast x + 156
R2= 097
+Qh
(W m
minus2)
Rnet minus G (W mminus2)
3519RMSE =
Qle
(b) KZ-Bal
Figure 2 Energy balance closure at the KZ-Ara site and the KZ-Bal site The slope of the fitted line represents energy closure ratio and 1198772 isthe coefficient of determination RMSE (Wmminus2) is the root mean square error The energy fluxes include sensible heat flux (119876ℎ) latent heatflux (119876le) net radiation (119877net) and ground heat flux (119866)
25 Statistical Analysis Energy balance ratio (EBR) [35] wasused to give an overall evaluation of energy balance closure byaveraging over random errors in the half-hourmeasurementsat two flux tower sites and it was calculated by
EBR =sum119899
119894=1(119876le + 119876ℎ)
sum119899
119894=1(119877net minus 119866)
(6)
where 119899 is the number of half hours of data The values ofEBR close to 1 indicate the best degree of energy balanceclosure Additionally coefficient of determination (1198772) rootmean square error (RMSE) slope (119887119904) and intercept (1198870) areused to justify the performance of the model
The Taylor diagram [37] was used to quantify the degreeof the sensitivities of the model to management parametersRMSE 1198772 and standard error (STD) are used in the TaylordiagramThe output of the model simulation is specified by asingle point with the STDbeing the polar axis and119877 the polarangle The ldquoreferencerdquo point represents observations and theother points refer to themodel results from the simulations ofsensitivity testsThe distances from the reference point to theother points representing the consequence of the relationshipindicate the RMSE The higher 119877 and the smaller the STDandRMSE the better the agreement betweenmodel and dataWhen comparing two simulations with different parametervalues the longer the distance between the two simulationpoints the greater the sensitivity to that parameter
3 Results
31 Energy Balance Closure The slopes of the linear regres-sion between the observed 119876le + 119876ℎ and 119877net minus 119866 were 076and 095 at KZ-Ara and KZ-Bal respectively The coefficient
of determination (1198772) of the observed 119876le + 119876ℎ and 119877net minus 119866was 091 and 097 and the root mean square error (RMSE)was 4645 and 3519Wmminus2 respectively (Figure 2) Energybalance ratio (EBR) at KZ-Ara and KZ-Bal was 111 and 106respectively
32 Modelled 119877119899119890119905 119876119897119890 119876ℎ and 119879119904119900119894119897 Figure 3 shows thecomparisons between the measurements and the simulationsof the reference model (S0) and the model with refinedroughness length (S1) for 119877net 119876le and 119876ℎ at the twoKazakhstan sites The reference model (S0) significantlyunderestimated the latent heat flux and overestimated thesensible heat flux at both sites However the performanceof the refined roughness length (S1) was largely improvedin simulating turbulent heat fluxes The latent heat flux wasincreased and sensible heat flux was decreased obviouslyAt the KZ-Ara site RMSE for 119877net decreased from 636 to395Wmminus2 1198772 values for 119876le given by the two simulations(S0 and S1) were 036 and 061 respectively and RMSEdecreased from 4115 in S0 to 333Wmminus2 in S1 (Table 4)For 119876ℎ simulations the RMSE for S0 was 12124Wmminus2as compared to 5847Wmminus2 for S1 (Table 4) The resultsindicated that the simulation with refined roughness length(S1) significantly improved the performance of the model forboth 119877net and 119876le and particularly for 119876ℎ At the KZ-Bal sitethe performance of the simulation with refined roughness(S1) was also greatly improved 1198772 values for 119876le given by S0and S1 were 09 and 092 respectively and RMSE decreasedfrom 4642 in S0 to 4311Wmminus2 in S1 1198772 values for 119876ℎgiven by the two simulations were 07 and 067 respectivelyand RMSE decreased from 7738 in S0 to 3797Wmminus2 in S1(Table 4 Figure 3)
6 Advances in Meteorology
Table 4 Model performance for simulating 119877net 119876le and 119876ℎ indicated by coefficient of determination (1198772) slope (119887119904) intercept (119887
0) and
root mean square error (RMSE Wmminus2) of linear regressions between model and observed data at the KZ-Ara site and the KZ-Bal site
Variables Reference model (S0) Refined roughness length (S1)1198772 RMSE 119887119904 1198870 119877
2 RMSE 119887119904 1198870
Site KZ-Ara119877net 097 636 112 2848 096 395 099 712119876le 036 4115 074 966 061 333 102 467119876ℎ 089 12124 186 566 086 5847 106 4693
Site KZ-Bal119877net 098 4386 104 227 098 3307 096 139119876le 09 4642 082 3057 092 4311 089 2867119876ℎ 07 7738 158 5087 067 3797 076 2908
To further investigate the effects of refined roughnesson the energy fluxes simulations Figure 4 shows the meandiurnal turbulent fluxes during growing seasons at the twosites Diurnal variations of the three components of energyfluxes showed typical characteristics at the KZ-Ara site thatis 119877net gt 119876ℎ gt 119876le (Figure 4(a)) The KZ-Bal site is locatedbetween oasis croplands and original deserts Although theamount of the average annual precipitation is similar to KZ-Ara site the characteristic of the energy fluxes allocationshowed higher latent heat flux and lower sensible heat fluxcompared to theKZ-Ara site (Figure 4(b)) At theKZ-Ara siteS0 overestimated 119877net with the peak value of 520Wmminus2 ascompared to the observed peak value 450Wmminus2 Howeverthe simulation with refined roughness (S1) produced a goodagreement for 119877net between the simulation and the measure-ments In addition S0 overestimated 119876ℎ with the peak valueof 400Wmminus2 at noontime as compared to 200Wmminus2 for themeasurements S1 improved the simulation for 119876ℎ as well AtKZ-Bal site both S0 and S1 agreed better with observationsfor 119877net and 119876le S0 significantly overestimated 119876ℎ at thedaytimeHowever the simulationwith refined roughness (S1)produced better agreement with the observations
Figure 5 showed the comparisons of soil temperaturebetween the observation and the simulations of the referencemodel (S0) and the model with refined roughness length (S1)at each site Although there are significant challenges when itcomes to validating soil temperature fromLSMbecause of thehigh sensitivities of simulated andmeasured soil temperatureto soil texture moisture conditions and the limitations ofthe measurement the results from the reference model (S0)and the model with refined roughness length (S1) appearedto be realistic and generally captured the seasonal variationfor soil temperature It was found that the modelled 119879soil inMay was underestimated and significantly improved in Juneand July but overestimated in August However S1 producedhigher values of soil temperature than S0 Additionally theCoLM simulated a smaller vertical soil temperature gradientas shown that the contour for the simulationswasmore sparsecompared with the observation
33 Sensitivity of CoLM to LAI 1198850119898 and 120572 In orderto investigate the sensitivities of CoLM the values of theparameters were adjusted in large ranges Key parameters
consist of LAI 1198850119898 and 120572 with regard to the simulationof 119877net 119876le 119876ℎ and 119879soil at the two EC sites A referencemodel (S0) and six independent sensitivity tests (S2ndashS7 seeTable 3) were conducted at each siteThis research only chosethe modelled soil temperatures at the depth of 20 cm forsensitivity tests in this study
The modelled 119876le and 119876ℎ in CoLM were divided intotwo parts the fluxes on vegetation leaves and the fluxes onthe ground Thus 119876le showed a more complicated sensitivityto LAI Since the leaf temperature increased and groundtemperature decreased with the increase in LAI the evap-otranspiration from the leaves improved and the groundevaporation weakened but the total latent heat flux wasincreased and exhibited a strong sensitivity for 119876le at KZ-Ara When it came to KZ-Bal the vegetation coverage waslarger than KZ-Ara while continuing to increase LAI hasslightly further improved the performance of 119876le Similarly119876ℎ and 119879soil showed a strong sensitivity to LAI The more thevegetation the more the solar radiation intercepted When itcame to1198850119898 the values of 119877net and119876ℎ were increased but119876leand 119879soil were decreased with the increase in 1198850119898 Figure 6indicated that 119876le 119876ℎ and 119879soil were highly sensitive to 1198850119898Taking the KZ-Ara site as an example 1198772 values for themodelled 119876le improved from 04 to 07 when 1198850119898 decreasedfrom 05 (S4) to 0005 (S5) Figure 6 also showed a strongsensitivity of albedo to 119877net 119876ℎ and 119879soil The imprecisesettings of albedomay enlarge the errors for sensible heat fluxin CoLMGround sensible heat flux was negatively correlatedto albedo The increase in the surface albedo decreasedthe solar radiation absorbed by soil and soil temperaturewas decreased However 119876le was slightly sensitive to albedo(Figure 6)
These sensitivity analysis results demonstrated that theimprovement in model performance observed in S2ndashS7 wassignificantly affected by the values of these three parametersin CoLM further justifying the significance of these keyparameters (LAI 1198850119898 and albedo) to the Common LandModel
4 Discussion
Arid and semiarid areas cover approximately one-third ofthe global terrestrial land surfaces [15] Central Asia has
Advances in Meteorology 7
KZ-AraSi
mul
atio
ns (W
mminus2)
S0
minus200
0
200
400
600
800
400 600 8000 200minus200
Observations (W mminus2)
(a) 119877net
KZ-AraS1
minus200
0
200
400
600
800
200 400 600 800minus200 0
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(b) 119877net
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400minus100
minus100
0
100
200
300
400
KZ-AraS0
Observations (W mminus2)
(c) 119876le
100
minus100
0
100
200
300
400
200 300 400minus100 0
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(d) 119876le
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
minus100
0
100
200
300
400
500
KZ-AraS0
Observations (W mminus2)
(e) 119876ℎ
minus100
0
100
200
300
400
500
minus100 0 100 200 300 400 500
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(f) 119876ℎKZ-Bal
Sim
ulat
ions
(W m
minus2)
S0
0 200 400 600 800minus200
minus200
0
200
400
600
800
Observations (W mminus2)
(g) 119877net
KZ-BalS1
minus200
0
200
400
600
800
minus200 0 200 400 600 800
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(h) 119877net
Figure 3 Continued
8 Advances in Meteorology
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
0
200
400
Observations (W mminus2)
KZ-BalS0
(i) 119876le
KZ-BalS1
0
200
400
minus100 0 100 200 300 400 500
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(j) 119876le
Sim
ulat
ions
(W m
minus2)
minus100
0
100
200
300
0 100 300minus100
Observations (W mminus2)
KZ-BalS0
200
(k) 119876ℎ
minus100
0
100
200
300
0 100 200 300minus100
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
KZ-BalS1
(l) 119876ℎ
Figure 3 Comparison between the measured half-hourly net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the valuessimulated by the reference model (S0) and the model with refined roughness length (S1) at the KZ-Ara and KZ-Bal sites The solid red linerepresents the linear regression between the simulation and the observed data and the dashed line represents a 1 1 relationship between thedatasets
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(a) KZ-Ara
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(b) KZ-Bal
Figure 4 Comparison between the measured net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the values simulatedby the reference model (S0) and themodel with refined roughness length (S1) on a diurnal course at the KZ-Ara and KZ-Bal sitesThe diurnalflux values were calculated as the mean values of all data at same measurement time in a day for the entire time period
Advances in Meteorology 9
20
30
40
50
60
70
80
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
Soil
dept
h (c
m)
DOY in 2012
10
20
30
40
(a)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
40555555555555555555555555555555555 1
(b)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
405555555555555555555 1
(c)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(d)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(e)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(f)
Figure 5 Panels (a) and (b) are the isograms changed with time and depth variation for the reference model (S0) and the model with refinedroughness length (S1) and (c) shows the corresponding observed data at the depth of 20ndash80 cm below ground at the KZ-Ara site panels (d)and (e) are the isograms for S0 and S1 and (f) shows the observations at the KZ-Bal site
vast proportion of dryland ecosystems where climate wasfeatured as hot and dry during growing season [19] and thevegetation is sparseThe characteristic of dryland ecosystemsis significantly different from those in humid areas in termsof surface energy budget Many researchers have focused onecosystem functioning and structure in response to climatechange Kariyeva et al [38] examined spatiotemporal changepatterns and interactions between terrestrial phenology andclimate change in Central Asia during the period 1981ndash2008Lioubimtseva et al [19] have found that local and regionalhuman impacts in arid zones can significantly modify landsurface albedo as well as water exchange and nutrient cyclesthat could have essential impacts on the climate at both theregional and global scales Such kind of analyses advancedour understanding on the response of dryland ecosystembehaviour and functioning to climate change Howeverapplying LSM into dryland ecosystems was far more limitedRecently the CoLM has been validated at one desert shrubecosystem in Xinjiang China and the results found that rootfraction and root water uptake have important effects on theperformance of CoLM in simulating energy and water fluxes[13 21] In this study a refined parameterization of CoLMhasbeen evaluated at two newly built EC sites in Kazakhstan
The most commonly used technique to obtain landsurface turbulent fluxes is measurement of eddy covariancesystem and the analyzer was based on flux footprint modelsThe footprint concept is the probability that a scalar comingfrom a given elemental source reaches the measurementpoint Footprint models describe the relationship between
the spatial distribution of surface sources and the measuredsignal using footprint functions Several flux footprintmodelshave been designed [39ndash42] But most of them cannotaccount for inhomogeneous turbulence or require largercomputational resources Gockede et al [43] improved anEulerian footprint model use of satellite maps for explicitassignment of surface type Gockede et al [44] and Rebmannet al [45] applied this newmodel at the EC sites and obtainedsatisfactory results At present footprint models are used toestimate the source areas contributing to the flux observa-tions In addition they provide a tool for quality control ofthe flux measurements and provide guidance in designingexperiments [46] Thus the footprint models have consider-able potential in microclimatology investigations especiallyin studies which include nonhomogeneous surfaces
The momentum roughness length (1198850119898) thermal rough-ness length (1198850ℎ) and the water vapor roughness length(1198850119908) are crucial parameters for calculating momentum andheat fluxes in bulk transfer equations which is one of theessential components in LSMs It has been widely observedthat 1198850119898 differs from 1198850ℎ and 1198850119908 [4 47] Unfortunately1198850119898 1198850ℎ and 1198850119908 up to date are still treated as constants inmost LSMs Inaccurate estimates of roughness length wouldenlarge the bias of simulated energy andwater fluxes in LSMsMany researchers have found that roughness length stronglydepended on surface heterogeneity vegetation height andcoverage [48 49] Therefore the values of roughness lengthvary considerably in different geographical context or veg-etation types [50ndash54] Dryland ecosystems were sparsely
10 Advances in Meteorology
20
260
200
140
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
80
140
80
60
40
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
70
50
30
(a) (b)
200
150
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
40
80
120
8
6
4
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
4
2
(c) (d)
KZ-Ara
2
03
06
08
09
095
099
280
0
160
220
Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs20
60
100
150
120
90
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
60
90
80
60
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
50
70
ObsS0S2S3
S4S5S6S7
0
2
25
302
0405
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
3
25
15
ObsS0S2S3
S4S5S6S7
(a) (b)
(c) (d)
KZ-Bal
Figure 6 Taylor diagramplot of the simulated119877net (a)119876le (b)119876ℎ (c) and119879soil (d) computed from a referencemodel (S0) and six independentsensitivity tests (S2ndashS7 see Table 3) from the CoLM against EC observations at the two EC sites Standard deviation (STD Wmminus2) iscalculated as the simulated variables divided by the observed data ldquoObsrdquo refers to observed data points Root mean square error (RMSEWmminus2) is represented by green lines 119877 is the correlation coefficient The higher the 119877 and the smaller the STD and RMSE the better theagreement between model and data When comparing two simulations with different parameter values the longer the distance between thetwo simulation points the greater the sensitivity to that parameter
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
2 Advances in Meteorology
Table 1 Location soil vegetation climate factors and flux time series at KZ-Ara site and KZ-Bal site
Site Lon Lat Soil Dominant vegetation Temperature (∘C) Precipitation (mm) Simulation periodKZ-Ara 61∘051015840E 45∘581015840N Alkaline Shrub 83 1405 MayndashAug 2012KZ-Bal 76∘391015840E 44∘341015840N Alkaline Grass 57 1402 MayndashSep 2012
ground surface temperature are less dependent on vegeta-tion parameters and soil hydraulic parameters but moreon energy-related soil parameters and surface parameters(surface albedo the ground surface emissivity aerodynamicroughness length and thermal roughness length) [16] Unfor-tunately application of LSMs in arid and semiarid ecosystemswas largely limited For example recent studies have foundthat CoLM extremely underpredicted latent heat fluxes indryland areas [13 17] Prediction errors can be derived fromthe forcing data physics processes and the parameterizationsof land characteristic (which mainly included vegetation andsoil effects) [2 18] Vegetation effects consist of differentplant types that differ in leaf areas root profile heightoptical properties stomatal conductance and roughnesslength Soil properties can be soil texture (percent of sandand clay) and soil thermal and hydraulic parameters Theerrors resulting from parameterizations of these processeswill cause large errors for LSMs in modelling the turbulentfluxes
TheCentral Asia dryland area is characterized by a typicalarid continental climate complex terrain sparse vegetationand high heterogeneity in land surface which inevitablyproposes high challenges on the application of LSMs [19]Investigating the energy and water exchange between landsurface and atmosphere was of great importance to hydro-logical and ecological research [15 20] Few recent studieshave evaluated the performance of the CoLM to reproduceenergy and water vapor fluxes in Chinarsquos desert ecosystem[13 21]They have found that root distribution and root wateruptake function have significant effects on the performanceof CoLM in estimating energy and water fluxes However120572 1198850119898 and LAI are also expected to significantly impactthe performance of LSMs which largely influence energybudget thermodynamic resistance and turbulence fluxesThe errors resulting from these critical parameterizationswill cause large error and uncertainties for LSMs in theestimates of sensible and latent heat fluxes Unfortunately acomprehensive evaluation of LSMs and their uncertaintiescaused by 120572 1198850119898 and LAI was never investigated in CentralAsia dryland ecosystems
Therefore the main objectives of this research are (1) toevaluate the performance of CoLM of two desert ecosystemsin Kazakhstan using EC observations during the growingseasons for the purpose of investigating how well the CoLMis able to simulate the energy and water fluxes over CentralAsia land surface and (2) to analyze model sensitivities toalbedo roughness length and leaf area index
2 Materials and Methods
21 Site Description Two sets of EC equipment were estab-lished in representative dryland ecosystem in Kazakhstan
in 2012 (Figure 1) One site is close to Aral Sea and theother is close to Balkhash Lake The Aral Sea site (KZ-Ara 6108∘E 4596∘N) is located northeast of the Aral Seaand at the edge of Aralkum Desert Within radius of 5 kmof the KZ-Ara site the dominant plant species are weed-grass and reed vegetation in combination with xerophyticand halophytic desert semishrubs and shrubs The averagefraction of vegetation coverage was about 30The dominantsoil type was solonchak Historical climatic records with longterm observations showed that mean annual precipitation is140mm and mean annual air temperature is 83∘C (Table 1)The Balkhash Lake site (KZ-Bal 7663∘E 4457∘N) is locatedbetween the Balkhash Lake and the Kapchagay Reservoirand between original deserts and oasis croplands Withinradius of 5 km of the KZ-Bal site the dominant plantspecies are irrigated crops grasses and desert shrubs Theaverage fraction of vegetation coverage was about 40 Thesoil in the KZ-Bal site is saline solonchak [22] Historicalclimatic records with long term observations showed thatmean annual precipitation is 140mm and mean annual airtemperature is 57∘C (Table 1) For detailed information aboutthe two sites refer to Li et al [23]
22 Eddy Covariance and Ancillary Measurements At eachsite a set of EC systems was used to measure energy watervapor and CO2 fluxes between the land surface and theatmosphere The eddy covariance consists of an open pathgas analyzer (LI-7500 LICOR) and a three-dimensionalsonic anemometer thermometer (Wind Master Pro GillInstruments Lymington UK) The EC system also measuresair temperature and humidity (HMP45C Campbell) pre-cipitation (TE525MM Texas Electronics Dallas TX USA)short-wave and long-wave radiation (CNR-1 Kipp amp ZonenDelft Netherlands) soil temperature (TCAV Campbell)soil moisture content (CS616 Campbell Sci) and soil heatflux (HFP01 Hukseflux Delft Netherlands) The open pathgas analyzer and the three-dimensional sonic anemometerthermometer are mounted at 20m above the ground Soiltemperature soilmoisture content and soil heat flux (119866) weremeasured at 20 40 60 and 80 cm depth below the groundTherefore this study chose the interpolated soil temperaturein CoLMon the corresponding depth for comparisonsThesedata were logged every 30min by the data logger at 10Hz andall variables were carried out with data processing and gapfillingThe data was used for the analysis in this study duringgrowing seasons at the two sites and covered the period from30April to 18 August 2012 at the KZ-Ara site and from 23Mayto 6 September 2012 at the KZ-Bal site
23 Common Land Model CoLM [18] is one of the widelyused land surface models which was originally proposed toprovide a framework for a truly community developed land
Advances in Meteorology 3
KZ-BalKZ-Ara
Figure 1 Study area and location of the two eddy covariance (EC) sites in Kazakhstan
component of the National Centre for Atmospheric Research(NCAR) Community Climate System Model (CCSM) [2425] and it was subsequently developed by an open collabora-tion of many scientists A variety of comprehensive multiyearpoint observational data over different regions of the worldhave been used [26ndash28] These data were included in theproject for the Intercomparison of Land ParameterizationSchemes [29] and the results from these extensive tests wereuseful for CoLM improvement
CoLM simulates the exchange of energy mass andmomentum between the atmosphere and terrestrial bio-sphere through a wide range of ground and canopy biogeo-physical processes and allows for the simulation of explicitbiophysical mechanisms including the representation ofthermodynamic hydrologic and physiological processes [1830] In CoLM soil temperatures are predicted using a heatdiffusion equation in 10 soil layers and the soil thermal con-ductivity depends on soil water density Surface evapotranspi-ration consists of evaporation of precipitation intercepted byleaves canopy transpiration and bare soil evaporation Thesensible heat and water vapor fluxes between the surface andthe reference height can be written in the following form
119876ℎ = minus120588119886119888119901
(120579119886 minus 120579119904)
119903119886ℎ
119864 = minus120588119886
(119902119886 minus 119902119904)
119903119886119908
(1)
where 119876ℎ is sensible heat flux and 119864 is water vaporflux 120588119886 is air density and 119888119901 is specific heat for dry air(100467 J kgminus1 Kminus1) 120579119886 and 120579119904 are air potential temperatureat reference height and surface potential temperature respec-tively 119902119886 and 119902119904 are water vapor specific humidity at referenceheight and surface air humidity respectively 119903119886ℎ and 119903119886119908
are the aerodynamic resistances for heat and water vaporrespectively which are crucial parameters in determining119876ℎ and 119864 calculated by Monin-Obukhov similarity theory[31] and mainly dependent on the thermal roughness length(1198850ℎ) and 1198850119898 [16] as well as the meteorological forcingvariables in CoLM such as air temperature wind speed andprecipitation
24 Model Simulations and Sensitivity Analysis The versionof CoLM which integrated an optimal root water uptakefunction for dryland ecosystem [21] was used (termed S0)in this study To investigate the effect of roughness lengthon the performance of CoLM a simulation driven by anempirical estimate of roughness length (S1) was conducted(Table 2) In S0 simulation momentum roughness lengththermal roughness length and the water vapor roughnesslength (1198850119908) were all defined the same in CoLM for baresoil (1198850119898 = 1198850ℎ = 1198850119908 = 005) However some researchin dryland ecosystems indicated that the abovementionedroughness length values have different magnitude [4 32ndash34]As in the previous studies Chen et al [16] found that the
4 Advances in Meteorology
Table 2 The parameterizations of the reference model (S0) and the model with refined roughness length (S1) in CLM at KZ-Ara site andKZ-Bal site RWUF is an optimal root water uptake function for dryland ecosystem (Jing et al 2014 [21]) and119882119888 119882119909 and 119898 are empiricalconstants in RWUF
Site Simulation Land cover type Vegetation fraction RWUF Roughness length for bare soil(119882119888119882119909119898) 1198850119898 1198850ℎ 1198850119908
KZ-Ara S0 Mixed shrub and grassland 30 (08054) 5119864 minus 2 5119864 minus 2 5119864 minus 2
S1 S0 S0 S0 2119864 minus 3 2119864 minus 4 2119864 minus 4
KZ-Bal S0(1) Irrigated cropland and pasture (35)(2) Herbaceous wetland (44)(3) Shrub (21)
40 (03074) 5119864 minus 2 5119864 minus 2 5119864 minus 2
S1 S0 S0 S0 2119864 minus 3 2119864 minus 4 2119864 minus 4
Table 3 A baseline of reference exprement (S0) and six different configurations of CLMat each site (S2ndashS7) as used in this study for sensitivitytesting Two sets of leaf area index (LAI) average roughness length (1198850119898) and average albedo (120572) were used in CLM ldquo119863rdquo refers to the modeldefault parameter setting
KZ-Ara KZ-BalSimulation LAI 1198850119898 120572 Simulation LAI 1198850119898 120572
S0 119863 119863(005) 119863 S0 119863 119863(01) 119863
S2 119863 lowast 07 119863 119863 S2 119863 lowast 07 119863 119863
S3 119863 lowast 13 119863 119863 S3 119863 lowast 13 119863 119863
S4 119863 0005 119863 S4 119863 001 119863
S5 119863 05 119863 S5 119863 1 119863
S6 119863 119863 119863 lowast 07 S6 119863 119863 119863 lowast 07
S7 119863 119863 119863 lowast 13 S7 119863 119863 119863 lowast 13
surface flux and temperature have different sensitivities to1198850ℎand 1198850119898 and the sensible heat flux is very sensitive to theparameterization schemes of 1198850ℎ in arid regions [16]
A common method to calculate 1198850119898 and 1198850ℎ is
1198850119898 = 119911119890(minus119896119880119906
lowast)minus120593119898(119911119871)
1198850ℎ = 119911119890(minus119896(119879minus119879
119904)119879119904)minus120593ℎ(119911119871)
(2)
where 119911119871 is a stability parameter (119871 is the Monin-Obukhovlength and 119911 is the observational height) and 120593119898(119911119871) isthe stability function of wind profile and becomes 0 at theneutral condition 119880 is the average wind speed and 119906lowast isthe surface friction velocity 120593ℎ(119911119871) is the stability functionof the temperature profile and becomes 0 at the neutralcondition 119896 is vonKarman constant and equals 04 generally119879 is the air temperature and119879119904 is the surface temperatureTherelationship between 1198850119898 and 1198850ℎ can be described as
ln(1198850119898
1198850ℎ
) = 119896119887minus1 (3)
where 119896119887minus1 can be obtained from the bulk transfer equationas
119896119887minus1=119896119906lowast (119879119904 minus 119879)
119867120588119888119901
minus [ln119911 minus 1198890
1198850119898
minus 120593ℎ (119911
119871)] (4)
where119867 is the observed sensible heat flux 120588 is the air density119888119901 is the specific heat for dry air and 1198890 is the zero plane
displacement Thus the relationship among 1198850119898 1198850ℎ and1198850119908 is described as
1198850ℎ = 1198850119908 = 1198850119898119890minus119896119887minus1
(5)
where the excess resistance to heat transfer 119896119887minus1 is importantto the sensible heat exchange between land surface andatmosphere and there are linear correlations between 119896119887
minus1
and surface temperature1198850119898 is physically related to the geometric roughness of
surface elements and can be derived from the wind speedand temperature profiles Bao et al [4] and Yang et al [34]argued that this scheme overestimated 1198850ℎ and would mis-estimate the energy and water fluxes Momentum transportis more efficient than heat transport due to the influence ofpressure fluctuation because individual roughness elementsmay enhance the momentum flux through form drag withlittle contribution to the area-averaged heat flux [35]
An appropriatemethod suggested by Zhang et al [36] wasemployed to estimate1198850119898 at desert ecosystems In desert1198850119898was estimated as 00019 plusmn 00071m and 1198850ℎ and 1198850119908 were atsame order but almost onemagnitude lower than1198850119898 [36] Inthis study1198850119898 was set as 0002m and1198850ℎ = 1198850119908 = 00002mfor bare soil at the two sites (Table 2)
This research further investigated the sensitivities of theCoLM to LAI 1198850119898 and 120572 Therefore other six simulationsby increasing or decreasing the value of each parameter wereconducted The specifications of all simulations are listedin Table 3 In order to avoid the cross-influence of eachparameter all sensitivity tests took the simulation S0 as areference
Advances in Meteorology 5
0 200100 300 500400 600 700minus100
minus100
0
100
200
300
400
500
600
700
Qle+Qh
(W m
minus2)
y = 076 lowast x + 2895
R2= 091
RMSE = 4645
Rnet minus G (W mminus2)
(a) KZ-Ara
minus100 5004003002001000 600 700
minus100
0
100
200
300
400
500
600
700
y = 095 lowast x + 156
R2= 097
+Qh
(W m
minus2)
Rnet minus G (W mminus2)
3519RMSE =
Qle
(b) KZ-Bal
Figure 2 Energy balance closure at the KZ-Ara site and the KZ-Bal site The slope of the fitted line represents energy closure ratio and 1198772 isthe coefficient of determination RMSE (Wmminus2) is the root mean square error The energy fluxes include sensible heat flux (119876ℎ) latent heatflux (119876le) net radiation (119877net) and ground heat flux (119866)
25 Statistical Analysis Energy balance ratio (EBR) [35] wasused to give an overall evaluation of energy balance closure byaveraging over random errors in the half-hourmeasurementsat two flux tower sites and it was calculated by
EBR =sum119899
119894=1(119876le + 119876ℎ)
sum119899
119894=1(119877net minus 119866)
(6)
where 119899 is the number of half hours of data The values ofEBR close to 1 indicate the best degree of energy balanceclosure Additionally coefficient of determination (1198772) rootmean square error (RMSE) slope (119887119904) and intercept (1198870) areused to justify the performance of the model
The Taylor diagram [37] was used to quantify the degreeof the sensitivities of the model to management parametersRMSE 1198772 and standard error (STD) are used in the TaylordiagramThe output of the model simulation is specified by asingle point with the STDbeing the polar axis and119877 the polarangle The ldquoreferencerdquo point represents observations and theother points refer to themodel results from the simulations ofsensitivity testsThe distances from the reference point to theother points representing the consequence of the relationshipindicate the RMSE The higher 119877 and the smaller the STDandRMSE the better the agreement betweenmodel and dataWhen comparing two simulations with different parametervalues the longer the distance between the two simulationpoints the greater the sensitivity to that parameter
3 Results
31 Energy Balance Closure The slopes of the linear regres-sion between the observed 119876le + 119876ℎ and 119877net minus 119866 were 076and 095 at KZ-Ara and KZ-Bal respectively The coefficient
of determination (1198772) of the observed 119876le + 119876ℎ and 119877net minus 119866was 091 and 097 and the root mean square error (RMSE)was 4645 and 3519Wmminus2 respectively (Figure 2) Energybalance ratio (EBR) at KZ-Ara and KZ-Bal was 111 and 106respectively
32 Modelled 119877119899119890119905 119876119897119890 119876ℎ and 119879119904119900119894119897 Figure 3 shows thecomparisons between the measurements and the simulationsof the reference model (S0) and the model with refinedroughness length (S1) for 119877net 119876le and 119876ℎ at the twoKazakhstan sites The reference model (S0) significantlyunderestimated the latent heat flux and overestimated thesensible heat flux at both sites However the performanceof the refined roughness length (S1) was largely improvedin simulating turbulent heat fluxes The latent heat flux wasincreased and sensible heat flux was decreased obviouslyAt the KZ-Ara site RMSE for 119877net decreased from 636 to395Wmminus2 1198772 values for 119876le given by the two simulations(S0 and S1) were 036 and 061 respectively and RMSEdecreased from 4115 in S0 to 333Wmminus2 in S1 (Table 4)For 119876ℎ simulations the RMSE for S0 was 12124Wmminus2as compared to 5847Wmminus2 for S1 (Table 4) The resultsindicated that the simulation with refined roughness length(S1) significantly improved the performance of the model forboth 119877net and 119876le and particularly for 119876ℎ At the KZ-Bal sitethe performance of the simulation with refined roughness(S1) was also greatly improved 1198772 values for 119876le given by S0and S1 were 09 and 092 respectively and RMSE decreasedfrom 4642 in S0 to 4311Wmminus2 in S1 1198772 values for 119876ℎgiven by the two simulations were 07 and 067 respectivelyand RMSE decreased from 7738 in S0 to 3797Wmminus2 in S1(Table 4 Figure 3)
6 Advances in Meteorology
Table 4 Model performance for simulating 119877net 119876le and 119876ℎ indicated by coefficient of determination (1198772) slope (119887119904) intercept (119887
0) and
root mean square error (RMSE Wmminus2) of linear regressions between model and observed data at the KZ-Ara site and the KZ-Bal site
Variables Reference model (S0) Refined roughness length (S1)1198772 RMSE 119887119904 1198870 119877
2 RMSE 119887119904 1198870
Site KZ-Ara119877net 097 636 112 2848 096 395 099 712119876le 036 4115 074 966 061 333 102 467119876ℎ 089 12124 186 566 086 5847 106 4693
Site KZ-Bal119877net 098 4386 104 227 098 3307 096 139119876le 09 4642 082 3057 092 4311 089 2867119876ℎ 07 7738 158 5087 067 3797 076 2908
To further investigate the effects of refined roughnesson the energy fluxes simulations Figure 4 shows the meandiurnal turbulent fluxes during growing seasons at the twosites Diurnal variations of the three components of energyfluxes showed typical characteristics at the KZ-Ara site thatis 119877net gt 119876ℎ gt 119876le (Figure 4(a)) The KZ-Bal site is locatedbetween oasis croplands and original deserts Although theamount of the average annual precipitation is similar to KZ-Ara site the characteristic of the energy fluxes allocationshowed higher latent heat flux and lower sensible heat fluxcompared to theKZ-Ara site (Figure 4(b)) At theKZ-Ara siteS0 overestimated 119877net with the peak value of 520Wmminus2 ascompared to the observed peak value 450Wmminus2 Howeverthe simulation with refined roughness (S1) produced a goodagreement for 119877net between the simulation and the measure-ments In addition S0 overestimated 119876ℎ with the peak valueof 400Wmminus2 at noontime as compared to 200Wmminus2 for themeasurements S1 improved the simulation for 119876ℎ as well AtKZ-Bal site both S0 and S1 agreed better with observationsfor 119877net and 119876le S0 significantly overestimated 119876ℎ at thedaytimeHowever the simulationwith refined roughness (S1)produced better agreement with the observations
Figure 5 showed the comparisons of soil temperaturebetween the observation and the simulations of the referencemodel (S0) and the model with refined roughness length (S1)at each site Although there are significant challenges when itcomes to validating soil temperature fromLSMbecause of thehigh sensitivities of simulated andmeasured soil temperatureto soil texture moisture conditions and the limitations ofthe measurement the results from the reference model (S0)and the model with refined roughness length (S1) appearedto be realistic and generally captured the seasonal variationfor soil temperature It was found that the modelled 119879soil inMay was underestimated and significantly improved in Juneand July but overestimated in August However S1 producedhigher values of soil temperature than S0 Additionally theCoLM simulated a smaller vertical soil temperature gradientas shown that the contour for the simulationswasmore sparsecompared with the observation
33 Sensitivity of CoLM to LAI 1198850119898 and 120572 In orderto investigate the sensitivities of CoLM the values of theparameters were adjusted in large ranges Key parameters
consist of LAI 1198850119898 and 120572 with regard to the simulationof 119877net 119876le 119876ℎ and 119879soil at the two EC sites A referencemodel (S0) and six independent sensitivity tests (S2ndashS7 seeTable 3) were conducted at each siteThis research only chosethe modelled soil temperatures at the depth of 20 cm forsensitivity tests in this study
The modelled 119876le and 119876ℎ in CoLM were divided intotwo parts the fluxes on vegetation leaves and the fluxes onthe ground Thus 119876le showed a more complicated sensitivityto LAI Since the leaf temperature increased and groundtemperature decreased with the increase in LAI the evap-otranspiration from the leaves improved and the groundevaporation weakened but the total latent heat flux wasincreased and exhibited a strong sensitivity for 119876le at KZ-Ara When it came to KZ-Bal the vegetation coverage waslarger than KZ-Ara while continuing to increase LAI hasslightly further improved the performance of 119876le Similarly119876ℎ and 119879soil showed a strong sensitivity to LAI The more thevegetation the more the solar radiation intercepted When itcame to1198850119898 the values of 119877net and119876ℎ were increased but119876leand 119879soil were decreased with the increase in 1198850119898 Figure 6indicated that 119876le 119876ℎ and 119879soil were highly sensitive to 1198850119898Taking the KZ-Ara site as an example 1198772 values for themodelled 119876le improved from 04 to 07 when 1198850119898 decreasedfrom 05 (S4) to 0005 (S5) Figure 6 also showed a strongsensitivity of albedo to 119877net 119876ℎ and 119879soil The imprecisesettings of albedomay enlarge the errors for sensible heat fluxin CoLMGround sensible heat flux was negatively correlatedto albedo The increase in the surface albedo decreasedthe solar radiation absorbed by soil and soil temperaturewas decreased However 119876le was slightly sensitive to albedo(Figure 6)
These sensitivity analysis results demonstrated that theimprovement in model performance observed in S2ndashS7 wassignificantly affected by the values of these three parametersin CoLM further justifying the significance of these keyparameters (LAI 1198850119898 and albedo) to the Common LandModel
4 Discussion
Arid and semiarid areas cover approximately one-third ofthe global terrestrial land surfaces [15] Central Asia has
Advances in Meteorology 7
KZ-AraSi
mul
atio
ns (W
mminus2)
S0
minus200
0
200
400
600
800
400 600 8000 200minus200
Observations (W mminus2)
(a) 119877net
KZ-AraS1
minus200
0
200
400
600
800
200 400 600 800minus200 0
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(b) 119877net
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400minus100
minus100
0
100
200
300
400
KZ-AraS0
Observations (W mminus2)
(c) 119876le
100
minus100
0
100
200
300
400
200 300 400minus100 0
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(d) 119876le
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
minus100
0
100
200
300
400
500
KZ-AraS0
Observations (W mminus2)
(e) 119876ℎ
minus100
0
100
200
300
400
500
minus100 0 100 200 300 400 500
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(f) 119876ℎKZ-Bal
Sim
ulat
ions
(W m
minus2)
S0
0 200 400 600 800minus200
minus200
0
200
400
600
800
Observations (W mminus2)
(g) 119877net
KZ-BalS1
minus200
0
200
400
600
800
minus200 0 200 400 600 800
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(h) 119877net
Figure 3 Continued
8 Advances in Meteorology
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
0
200
400
Observations (W mminus2)
KZ-BalS0
(i) 119876le
KZ-BalS1
0
200
400
minus100 0 100 200 300 400 500
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(j) 119876le
Sim
ulat
ions
(W m
minus2)
minus100
0
100
200
300
0 100 300minus100
Observations (W mminus2)
KZ-BalS0
200
(k) 119876ℎ
minus100
0
100
200
300
0 100 200 300minus100
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
KZ-BalS1
(l) 119876ℎ
Figure 3 Comparison between the measured half-hourly net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the valuessimulated by the reference model (S0) and the model with refined roughness length (S1) at the KZ-Ara and KZ-Bal sites The solid red linerepresents the linear regression between the simulation and the observed data and the dashed line represents a 1 1 relationship between thedatasets
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(a) KZ-Ara
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(b) KZ-Bal
Figure 4 Comparison between the measured net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the values simulatedby the reference model (S0) and themodel with refined roughness length (S1) on a diurnal course at the KZ-Ara and KZ-Bal sitesThe diurnalflux values were calculated as the mean values of all data at same measurement time in a day for the entire time period
Advances in Meteorology 9
20
30
40
50
60
70
80
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
Soil
dept
h (c
m)
DOY in 2012
10
20
30
40
(a)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
40555555555555555555555555555555555 1
(b)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
405555555555555555555 1
(c)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(d)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(e)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(f)
Figure 5 Panels (a) and (b) are the isograms changed with time and depth variation for the reference model (S0) and the model with refinedroughness length (S1) and (c) shows the corresponding observed data at the depth of 20ndash80 cm below ground at the KZ-Ara site panels (d)and (e) are the isograms for S0 and S1 and (f) shows the observations at the KZ-Bal site
vast proportion of dryland ecosystems where climate wasfeatured as hot and dry during growing season [19] and thevegetation is sparseThe characteristic of dryland ecosystemsis significantly different from those in humid areas in termsof surface energy budget Many researchers have focused onecosystem functioning and structure in response to climatechange Kariyeva et al [38] examined spatiotemporal changepatterns and interactions between terrestrial phenology andclimate change in Central Asia during the period 1981ndash2008Lioubimtseva et al [19] have found that local and regionalhuman impacts in arid zones can significantly modify landsurface albedo as well as water exchange and nutrient cyclesthat could have essential impacts on the climate at both theregional and global scales Such kind of analyses advancedour understanding on the response of dryland ecosystembehaviour and functioning to climate change Howeverapplying LSM into dryland ecosystems was far more limitedRecently the CoLM has been validated at one desert shrubecosystem in Xinjiang China and the results found that rootfraction and root water uptake have important effects on theperformance of CoLM in simulating energy and water fluxes[13 21] In this study a refined parameterization of CoLMhasbeen evaluated at two newly built EC sites in Kazakhstan
The most commonly used technique to obtain landsurface turbulent fluxes is measurement of eddy covariancesystem and the analyzer was based on flux footprint modelsThe footprint concept is the probability that a scalar comingfrom a given elemental source reaches the measurementpoint Footprint models describe the relationship between
the spatial distribution of surface sources and the measuredsignal using footprint functions Several flux footprintmodelshave been designed [39ndash42] But most of them cannotaccount for inhomogeneous turbulence or require largercomputational resources Gockede et al [43] improved anEulerian footprint model use of satellite maps for explicitassignment of surface type Gockede et al [44] and Rebmannet al [45] applied this newmodel at the EC sites and obtainedsatisfactory results At present footprint models are used toestimate the source areas contributing to the flux observa-tions In addition they provide a tool for quality control ofthe flux measurements and provide guidance in designingexperiments [46] Thus the footprint models have consider-able potential in microclimatology investigations especiallyin studies which include nonhomogeneous surfaces
The momentum roughness length (1198850119898) thermal rough-ness length (1198850ℎ) and the water vapor roughness length(1198850119908) are crucial parameters for calculating momentum andheat fluxes in bulk transfer equations which is one of theessential components in LSMs It has been widely observedthat 1198850119898 differs from 1198850ℎ and 1198850119908 [4 47] Unfortunately1198850119898 1198850ℎ and 1198850119908 up to date are still treated as constants inmost LSMs Inaccurate estimates of roughness length wouldenlarge the bias of simulated energy andwater fluxes in LSMsMany researchers have found that roughness length stronglydepended on surface heterogeneity vegetation height andcoverage [48 49] Therefore the values of roughness lengthvary considerably in different geographical context or veg-etation types [50ndash54] Dryland ecosystems were sparsely
10 Advances in Meteorology
20
260
200
140
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
80
140
80
60
40
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
70
50
30
(a) (b)
200
150
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
40
80
120
8
6
4
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
4
2
(c) (d)
KZ-Ara
2
03
06
08
09
095
099
280
0
160
220
Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs20
60
100
150
120
90
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
60
90
80
60
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
50
70
ObsS0S2S3
S4S5S6S7
0
2
25
302
0405
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
3
25
15
ObsS0S2S3
S4S5S6S7
(a) (b)
(c) (d)
KZ-Bal
Figure 6 Taylor diagramplot of the simulated119877net (a)119876le (b)119876ℎ (c) and119879soil (d) computed from a referencemodel (S0) and six independentsensitivity tests (S2ndashS7 see Table 3) from the CoLM against EC observations at the two EC sites Standard deviation (STD Wmminus2) iscalculated as the simulated variables divided by the observed data ldquoObsrdquo refers to observed data points Root mean square error (RMSEWmminus2) is represented by green lines 119877 is the correlation coefficient The higher the 119877 and the smaller the STD and RMSE the better theagreement between model and data When comparing two simulations with different parameter values the longer the distance between thetwo simulation points the greater the sensitivity to that parameter
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
Advances in Meteorology 3
KZ-BalKZ-Ara
Figure 1 Study area and location of the two eddy covariance (EC) sites in Kazakhstan
component of the National Centre for Atmospheric Research(NCAR) Community Climate System Model (CCSM) [2425] and it was subsequently developed by an open collabora-tion of many scientists A variety of comprehensive multiyearpoint observational data over different regions of the worldhave been used [26ndash28] These data were included in theproject for the Intercomparison of Land ParameterizationSchemes [29] and the results from these extensive tests wereuseful for CoLM improvement
CoLM simulates the exchange of energy mass andmomentum between the atmosphere and terrestrial bio-sphere through a wide range of ground and canopy biogeo-physical processes and allows for the simulation of explicitbiophysical mechanisms including the representation ofthermodynamic hydrologic and physiological processes [1830] In CoLM soil temperatures are predicted using a heatdiffusion equation in 10 soil layers and the soil thermal con-ductivity depends on soil water density Surface evapotranspi-ration consists of evaporation of precipitation intercepted byleaves canopy transpiration and bare soil evaporation Thesensible heat and water vapor fluxes between the surface andthe reference height can be written in the following form
119876ℎ = minus120588119886119888119901
(120579119886 minus 120579119904)
119903119886ℎ
119864 = minus120588119886
(119902119886 minus 119902119904)
119903119886119908
(1)
where 119876ℎ is sensible heat flux and 119864 is water vaporflux 120588119886 is air density and 119888119901 is specific heat for dry air(100467 J kgminus1 Kminus1) 120579119886 and 120579119904 are air potential temperatureat reference height and surface potential temperature respec-tively 119902119886 and 119902119904 are water vapor specific humidity at referenceheight and surface air humidity respectively 119903119886ℎ and 119903119886119908
are the aerodynamic resistances for heat and water vaporrespectively which are crucial parameters in determining119876ℎ and 119864 calculated by Monin-Obukhov similarity theory[31] and mainly dependent on the thermal roughness length(1198850ℎ) and 1198850119898 [16] as well as the meteorological forcingvariables in CoLM such as air temperature wind speed andprecipitation
24 Model Simulations and Sensitivity Analysis The versionof CoLM which integrated an optimal root water uptakefunction for dryland ecosystem [21] was used (termed S0)in this study To investigate the effect of roughness lengthon the performance of CoLM a simulation driven by anempirical estimate of roughness length (S1) was conducted(Table 2) In S0 simulation momentum roughness lengththermal roughness length and the water vapor roughnesslength (1198850119908) were all defined the same in CoLM for baresoil (1198850119898 = 1198850ℎ = 1198850119908 = 005) However some researchin dryland ecosystems indicated that the abovementionedroughness length values have different magnitude [4 32ndash34]As in the previous studies Chen et al [16] found that the
4 Advances in Meteorology
Table 2 The parameterizations of the reference model (S0) and the model with refined roughness length (S1) in CLM at KZ-Ara site andKZ-Bal site RWUF is an optimal root water uptake function for dryland ecosystem (Jing et al 2014 [21]) and119882119888 119882119909 and 119898 are empiricalconstants in RWUF
Site Simulation Land cover type Vegetation fraction RWUF Roughness length for bare soil(119882119888119882119909119898) 1198850119898 1198850ℎ 1198850119908
KZ-Ara S0 Mixed shrub and grassland 30 (08054) 5119864 minus 2 5119864 minus 2 5119864 minus 2
S1 S0 S0 S0 2119864 minus 3 2119864 minus 4 2119864 minus 4
KZ-Bal S0(1) Irrigated cropland and pasture (35)(2) Herbaceous wetland (44)(3) Shrub (21)
40 (03074) 5119864 minus 2 5119864 minus 2 5119864 minus 2
S1 S0 S0 S0 2119864 minus 3 2119864 minus 4 2119864 minus 4
Table 3 A baseline of reference exprement (S0) and six different configurations of CLMat each site (S2ndashS7) as used in this study for sensitivitytesting Two sets of leaf area index (LAI) average roughness length (1198850119898) and average albedo (120572) were used in CLM ldquo119863rdquo refers to the modeldefault parameter setting
KZ-Ara KZ-BalSimulation LAI 1198850119898 120572 Simulation LAI 1198850119898 120572
S0 119863 119863(005) 119863 S0 119863 119863(01) 119863
S2 119863 lowast 07 119863 119863 S2 119863 lowast 07 119863 119863
S3 119863 lowast 13 119863 119863 S3 119863 lowast 13 119863 119863
S4 119863 0005 119863 S4 119863 001 119863
S5 119863 05 119863 S5 119863 1 119863
S6 119863 119863 119863 lowast 07 S6 119863 119863 119863 lowast 07
S7 119863 119863 119863 lowast 13 S7 119863 119863 119863 lowast 13
surface flux and temperature have different sensitivities to1198850ℎand 1198850119898 and the sensible heat flux is very sensitive to theparameterization schemes of 1198850ℎ in arid regions [16]
A common method to calculate 1198850119898 and 1198850ℎ is
1198850119898 = 119911119890(minus119896119880119906
lowast)minus120593119898(119911119871)
1198850ℎ = 119911119890(minus119896(119879minus119879
119904)119879119904)minus120593ℎ(119911119871)
(2)
where 119911119871 is a stability parameter (119871 is the Monin-Obukhovlength and 119911 is the observational height) and 120593119898(119911119871) isthe stability function of wind profile and becomes 0 at theneutral condition 119880 is the average wind speed and 119906lowast isthe surface friction velocity 120593ℎ(119911119871) is the stability functionof the temperature profile and becomes 0 at the neutralcondition 119896 is vonKarman constant and equals 04 generally119879 is the air temperature and119879119904 is the surface temperatureTherelationship between 1198850119898 and 1198850ℎ can be described as
ln(1198850119898
1198850ℎ
) = 119896119887minus1 (3)
where 119896119887minus1 can be obtained from the bulk transfer equationas
119896119887minus1=119896119906lowast (119879119904 minus 119879)
119867120588119888119901
minus [ln119911 minus 1198890
1198850119898
minus 120593ℎ (119911
119871)] (4)
where119867 is the observed sensible heat flux 120588 is the air density119888119901 is the specific heat for dry air and 1198890 is the zero plane
displacement Thus the relationship among 1198850119898 1198850ℎ and1198850119908 is described as
1198850ℎ = 1198850119908 = 1198850119898119890minus119896119887minus1
(5)
where the excess resistance to heat transfer 119896119887minus1 is importantto the sensible heat exchange between land surface andatmosphere and there are linear correlations between 119896119887
minus1
and surface temperature1198850119898 is physically related to the geometric roughness of
surface elements and can be derived from the wind speedand temperature profiles Bao et al [4] and Yang et al [34]argued that this scheme overestimated 1198850ℎ and would mis-estimate the energy and water fluxes Momentum transportis more efficient than heat transport due to the influence ofpressure fluctuation because individual roughness elementsmay enhance the momentum flux through form drag withlittle contribution to the area-averaged heat flux [35]
An appropriatemethod suggested by Zhang et al [36] wasemployed to estimate1198850119898 at desert ecosystems In desert1198850119898was estimated as 00019 plusmn 00071m and 1198850ℎ and 1198850119908 were atsame order but almost onemagnitude lower than1198850119898 [36] Inthis study1198850119898 was set as 0002m and1198850ℎ = 1198850119908 = 00002mfor bare soil at the two sites (Table 2)
This research further investigated the sensitivities of theCoLM to LAI 1198850119898 and 120572 Therefore other six simulationsby increasing or decreasing the value of each parameter wereconducted The specifications of all simulations are listedin Table 3 In order to avoid the cross-influence of eachparameter all sensitivity tests took the simulation S0 as areference
Advances in Meteorology 5
0 200100 300 500400 600 700minus100
minus100
0
100
200
300
400
500
600
700
Qle+Qh
(W m
minus2)
y = 076 lowast x + 2895
R2= 091
RMSE = 4645
Rnet minus G (W mminus2)
(a) KZ-Ara
minus100 5004003002001000 600 700
minus100
0
100
200
300
400
500
600
700
y = 095 lowast x + 156
R2= 097
+Qh
(W m
minus2)
Rnet minus G (W mminus2)
3519RMSE =
Qle
(b) KZ-Bal
Figure 2 Energy balance closure at the KZ-Ara site and the KZ-Bal site The slope of the fitted line represents energy closure ratio and 1198772 isthe coefficient of determination RMSE (Wmminus2) is the root mean square error The energy fluxes include sensible heat flux (119876ℎ) latent heatflux (119876le) net radiation (119877net) and ground heat flux (119866)
25 Statistical Analysis Energy balance ratio (EBR) [35] wasused to give an overall evaluation of energy balance closure byaveraging over random errors in the half-hourmeasurementsat two flux tower sites and it was calculated by
EBR =sum119899
119894=1(119876le + 119876ℎ)
sum119899
119894=1(119877net minus 119866)
(6)
where 119899 is the number of half hours of data The values ofEBR close to 1 indicate the best degree of energy balanceclosure Additionally coefficient of determination (1198772) rootmean square error (RMSE) slope (119887119904) and intercept (1198870) areused to justify the performance of the model
The Taylor diagram [37] was used to quantify the degreeof the sensitivities of the model to management parametersRMSE 1198772 and standard error (STD) are used in the TaylordiagramThe output of the model simulation is specified by asingle point with the STDbeing the polar axis and119877 the polarangle The ldquoreferencerdquo point represents observations and theother points refer to themodel results from the simulations ofsensitivity testsThe distances from the reference point to theother points representing the consequence of the relationshipindicate the RMSE The higher 119877 and the smaller the STDandRMSE the better the agreement betweenmodel and dataWhen comparing two simulations with different parametervalues the longer the distance between the two simulationpoints the greater the sensitivity to that parameter
3 Results
31 Energy Balance Closure The slopes of the linear regres-sion between the observed 119876le + 119876ℎ and 119877net minus 119866 were 076and 095 at KZ-Ara and KZ-Bal respectively The coefficient
of determination (1198772) of the observed 119876le + 119876ℎ and 119877net minus 119866was 091 and 097 and the root mean square error (RMSE)was 4645 and 3519Wmminus2 respectively (Figure 2) Energybalance ratio (EBR) at KZ-Ara and KZ-Bal was 111 and 106respectively
32 Modelled 119877119899119890119905 119876119897119890 119876ℎ and 119879119904119900119894119897 Figure 3 shows thecomparisons between the measurements and the simulationsof the reference model (S0) and the model with refinedroughness length (S1) for 119877net 119876le and 119876ℎ at the twoKazakhstan sites The reference model (S0) significantlyunderestimated the latent heat flux and overestimated thesensible heat flux at both sites However the performanceof the refined roughness length (S1) was largely improvedin simulating turbulent heat fluxes The latent heat flux wasincreased and sensible heat flux was decreased obviouslyAt the KZ-Ara site RMSE for 119877net decreased from 636 to395Wmminus2 1198772 values for 119876le given by the two simulations(S0 and S1) were 036 and 061 respectively and RMSEdecreased from 4115 in S0 to 333Wmminus2 in S1 (Table 4)For 119876ℎ simulations the RMSE for S0 was 12124Wmminus2as compared to 5847Wmminus2 for S1 (Table 4) The resultsindicated that the simulation with refined roughness length(S1) significantly improved the performance of the model forboth 119877net and 119876le and particularly for 119876ℎ At the KZ-Bal sitethe performance of the simulation with refined roughness(S1) was also greatly improved 1198772 values for 119876le given by S0and S1 were 09 and 092 respectively and RMSE decreasedfrom 4642 in S0 to 4311Wmminus2 in S1 1198772 values for 119876ℎgiven by the two simulations were 07 and 067 respectivelyand RMSE decreased from 7738 in S0 to 3797Wmminus2 in S1(Table 4 Figure 3)
6 Advances in Meteorology
Table 4 Model performance for simulating 119877net 119876le and 119876ℎ indicated by coefficient of determination (1198772) slope (119887119904) intercept (119887
0) and
root mean square error (RMSE Wmminus2) of linear regressions between model and observed data at the KZ-Ara site and the KZ-Bal site
Variables Reference model (S0) Refined roughness length (S1)1198772 RMSE 119887119904 1198870 119877
2 RMSE 119887119904 1198870
Site KZ-Ara119877net 097 636 112 2848 096 395 099 712119876le 036 4115 074 966 061 333 102 467119876ℎ 089 12124 186 566 086 5847 106 4693
Site KZ-Bal119877net 098 4386 104 227 098 3307 096 139119876le 09 4642 082 3057 092 4311 089 2867119876ℎ 07 7738 158 5087 067 3797 076 2908
To further investigate the effects of refined roughnesson the energy fluxes simulations Figure 4 shows the meandiurnal turbulent fluxes during growing seasons at the twosites Diurnal variations of the three components of energyfluxes showed typical characteristics at the KZ-Ara site thatis 119877net gt 119876ℎ gt 119876le (Figure 4(a)) The KZ-Bal site is locatedbetween oasis croplands and original deserts Although theamount of the average annual precipitation is similar to KZ-Ara site the characteristic of the energy fluxes allocationshowed higher latent heat flux and lower sensible heat fluxcompared to theKZ-Ara site (Figure 4(b)) At theKZ-Ara siteS0 overestimated 119877net with the peak value of 520Wmminus2 ascompared to the observed peak value 450Wmminus2 Howeverthe simulation with refined roughness (S1) produced a goodagreement for 119877net between the simulation and the measure-ments In addition S0 overestimated 119876ℎ with the peak valueof 400Wmminus2 at noontime as compared to 200Wmminus2 for themeasurements S1 improved the simulation for 119876ℎ as well AtKZ-Bal site both S0 and S1 agreed better with observationsfor 119877net and 119876le S0 significantly overestimated 119876ℎ at thedaytimeHowever the simulationwith refined roughness (S1)produced better agreement with the observations
Figure 5 showed the comparisons of soil temperaturebetween the observation and the simulations of the referencemodel (S0) and the model with refined roughness length (S1)at each site Although there are significant challenges when itcomes to validating soil temperature fromLSMbecause of thehigh sensitivities of simulated andmeasured soil temperatureto soil texture moisture conditions and the limitations ofthe measurement the results from the reference model (S0)and the model with refined roughness length (S1) appearedto be realistic and generally captured the seasonal variationfor soil temperature It was found that the modelled 119879soil inMay was underestimated and significantly improved in Juneand July but overestimated in August However S1 producedhigher values of soil temperature than S0 Additionally theCoLM simulated a smaller vertical soil temperature gradientas shown that the contour for the simulationswasmore sparsecompared with the observation
33 Sensitivity of CoLM to LAI 1198850119898 and 120572 In orderto investigate the sensitivities of CoLM the values of theparameters were adjusted in large ranges Key parameters
consist of LAI 1198850119898 and 120572 with regard to the simulationof 119877net 119876le 119876ℎ and 119879soil at the two EC sites A referencemodel (S0) and six independent sensitivity tests (S2ndashS7 seeTable 3) were conducted at each siteThis research only chosethe modelled soil temperatures at the depth of 20 cm forsensitivity tests in this study
The modelled 119876le and 119876ℎ in CoLM were divided intotwo parts the fluxes on vegetation leaves and the fluxes onthe ground Thus 119876le showed a more complicated sensitivityto LAI Since the leaf temperature increased and groundtemperature decreased with the increase in LAI the evap-otranspiration from the leaves improved and the groundevaporation weakened but the total latent heat flux wasincreased and exhibited a strong sensitivity for 119876le at KZ-Ara When it came to KZ-Bal the vegetation coverage waslarger than KZ-Ara while continuing to increase LAI hasslightly further improved the performance of 119876le Similarly119876ℎ and 119879soil showed a strong sensitivity to LAI The more thevegetation the more the solar radiation intercepted When itcame to1198850119898 the values of 119877net and119876ℎ were increased but119876leand 119879soil were decreased with the increase in 1198850119898 Figure 6indicated that 119876le 119876ℎ and 119879soil were highly sensitive to 1198850119898Taking the KZ-Ara site as an example 1198772 values for themodelled 119876le improved from 04 to 07 when 1198850119898 decreasedfrom 05 (S4) to 0005 (S5) Figure 6 also showed a strongsensitivity of albedo to 119877net 119876ℎ and 119879soil The imprecisesettings of albedomay enlarge the errors for sensible heat fluxin CoLMGround sensible heat flux was negatively correlatedto albedo The increase in the surface albedo decreasedthe solar radiation absorbed by soil and soil temperaturewas decreased However 119876le was slightly sensitive to albedo(Figure 6)
These sensitivity analysis results demonstrated that theimprovement in model performance observed in S2ndashS7 wassignificantly affected by the values of these three parametersin CoLM further justifying the significance of these keyparameters (LAI 1198850119898 and albedo) to the Common LandModel
4 Discussion
Arid and semiarid areas cover approximately one-third ofthe global terrestrial land surfaces [15] Central Asia has
Advances in Meteorology 7
KZ-AraSi
mul
atio
ns (W
mminus2)
S0
minus200
0
200
400
600
800
400 600 8000 200minus200
Observations (W mminus2)
(a) 119877net
KZ-AraS1
minus200
0
200
400
600
800
200 400 600 800minus200 0
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(b) 119877net
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400minus100
minus100
0
100
200
300
400
KZ-AraS0
Observations (W mminus2)
(c) 119876le
100
minus100
0
100
200
300
400
200 300 400minus100 0
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(d) 119876le
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
minus100
0
100
200
300
400
500
KZ-AraS0
Observations (W mminus2)
(e) 119876ℎ
minus100
0
100
200
300
400
500
minus100 0 100 200 300 400 500
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(f) 119876ℎKZ-Bal
Sim
ulat
ions
(W m
minus2)
S0
0 200 400 600 800minus200
minus200
0
200
400
600
800
Observations (W mminus2)
(g) 119877net
KZ-BalS1
minus200
0
200
400
600
800
minus200 0 200 400 600 800
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(h) 119877net
Figure 3 Continued
8 Advances in Meteorology
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
0
200
400
Observations (W mminus2)
KZ-BalS0
(i) 119876le
KZ-BalS1
0
200
400
minus100 0 100 200 300 400 500
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(j) 119876le
Sim
ulat
ions
(W m
minus2)
minus100
0
100
200
300
0 100 300minus100
Observations (W mminus2)
KZ-BalS0
200
(k) 119876ℎ
minus100
0
100
200
300
0 100 200 300minus100
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
KZ-BalS1
(l) 119876ℎ
Figure 3 Comparison between the measured half-hourly net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the valuessimulated by the reference model (S0) and the model with refined roughness length (S1) at the KZ-Ara and KZ-Bal sites The solid red linerepresents the linear regression between the simulation and the observed data and the dashed line represents a 1 1 relationship between thedatasets
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(a) KZ-Ara
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(b) KZ-Bal
Figure 4 Comparison between the measured net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the values simulatedby the reference model (S0) and themodel with refined roughness length (S1) on a diurnal course at the KZ-Ara and KZ-Bal sitesThe diurnalflux values were calculated as the mean values of all data at same measurement time in a day for the entire time period
Advances in Meteorology 9
20
30
40
50
60
70
80
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
Soil
dept
h (c
m)
DOY in 2012
10
20
30
40
(a)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
40555555555555555555555555555555555 1
(b)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
405555555555555555555 1
(c)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(d)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(e)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(f)
Figure 5 Panels (a) and (b) are the isograms changed with time and depth variation for the reference model (S0) and the model with refinedroughness length (S1) and (c) shows the corresponding observed data at the depth of 20ndash80 cm below ground at the KZ-Ara site panels (d)and (e) are the isograms for S0 and S1 and (f) shows the observations at the KZ-Bal site
vast proportion of dryland ecosystems where climate wasfeatured as hot and dry during growing season [19] and thevegetation is sparseThe characteristic of dryland ecosystemsis significantly different from those in humid areas in termsof surface energy budget Many researchers have focused onecosystem functioning and structure in response to climatechange Kariyeva et al [38] examined spatiotemporal changepatterns and interactions between terrestrial phenology andclimate change in Central Asia during the period 1981ndash2008Lioubimtseva et al [19] have found that local and regionalhuman impacts in arid zones can significantly modify landsurface albedo as well as water exchange and nutrient cyclesthat could have essential impacts on the climate at both theregional and global scales Such kind of analyses advancedour understanding on the response of dryland ecosystembehaviour and functioning to climate change Howeverapplying LSM into dryland ecosystems was far more limitedRecently the CoLM has been validated at one desert shrubecosystem in Xinjiang China and the results found that rootfraction and root water uptake have important effects on theperformance of CoLM in simulating energy and water fluxes[13 21] In this study a refined parameterization of CoLMhasbeen evaluated at two newly built EC sites in Kazakhstan
The most commonly used technique to obtain landsurface turbulent fluxes is measurement of eddy covariancesystem and the analyzer was based on flux footprint modelsThe footprint concept is the probability that a scalar comingfrom a given elemental source reaches the measurementpoint Footprint models describe the relationship between
the spatial distribution of surface sources and the measuredsignal using footprint functions Several flux footprintmodelshave been designed [39ndash42] But most of them cannotaccount for inhomogeneous turbulence or require largercomputational resources Gockede et al [43] improved anEulerian footprint model use of satellite maps for explicitassignment of surface type Gockede et al [44] and Rebmannet al [45] applied this newmodel at the EC sites and obtainedsatisfactory results At present footprint models are used toestimate the source areas contributing to the flux observa-tions In addition they provide a tool for quality control ofthe flux measurements and provide guidance in designingexperiments [46] Thus the footprint models have consider-able potential in microclimatology investigations especiallyin studies which include nonhomogeneous surfaces
The momentum roughness length (1198850119898) thermal rough-ness length (1198850ℎ) and the water vapor roughness length(1198850119908) are crucial parameters for calculating momentum andheat fluxes in bulk transfer equations which is one of theessential components in LSMs It has been widely observedthat 1198850119898 differs from 1198850ℎ and 1198850119908 [4 47] Unfortunately1198850119898 1198850ℎ and 1198850119908 up to date are still treated as constants inmost LSMs Inaccurate estimates of roughness length wouldenlarge the bias of simulated energy andwater fluxes in LSMsMany researchers have found that roughness length stronglydepended on surface heterogeneity vegetation height andcoverage [48 49] Therefore the values of roughness lengthvary considerably in different geographical context or veg-etation types [50ndash54] Dryland ecosystems were sparsely
10 Advances in Meteorology
20
260
200
140
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
80
140
80
60
40
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
70
50
30
(a) (b)
200
150
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
40
80
120
8
6
4
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
4
2
(c) (d)
KZ-Ara
2
03
06
08
09
095
099
280
0
160
220
Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs20
60
100
150
120
90
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
60
90
80
60
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
50
70
ObsS0S2S3
S4S5S6S7
0
2
25
302
0405
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
3
25
15
ObsS0S2S3
S4S5S6S7
(a) (b)
(c) (d)
KZ-Bal
Figure 6 Taylor diagramplot of the simulated119877net (a)119876le (b)119876ℎ (c) and119879soil (d) computed from a referencemodel (S0) and six independentsensitivity tests (S2ndashS7 see Table 3) from the CoLM against EC observations at the two EC sites Standard deviation (STD Wmminus2) iscalculated as the simulated variables divided by the observed data ldquoObsrdquo refers to observed data points Root mean square error (RMSEWmminus2) is represented by green lines 119877 is the correlation coefficient The higher the 119877 and the smaller the STD and RMSE the better theagreement between model and data When comparing two simulations with different parameter values the longer the distance between thetwo simulation points the greater the sensitivity to that parameter
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
4 Advances in Meteorology
Table 2 The parameterizations of the reference model (S0) and the model with refined roughness length (S1) in CLM at KZ-Ara site andKZ-Bal site RWUF is an optimal root water uptake function for dryland ecosystem (Jing et al 2014 [21]) and119882119888 119882119909 and 119898 are empiricalconstants in RWUF
Site Simulation Land cover type Vegetation fraction RWUF Roughness length for bare soil(119882119888119882119909119898) 1198850119898 1198850ℎ 1198850119908
KZ-Ara S0 Mixed shrub and grassland 30 (08054) 5119864 minus 2 5119864 minus 2 5119864 minus 2
S1 S0 S0 S0 2119864 minus 3 2119864 minus 4 2119864 minus 4
KZ-Bal S0(1) Irrigated cropland and pasture (35)(2) Herbaceous wetland (44)(3) Shrub (21)
40 (03074) 5119864 minus 2 5119864 minus 2 5119864 minus 2
S1 S0 S0 S0 2119864 minus 3 2119864 minus 4 2119864 minus 4
Table 3 A baseline of reference exprement (S0) and six different configurations of CLMat each site (S2ndashS7) as used in this study for sensitivitytesting Two sets of leaf area index (LAI) average roughness length (1198850119898) and average albedo (120572) were used in CLM ldquo119863rdquo refers to the modeldefault parameter setting
KZ-Ara KZ-BalSimulation LAI 1198850119898 120572 Simulation LAI 1198850119898 120572
S0 119863 119863(005) 119863 S0 119863 119863(01) 119863
S2 119863 lowast 07 119863 119863 S2 119863 lowast 07 119863 119863
S3 119863 lowast 13 119863 119863 S3 119863 lowast 13 119863 119863
S4 119863 0005 119863 S4 119863 001 119863
S5 119863 05 119863 S5 119863 1 119863
S6 119863 119863 119863 lowast 07 S6 119863 119863 119863 lowast 07
S7 119863 119863 119863 lowast 13 S7 119863 119863 119863 lowast 13
surface flux and temperature have different sensitivities to1198850ℎand 1198850119898 and the sensible heat flux is very sensitive to theparameterization schemes of 1198850ℎ in arid regions [16]
A common method to calculate 1198850119898 and 1198850ℎ is
1198850119898 = 119911119890(minus119896119880119906
lowast)minus120593119898(119911119871)
1198850ℎ = 119911119890(minus119896(119879minus119879
119904)119879119904)minus120593ℎ(119911119871)
(2)
where 119911119871 is a stability parameter (119871 is the Monin-Obukhovlength and 119911 is the observational height) and 120593119898(119911119871) isthe stability function of wind profile and becomes 0 at theneutral condition 119880 is the average wind speed and 119906lowast isthe surface friction velocity 120593ℎ(119911119871) is the stability functionof the temperature profile and becomes 0 at the neutralcondition 119896 is vonKarman constant and equals 04 generally119879 is the air temperature and119879119904 is the surface temperatureTherelationship between 1198850119898 and 1198850ℎ can be described as
ln(1198850119898
1198850ℎ
) = 119896119887minus1 (3)
where 119896119887minus1 can be obtained from the bulk transfer equationas
119896119887minus1=119896119906lowast (119879119904 minus 119879)
119867120588119888119901
minus [ln119911 minus 1198890
1198850119898
minus 120593ℎ (119911
119871)] (4)
where119867 is the observed sensible heat flux 120588 is the air density119888119901 is the specific heat for dry air and 1198890 is the zero plane
displacement Thus the relationship among 1198850119898 1198850ℎ and1198850119908 is described as
1198850ℎ = 1198850119908 = 1198850119898119890minus119896119887minus1
(5)
where the excess resistance to heat transfer 119896119887minus1 is importantto the sensible heat exchange between land surface andatmosphere and there are linear correlations between 119896119887
minus1
and surface temperature1198850119898 is physically related to the geometric roughness of
surface elements and can be derived from the wind speedand temperature profiles Bao et al [4] and Yang et al [34]argued that this scheme overestimated 1198850ℎ and would mis-estimate the energy and water fluxes Momentum transportis more efficient than heat transport due to the influence ofpressure fluctuation because individual roughness elementsmay enhance the momentum flux through form drag withlittle contribution to the area-averaged heat flux [35]
An appropriatemethod suggested by Zhang et al [36] wasemployed to estimate1198850119898 at desert ecosystems In desert1198850119898was estimated as 00019 plusmn 00071m and 1198850ℎ and 1198850119908 were atsame order but almost onemagnitude lower than1198850119898 [36] Inthis study1198850119898 was set as 0002m and1198850ℎ = 1198850119908 = 00002mfor bare soil at the two sites (Table 2)
This research further investigated the sensitivities of theCoLM to LAI 1198850119898 and 120572 Therefore other six simulationsby increasing or decreasing the value of each parameter wereconducted The specifications of all simulations are listedin Table 3 In order to avoid the cross-influence of eachparameter all sensitivity tests took the simulation S0 as areference
Advances in Meteorology 5
0 200100 300 500400 600 700minus100
minus100
0
100
200
300
400
500
600
700
Qle+Qh
(W m
minus2)
y = 076 lowast x + 2895
R2= 091
RMSE = 4645
Rnet minus G (W mminus2)
(a) KZ-Ara
minus100 5004003002001000 600 700
minus100
0
100
200
300
400
500
600
700
y = 095 lowast x + 156
R2= 097
+Qh
(W m
minus2)
Rnet minus G (W mminus2)
3519RMSE =
Qle
(b) KZ-Bal
Figure 2 Energy balance closure at the KZ-Ara site and the KZ-Bal site The slope of the fitted line represents energy closure ratio and 1198772 isthe coefficient of determination RMSE (Wmminus2) is the root mean square error The energy fluxes include sensible heat flux (119876ℎ) latent heatflux (119876le) net radiation (119877net) and ground heat flux (119866)
25 Statistical Analysis Energy balance ratio (EBR) [35] wasused to give an overall evaluation of energy balance closure byaveraging over random errors in the half-hourmeasurementsat two flux tower sites and it was calculated by
EBR =sum119899
119894=1(119876le + 119876ℎ)
sum119899
119894=1(119877net minus 119866)
(6)
where 119899 is the number of half hours of data The values ofEBR close to 1 indicate the best degree of energy balanceclosure Additionally coefficient of determination (1198772) rootmean square error (RMSE) slope (119887119904) and intercept (1198870) areused to justify the performance of the model
The Taylor diagram [37] was used to quantify the degreeof the sensitivities of the model to management parametersRMSE 1198772 and standard error (STD) are used in the TaylordiagramThe output of the model simulation is specified by asingle point with the STDbeing the polar axis and119877 the polarangle The ldquoreferencerdquo point represents observations and theother points refer to themodel results from the simulations ofsensitivity testsThe distances from the reference point to theother points representing the consequence of the relationshipindicate the RMSE The higher 119877 and the smaller the STDandRMSE the better the agreement betweenmodel and dataWhen comparing two simulations with different parametervalues the longer the distance between the two simulationpoints the greater the sensitivity to that parameter
3 Results
31 Energy Balance Closure The slopes of the linear regres-sion between the observed 119876le + 119876ℎ and 119877net minus 119866 were 076and 095 at KZ-Ara and KZ-Bal respectively The coefficient
of determination (1198772) of the observed 119876le + 119876ℎ and 119877net minus 119866was 091 and 097 and the root mean square error (RMSE)was 4645 and 3519Wmminus2 respectively (Figure 2) Energybalance ratio (EBR) at KZ-Ara and KZ-Bal was 111 and 106respectively
32 Modelled 119877119899119890119905 119876119897119890 119876ℎ and 119879119904119900119894119897 Figure 3 shows thecomparisons between the measurements and the simulationsof the reference model (S0) and the model with refinedroughness length (S1) for 119877net 119876le and 119876ℎ at the twoKazakhstan sites The reference model (S0) significantlyunderestimated the latent heat flux and overestimated thesensible heat flux at both sites However the performanceof the refined roughness length (S1) was largely improvedin simulating turbulent heat fluxes The latent heat flux wasincreased and sensible heat flux was decreased obviouslyAt the KZ-Ara site RMSE for 119877net decreased from 636 to395Wmminus2 1198772 values for 119876le given by the two simulations(S0 and S1) were 036 and 061 respectively and RMSEdecreased from 4115 in S0 to 333Wmminus2 in S1 (Table 4)For 119876ℎ simulations the RMSE for S0 was 12124Wmminus2as compared to 5847Wmminus2 for S1 (Table 4) The resultsindicated that the simulation with refined roughness length(S1) significantly improved the performance of the model forboth 119877net and 119876le and particularly for 119876ℎ At the KZ-Bal sitethe performance of the simulation with refined roughness(S1) was also greatly improved 1198772 values for 119876le given by S0and S1 were 09 and 092 respectively and RMSE decreasedfrom 4642 in S0 to 4311Wmminus2 in S1 1198772 values for 119876ℎgiven by the two simulations were 07 and 067 respectivelyand RMSE decreased from 7738 in S0 to 3797Wmminus2 in S1(Table 4 Figure 3)
6 Advances in Meteorology
Table 4 Model performance for simulating 119877net 119876le and 119876ℎ indicated by coefficient of determination (1198772) slope (119887119904) intercept (119887
0) and
root mean square error (RMSE Wmminus2) of linear regressions between model and observed data at the KZ-Ara site and the KZ-Bal site
Variables Reference model (S0) Refined roughness length (S1)1198772 RMSE 119887119904 1198870 119877
2 RMSE 119887119904 1198870
Site KZ-Ara119877net 097 636 112 2848 096 395 099 712119876le 036 4115 074 966 061 333 102 467119876ℎ 089 12124 186 566 086 5847 106 4693
Site KZ-Bal119877net 098 4386 104 227 098 3307 096 139119876le 09 4642 082 3057 092 4311 089 2867119876ℎ 07 7738 158 5087 067 3797 076 2908
To further investigate the effects of refined roughnesson the energy fluxes simulations Figure 4 shows the meandiurnal turbulent fluxes during growing seasons at the twosites Diurnal variations of the three components of energyfluxes showed typical characteristics at the KZ-Ara site thatis 119877net gt 119876ℎ gt 119876le (Figure 4(a)) The KZ-Bal site is locatedbetween oasis croplands and original deserts Although theamount of the average annual precipitation is similar to KZ-Ara site the characteristic of the energy fluxes allocationshowed higher latent heat flux and lower sensible heat fluxcompared to theKZ-Ara site (Figure 4(b)) At theKZ-Ara siteS0 overestimated 119877net with the peak value of 520Wmminus2 ascompared to the observed peak value 450Wmminus2 Howeverthe simulation with refined roughness (S1) produced a goodagreement for 119877net between the simulation and the measure-ments In addition S0 overestimated 119876ℎ with the peak valueof 400Wmminus2 at noontime as compared to 200Wmminus2 for themeasurements S1 improved the simulation for 119876ℎ as well AtKZ-Bal site both S0 and S1 agreed better with observationsfor 119877net and 119876le S0 significantly overestimated 119876ℎ at thedaytimeHowever the simulationwith refined roughness (S1)produced better agreement with the observations
Figure 5 showed the comparisons of soil temperaturebetween the observation and the simulations of the referencemodel (S0) and the model with refined roughness length (S1)at each site Although there are significant challenges when itcomes to validating soil temperature fromLSMbecause of thehigh sensitivities of simulated andmeasured soil temperatureto soil texture moisture conditions and the limitations ofthe measurement the results from the reference model (S0)and the model with refined roughness length (S1) appearedto be realistic and generally captured the seasonal variationfor soil temperature It was found that the modelled 119879soil inMay was underestimated and significantly improved in Juneand July but overestimated in August However S1 producedhigher values of soil temperature than S0 Additionally theCoLM simulated a smaller vertical soil temperature gradientas shown that the contour for the simulationswasmore sparsecompared with the observation
33 Sensitivity of CoLM to LAI 1198850119898 and 120572 In orderto investigate the sensitivities of CoLM the values of theparameters were adjusted in large ranges Key parameters
consist of LAI 1198850119898 and 120572 with regard to the simulationof 119877net 119876le 119876ℎ and 119879soil at the two EC sites A referencemodel (S0) and six independent sensitivity tests (S2ndashS7 seeTable 3) were conducted at each siteThis research only chosethe modelled soil temperatures at the depth of 20 cm forsensitivity tests in this study
The modelled 119876le and 119876ℎ in CoLM were divided intotwo parts the fluxes on vegetation leaves and the fluxes onthe ground Thus 119876le showed a more complicated sensitivityto LAI Since the leaf temperature increased and groundtemperature decreased with the increase in LAI the evap-otranspiration from the leaves improved and the groundevaporation weakened but the total latent heat flux wasincreased and exhibited a strong sensitivity for 119876le at KZ-Ara When it came to KZ-Bal the vegetation coverage waslarger than KZ-Ara while continuing to increase LAI hasslightly further improved the performance of 119876le Similarly119876ℎ and 119879soil showed a strong sensitivity to LAI The more thevegetation the more the solar radiation intercepted When itcame to1198850119898 the values of 119877net and119876ℎ were increased but119876leand 119879soil were decreased with the increase in 1198850119898 Figure 6indicated that 119876le 119876ℎ and 119879soil were highly sensitive to 1198850119898Taking the KZ-Ara site as an example 1198772 values for themodelled 119876le improved from 04 to 07 when 1198850119898 decreasedfrom 05 (S4) to 0005 (S5) Figure 6 also showed a strongsensitivity of albedo to 119877net 119876ℎ and 119879soil The imprecisesettings of albedomay enlarge the errors for sensible heat fluxin CoLMGround sensible heat flux was negatively correlatedto albedo The increase in the surface albedo decreasedthe solar radiation absorbed by soil and soil temperaturewas decreased However 119876le was slightly sensitive to albedo(Figure 6)
These sensitivity analysis results demonstrated that theimprovement in model performance observed in S2ndashS7 wassignificantly affected by the values of these three parametersin CoLM further justifying the significance of these keyparameters (LAI 1198850119898 and albedo) to the Common LandModel
4 Discussion
Arid and semiarid areas cover approximately one-third ofthe global terrestrial land surfaces [15] Central Asia has
Advances in Meteorology 7
KZ-AraSi
mul
atio
ns (W
mminus2)
S0
minus200
0
200
400
600
800
400 600 8000 200minus200
Observations (W mminus2)
(a) 119877net
KZ-AraS1
minus200
0
200
400
600
800
200 400 600 800minus200 0
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(b) 119877net
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400minus100
minus100
0
100
200
300
400
KZ-AraS0
Observations (W mminus2)
(c) 119876le
100
minus100
0
100
200
300
400
200 300 400minus100 0
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(d) 119876le
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
minus100
0
100
200
300
400
500
KZ-AraS0
Observations (W mminus2)
(e) 119876ℎ
minus100
0
100
200
300
400
500
minus100 0 100 200 300 400 500
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(f) 119876ℎKZ-Bal
Sim
ulat
ions
(W m
minus2)
S0
0 200 400 600 800minus200
minus200
0
200
400
600
800
Observations (W mminus2)
(g) 119877net
KZ-BalS1
minus200
0
200
400
600
800
minus200 0 200 400 600 800
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(h) 119877net
Figure 3 Continued
8 Advances in Meteorology
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
0
200
400
Observations (W mminus2)
KZ-BalS0
(i) 119876le
KZ-BalS1
0
200
400
minus100 0 100 200 300 400 500
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(j) 119876le
Sim
ulat
ions
(W m
minus2)
minus100
0
100
200
300
0 100 300minus100
Observations (W mminus2)
KZ-BalS0
200
(k) 119876ℎ
minus100
0
100
200
300
0 100 200 300minus100
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
KZ-BalS1
(l) 119876ℎ
Figure 3 Comparison between the measured half-hourly net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the valuessimulated by the reference model (S0) and the model with refined roughness length (S1) at the KZ-Ara and KZ-Bal sites The solid red linerepresents the linear regression between the simulation and the observed data and the dashed line represents a 1 1 relationship between thedatasets
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(a) KZ-Ara
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(b) KZ-Bal
Figure 4 Comparison between the measured net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the values simulatedby the reference model (S0) and themodel with refined roughness length (S1) on a diurnal course at the KZ-Ara and KZ-Bal sitesThe diurnalflux values were calculated as the mean values of all data at same measurement time in a day for the entire time period
Advances in Meteorology 9
20
30
40
50
60
70
80
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
Soil
dept
h (c
m)
DOY in 2012
10
20
30
40
(a)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
40555555555555555555555555555555555 1
(b)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
405555555555555555555 1
(c)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(d)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(e)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(f)
Figure 5 Panels (a) and (b) are the isograms changed with time and depth variation for the reference model (S0) and the model with refinedroughness length (S1) and (c) shows the corresponding observed data at the depth of 20ndash80 cm below ground at the KZ-Ara site panels (d)and (e) are the isograms for S0 and S1 and (f) shows the observations at the KZ-Bal site
vast proportion of dryland ecosystems where climate wasfeatured as hot and dry during growing season [19] and thevegetation is sparseThe characteristic of dryland ecosystemsis significantly different from those in humid areas in termsof surface energy budget Many researchers have focused onecosystem functioning and structure in response to climatechange Kariyeva et al [38] examined spatiotemporal changepatterns and interactions between terrestrial phenology andclimate change in Central Asia during the period 1981ndash2008Lioubimtseva et al [19] have found that local and regionalhuman impacts in arid zones can significantly modify landsurface albedo as well as water exchange and nutrient cyclesthat could have essential impacts on the climate at both theregional and global scales Such kind of analyses advancedour understanding on the response of dryland ecosystembehaviour and functioning to climate change Howeverapplying LSM into dryland ecosystems was far more limitedRecently the CoLM has been validated at one desert shrubecosystem in Xinjiang China and the results found that rootfraction and root water uptake have important effects on theperformance of CoLM in simulating energy and water fluxes[13 21] In this study a refined parameterization of CoLMhasbeen evaluated at two newly built EC sites in Kazakhstan
The most commonly used technique to obtain landsurface turbulent fluxes is measurement of eddy covariancesystem and the analyzer was based on flux footprint modelsThe footprint concept is the probability that a scalar comingfrom a given elemental source reaches the measurementpoint Footprint models describe the relationship between
the spatial distribution of surface sources and the measuredsignal using footprint functions Several flux footprintmodelshave been designed [39ndash42] But most of them cannotaccount for inhomogeneous turbulence or require largercomputational resources Gockede et al [43] improved anEulerian footprint model use of satellite maps for explicitassignment of surface type Gockede et al [44] and Rebmannet al [45] applied this newmodel at the EC sites and obtainedsatisfactory results At present footprint models are used toestimate the source areas contributing to the flux observa-tions In addition they provide a tool for quality control ofthe flux measurements and provide guidance in designingexperiments [46] Thus the footprint models have consider-able potential in microclimatology investigations especiallyin studies which include nonhomogeneous surfaces
The momentum roughness length (1198850119898) thermal rough-ness length (1198850ℎ) and the water vapor roughness length(1198850119908) are crucial parameters for calculating momentum andheat fluxes in bulk transfer equations which is one of theessential components in LSMs It has been widely observedthat 1198850119898 differs from 1198850ℎ and 1198850119908 [4 47] Unfortunately1198850119898 1198850ℎ and 1198850119908 up to date are still treated as constants inmost LSMs Inaccurate estimates of roughness length wouldenlarge the bias of simulated energy andwater fluxes in LSMsMany researchers have found that roughness length stronglydepended on surface heterogeneity vegetation height andcoverage [48 49] Therefore the values of roughness lengthvary considerably in different geographical context or veg-etation types [50ndash54] Dryland ecosystems were sparsely
10 Advances in Meteorology
20
260
200
140
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
80
140
80
60
40
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
70
50
30
(a) (b)
200
150
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
40
80
120
8
6
4
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
4
2
(c) (d)
KZ-Ara
2
03
06
08
09
095
099
280
0
160
220
Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs20
60
100
150
120
90
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
60
90
80
60
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
50
70
ObsS0S2S3
S4S5S6S7
0
2
25
302
0405
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
3
25
15
ObsS0S2S3
S4S5S6S7
(a) (b)
(c) (d)
KZ-Bal
Figure 6 Taylor diagramplot of the simulated119877net (a)119876le (b)119876ℎ (c) and119879soil (d) computed from a referencemodel (S0) and six independentsensitivity tests (S2ndashS7 see Table 3) from the CoLM against EC observations at the two EC sites Standard deviation (STD Wmminus2) iscalculated as the simulated variables divided by the observed data ldquoObsrdquo refers to observed data points Root mean square error (RMSEWmminus2) is represented by green lines 119877 is the correlation coefficient The higher the 119877 and the smaller the STD and RMSE the better theagreement between model and data When comparing two simulations with different parameter values the longer the distance between thetwo simulation points the greater the sensitivity to that parameter
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
Advances in Meteorology 5
0 200100 300 500400 600 700minus100
minus100
0
100
200
300
400
500
600
700
Qle+Qh
(W m
minus2)
y = 076 lowast x + 2895
R2= 091
RMSE = 4645
Rnet minus G (W mminus2)
(a) KZ-Ara
minus100 5004003002001000 600 700
minus100
0
100
200
300
400
500
600
700
y = 095 lowast x + 156
R2= 097
+Qh
(W m
minus2)
Rnet minus G (W mminus2)
3519RMSE =
Qle
(b) KZ-Bal
Figure 2 Energy balance closure at the KZ-Ara site and the KZ-Bal site The slope of the fitted line represents energy closure ratio and 1198772 isthe coefficient of determination RMSE (Wmminus2) is the root mean square error The energy fluxes include sensible heat flux (119876ℎ) latent heatflux (119876le) net radiation (119877net) and ground heat flux (119866)
25 Statistical Analysis Energy balance ratio (EBR) [35] wasused to give an overall evaluation of energy balance closure byaveraging over random errors in the half-hourmeasurementsat two flux tower sites and it was calculated by
EBR =sum119899
119894=1(119876le + 119876ℎ)
sum119899
119894=1(119877net minus 119866)
(6)
where 119899 is the number of half hours of data The values ofEBR close to 1 indicate the best degree of energy balanceclosure Additionally coefficient of determination (1198772) rootmean square error (RMSE) slope (119887119904) and intercept (1198870) areused to justify the performance of the model
The Taylor diagram [37] was used to quantify the degreeof the sensitivities of the model to management parametersRMSE 1198772 and standard error (STD) are used in the TaylordiagramThe output of the model simulation is specified by asingle point with the STDbeing the polar axis and119877 the polarangle The ldquoreferencerdquo point represents observations and theother points refer to themodel results from the simulations ofsensitivity testsThe distances from the reference point to theother points representing the consequence of the relationshipindicate the RMSE The higher 119877 and the smaller the STDandRMSE the better the agreement betweenmodel and dataWhen comparing two simulations with different parametervalues the longer the distance between the two simulationpoints the greater the sensitivity to that parameter
3 Results
31 Energy Balance Closure The slopes of the linear regres-sion between the observed 119876le + 119876ℎ and 119877net minus 119866 were 076and 095 at KZ-Ara and KZ-Bal respectively The coefficient
of determination (1198772) of the observed 119876le + 119876ℎ and 119877net minus 119866was 091 and 097 and the root mean square error (RMSE)was 4645 and 3519Wmminus2 respectively (Figure 2) Energybalance ratio (EBR) at KZ-Ara and KZ-Bal was 111 and 106respectively
32 Modelled 119877119899119890119905 119876119897119890 119876ℎ and 119879119904119900119894119897 Figure 3 shows thecomparisons between the measurements and the simulationsof the reference model (S0) and the model with refinedroughness length (S1) for 119877net 119876le and 119876ℎ at the twoKazakhstan sites The reference model (S0) significantlyunderestimated the latent heat flux and overestimated thesensible heat flux at both sites However the performanceof the refined roughness length (S1) was largely improvedin simulating turbulent heat fluxes The latent heat flux wasincreased and sensible heat flux was decreased obviouslyAt the KZ-Ara site RMSE for 119877net decreased from 636 to395Wmminus2 1198772 values for 119876le given by the two simulations(S0 and S1) were 036 and 061 respectively and RMSEdecreased from 4115 in S0 to 333Wmminus2 in S1 (Table 4)For 119876ℎ simulations the RMSE for S0 was 12124Wmminus2as compared to 5847Wmminus2 for S1 (Table 4) The resultsindicated that the simulation with refined roughness length(S1) significantly improved the performance of the model forboth 119877net and 119876le and particularly for 119876ℎ At the KZ-Bal sitethe performance of the simulation with refined roughness(S1) was also greatly improved 1198772 values for 119876le given by S0and S1 were 09 and 092 respectively and RMSE decreasedfrom 4642 in S0 to 4311Wmminus2 in S1 1198772 values for 119876ℎgiven by the two simulations were 07 and 067 respectivelyand RMSE decreased from 7738 in S0 to 3797Wmminus2 in S1(Table 4 Figure 3)
6 Advances in Meteorology
Table 4 Model performance for simulating 119877net 119876le and 119876ℎ indicated by coefficient of determination (1198772) slope (119887119904) intercept (119887
0) and
root mean square error (RMSE Wmminus2) of linear regressions between model and observed data at the KZ-Ara site and the KZ-Bal site
Variables Reference model (S0) Refined roughness length (S1)1198772 RMSE 119887119904 1198870 119877
2 RMSE 119887119904 1198870
Site KZ-Ara119877net 097 636 112 2848 096 395 099 712119876le 036 4115 074 966 061 333 102 467119876ℎ 089 12124 186 566 086 5847 106 4693
Site KZ-Bal119877net 098 4386 104 227 098 3307 096 139119876le 09 4642 082 3057 092 4311 089 2867119876ℎ 07 7738 158 5087 067 3797 076 2908
To further investigate the effects of refined roughnesson the energy fluxes simulations Figure 4 shows the meandiurnal turbulent fluxes during growing seasons at the twosites Diurnal variations of the three components of energyfluxes showed typical characteristics at the KZ-Ara site thatis 119877net gt 119876ℎ gt 119876le (Figure 4(a)) The KZ-Bal site is locatedbetween oasis croplands and original deserts Although theamount of the average annual precipitation is similar to KZ-Ara site the characteristic of the energy fluxes allocationshowed higher latent heat flux and lower sensible heat fluxcompared to theKZ-Ara site (Figure 4(b)) At theKZ-Ara siteS0 overestimated 119877net with the peak value of 520Wmminus2 ascompared to the observed peak value 450Wmminus2 Howeverthe simulation with refined roughness (S1) produced a goodagreement for 119877net between the simulation and the measure-ments In addition S0 overestimated 119876ℎ with the peak valueof 400Wmminus2 at noontime as compared to 200Wmminus2 for themeasurements S1 improved the simulation for 119876ℎ as well AtKZ-Bal site both S0 and S1 agreed better with observationsfor 119877net and 119876le S0 significantly overestimated 119876ℎ at thedaytimeHowever the simulationwith refined roughness (S1)produced better agreement with the observations
Figure 5 showed the comparisons of soil temperaturebetween the observation and the simulations of the referencemodel (S0) and the model with refined roughness length (S1)at each site Although there are significant challenges when itcomes to validating soil temperature fromLSMbecause of thehigh sensitivities of simulated andmeasured soil temperatureto soil texture moisture conditions and the limitations ofthe measurement the results from the reference model (S0)and the model with refined roughness length (S1) appearedto be realistic and generally captured the seasonal variationfor soil temperature It was found that the modelled 119879soil inMay was underestimated and significantly improved in Juneand July but overestimated in August However S1 producedhigher values of soil temperature than S0 Additionally theCoLM simulated a smaller vertical soil temperature gradientas shown that the contour for the simulationswasmore sparsecompared with the observation
33 Sensitivity of CoLM to LAI 1198850119898 and 120572 In orderto investigate the sensitivities of CoLM the values of theparameters were adjusted in large ranges Key parameters
consist of LAI 1198850119898 and 120572 with regard to the simulationof 119877net 119876le 119876ℎ and 119879soil at the two EC sites A referencemodel (S0) and six independent sensitivity tests (S2ndashS7 seeTable 3) were conducted at each siteThis research only chosethe modelled soil temperatures at the depth of 20 cm forsensitivity tests in this study
The modelled 119876le and 119876ℎ in CoLM were divided intotwo parts the fluxes on vegetation leaves and the fluxes onthe ground Thus 119876le showed a more complicated sensitivityto LAI Since the leaf temperature increased and groundtemperature decreased with the increase in LAI the evap-otranspiration from the leaves improved and the groundevaporation weakened but the total latent heat flux wasincreased and exhibited a strong sensitivity for 119876le at KZ-Ara When it came to KZ-Bal the vegetation coverage waslarger than KZ-Ara while continuing to increase LAI hasslightly further improved the performance of 119876le Similarly119876ℎ and 119879soil showed a strong sensitivity to LAI The more thevegetation the more the solar radiation intercepted When itcame to1198850119898 the values of 119877net and119876ℎ were increased but119876leand 119879soil were decreased with the increase in 1198850119898 Figure 6indicated that 119876le 119876ℎ and 119879soil were highly sensitive to 1198850119898Taking the KZ-Ara site as an example 1198772 values for themodelled 119876le improved from 04 to 07 when 1198850119898 decreasedfrom 05 (S4) to 0005 (S5) Figure 6 also showed a strongsensitivity of albedo to 119877net 119876ℎ and 119879soil The imprecisesettings of albedomay enlarge the errors for sensible heat fluxin CoLMGround sensible heat flux was negatively correlatedto albedo The increase in the surface albedo decreasedthe solar radiation absorbed by soil and soil temperaturewas decreased However 119876le was slightly sensitive to albedo(Figure 6)
These sensitivity analysis results demonstrated that theimprovement in model performance observed in S2ndashS7 wassignificantly affected by the values of these three parametersin CoLM further justifying the significance of these keyparameters (LAI 1198850119898 and albedo) to the Common LandModel
4 Discussion
Arid and semiarid areas cover approximately one-third ofthe global terrestrial land surfaces [15] Central Asia has
Advances in Meteorology 7
KZ-AraSi
mul
atio
ns (W
mminus2)
S0
minus200
0
200
400
600
800
400 600 8000 200minus200
Observations (W mminus2)
(a) 119877net
KZ-AraS1
minus200
0
200
400
600
800
200 400 600 800minus200 0
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(b) 119877net
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400minus100
minus100
0
100
200
300
400
KZ-AraS0
Observations (W mminus2)
(c) 119876le
100
minus100
0
100
200
300
400
200 300 400minus100 0
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(d) 119876le
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
minus100
0
100
200
300
400
500
KZ-AraS0
Observations (W mminus2)
(e) 119876ℎ
minus100
0
100
200
300
400
500
minus100 0 100 200 300 400 500
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(f) 119876ℎKZ-Bal
Sim
ulat
ions
(W m
minus2)
S0
0 200 400 600 800minus200
minus200
0
200
400
600
800
Observations (W mminus2)
(g) 119877net
KZ-BalS1
minus200
0
200
400
600
800
minus200 0 200 400 600 800
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(h) 119877net
Figure 3 Continued
8 Advances in Meteorology
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
0
200
400
Observations (W mminus2)
KZ-BalS0
(i) 119876le
KZ-BalS1
0
200
400
minus100 0 100 200 300 400 500
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(j) 119876le
Sim
ulat
ions
(W m
minus2)
minus100
0
100
200
300
0 100 300minus100
Observations (W mminus2)
KZ-BalS0
200
(k) 119876ℎ
minus100
0
100
200
300
0 100 200 300minus100
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
KZ-BalS1
(l) 119876ℎ
Figure 3 Comparison between the measured half-hourly net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the valuessimulated by the reference model (S0) and the model with refined roughness length (S1) at the KZ-Ara and KZ-Bal sites The solid red linerepresents the linear regression between the simulation and the observed data and the dashed line represents a 1 1 relationship between thedatasets
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(a) KZ-Ara
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(b) KZ-Bal
Figure 4 Comparison between the measured net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the values simulatedby the reference model (S0) and themodel with refined roughness length (S1) on a diurnal course at the KZ-Ara and KZ-Bal sitesThe diurnalflux values were calculated as the mean values of all data at same measurement time in a day for the entire time period
Advances in Meteorology 9
20
30
40
50
60
70
80
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
Soil
dept
h (c
m)
DOY in 2012
10
20
30
40
(a)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
40555555555555555555555555555555555 1
(b)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
405555555555555555555 1
(c)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(d)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(e)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(f)
Figure 5 Panels (a) and (b) are the isograms changed with time and depth variation for the reference model (S0) and the model with refinedroughness length (S1) and (c) shows the corresponding observed data at the depth of 20ndash80 cm below ground at the KZ-Ara site panels (d)and (e) are the isograms for S0 and S1 and (f) shows the observations at the KZ-Bal site
vast proportion of dryland ecosystems where climate wasfeatured as hot and dry during growing season [19] and thevegetation is sparseThe characteristic of dryland ecosystemsis significantly different from those in humid areas in termsof surface energy budget Many researchers have focused onecosystem functioning and structure in response to climatechange Kariyeva et al [38] examined spatiotemporal changepatterns and interactions between terrestrial phenology andclimate change in Central Asia during the period 1981ndash2008Lioubimtseva et al [19] have found that local and regionalhuman impacts in arid zones can significantly modify landsurface albedo as well as water exchange and nutrient cyclesthat could have essential impacts on the climate at both theregional and global scales Such kind of analyses advancedour understanding on the response of dryland ecosystembehaviour and functioning to climate change Howeverapplying LSM into dryland ecosystems was far more limitedRecently the CoLM has been validated at one desert shrubecosystem in Xinjiang China and the results found that rootfraction and root water uptake have important effects on theperformance of CoLM in simulating energy and water fluxes[13 21] In this study a refined parameterization of CoLMhasbeen evaluated at two newly built EC sites in Kazakhstan
The most commonly used technique to obtain landsurface turbulent fluxes is measurement of eddy covariancesystem and the analyzer was based on flux footprint modelsThe footprint concept is the probability that a scalar comingfrom a given elemental source reaches the measurementpoint Footprint models describe the relationship between
the spatial distribution of surface sources and the measuredsignal using footprint functions Several flux footprintmodelshave been designed [39ndash42] But most of them cannotaccount for inhomogeneous turbulence or require largercomputational resources Gockede et al [43] improved anEulerian footprint model use of satellite maps for explicitassignment of surface type Gockede et al [44] and Rebmannet al [45] applied this newmodel at the EC sites and obtainedsatisfactory results At present footprint models are used toestimate the source areas contributing to the flux observa-tions In addition they provide a tool for quality control ofthe flux measurements and provide guidance in designingexperiments [46] Thus the footprint models have consider-able potential in microclimatology investigations especiallyin studies which include nonhomogeneous surfaces
The momentum roughness length (1198850119898) thermal rough-ness length (1198850ℎ) and the water vapor roughness length(1198850119908) are crucial parameters for calculating momentum andheat fluxes in bulk transfer equations which is one of theessential components in LSMs It has been widely observedthat 1198850119898 differs from 1198850ℎ and 1198850119908 [4 47] Unfortunately1198850119898 1198850ℎ and 1198850119908 up to date are still treated as constants inmost LSMs Inaccurate estimates of roughness length wouldenlarge the bias of simulated energy andwater fluxes in LSMsMany researchers have found that roughness length stronglydepended on surface heterogeneity vegetation height andcoverage [48 49] Therefore the values of roughness lengthvary considerably in different geographical context or veg-etation types [50ndash54] Dryland ecosystems were sparsely
10 Advances in Meteorology
20
260
200
140
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
80
140
80
60
40
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
70
50
30
(a) (b)
200
150
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
40
80
120
8
6
4
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
4
2
(c) (d)
KZ-Ara
2
03
06
08
09
095
099
280
0
160
220
Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs20
60
100
150
120
90
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
60
90
80
60
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
50
70
ObsS0S2S3
S4S5S6S7
0
2
25
302
0405
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
3
25
15
ObsS0S2S3
S4S5S6S7
(a) (b)
(c) (d)
KZ-Bal
Figure 6 Taylor diagramplot of the simulated119877net (a)119876le (b)119876ℎ (c) and119879soil (d) computed from a referencemodel (S0) and six independentsensitivity tests (S2ndashS7 see Table 3) from the CoLM against EC observations at the two EC sites Standard deviation (STD Wmminus2) iscalculated as the simulated variables divided by the observed data ldquoObsrdquo refers to observed data points Root mean square error (RMSEWmminus2) is represented by green lines 119877 is the correlation coefficient The higher the 119877 and the smaller the STD and RMSE the better theagreement between model and data When comparing two simulations with different parameter values the longer the distance between thetwo simulation points the greater the sensitivity to that parameter
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
6 Advances in Meteorology
Table 4 Model performance for simulating 119877net 119876le and 119876ℎ indicated by coefficient of determination (1198772) slope (119887119904) intercept (119887
0) and
root mean square error (RMSE Wmminus2) of linear regressions between model and observed data at the KZ-Ara site and the KZ-Bal site
Variables Reference model (S0) Refined roughness length (S1)1198772 RMSE 119887119904 1198870 119877
2 RMSE 119887119904 1198870
Site KZ-Ara119877net 097 636 112 2848 096 395 099 712119876le 036 4115 074 966 061 333 102 467119876ℎ 089 12124 186 566 086 5847 106 4693
Site KZ-Bal119877net 098 4386 104 227 098 3307 096 139119876le 09 4642 082 3057 092 4311 089 2867119876ℎ 07 7738 158 5087 067 3797 076 2908
To further investigate the effects of refined roughnesson the energy fluxes simulations Figure 4 shows the meandiurnal turbulent fluxes during growing seasons at the twosites Diurnal variations of the three components of energyfluxes showed typical characteristics at the KZ-Ara site thatis 119877net gt 119876ℎ gt 119876le (Figure 4(a)) The KZ-Bal site is locatedbetween oasis croplands and original deserts Although theamount of the average annual precipitation is similar to KZ-Ara site the characteristic of the energy fluxes allocationshowed higher latent heat flux and lower sensible heat fluxcompared to theKZ-Ara site (Figure 4(b)) At theKZ-Ara siteS0 overestimated 119877net with the peak value of 520Wmminus2 ascompared to the observed peak value 450Wmminus2 Howeverthe simulation with refined roughness (S1) produced a goodagreement for 119877net between the simulation and the measure-ments In addition S0 overestimated 119876ℎ with the peak valueof 400Wmminus2 at noontime as compared to 200Wmminus2 for themeasurements S1 improved the simulation for 119876ℎ as well AtKZ-Bal site both S0 and S1 agreed better with observationsfor 119877net and 119876le S0 significantly overestimated 119876ℎ at thedaytimeHowever the simulationwith refined roughness (S1)produced better agreement with the observations
Figure 5 showed the comparisons of soil temperaturebetween the observation and the simulations of the referencemodel (S0) and the model with refined roughness length (S1)at each site Although there are significant challenges when itcomes to validating soil temperature fromLSMbecause of thehigh sensitivities of simulated andmeasured soil temperatureto soil texture moisture conditions and the limitations ofthe measurement the results from the reference model (S0)and the model with refined roughness length (S1) appearedto be realistic and generally captured the seasonal variationfor soil temperature It was found that the modelled 119879soil inMay was underestimated and significantly improved in Juneand July but overestimated in August However S1 producedhigher values of soil temperature than S0 Additionally theCoLM simulated a smaller vertical soil temperature gradientas shown that the contour for the simulationswasmore sparsecompared with the observation
33 Sensitivity of CoLM to LAI 1198850119898 and 120572 In orderto investigate the sensitivities of CoLM the values of theparameters were adjusted in large ranges Key parameters
consist of LAI 1198850119898 and 120572 with regard to the simulationof 119877net 119876le 119876ℎ and 119879soil at the two EC sites A referencemodel (S0) and six independent sensitivity tests (S2ndashS7 seeTable 3) were conducted at each siteThis research only chosethe modelled soil temperatures at the depth of 20 cm forsensitivity tests in this study
The modelled 119876le and 119876ℎ in CoLM were divided intotwo parts the fluxes on vegetation leaves and the fluxes onthe ground Thus 119876le showed a more complicated sensitivityto LAI Since the leaf temperature increased and groundtemperature decreased with the increase in LAI the evap-otranspiration from the leaves improved and the groundevaporation weakened but the total latent heat flux wasincreased and exhibited a strong sensitivity for 119876le at KZ-Ara When it came to KZ-Bal the vegetation coverage waslarger than KZ-Ara while continuing to increase LAI hasslightly further improved the performance of 119876le Similarly119876ℎ and 119879soil showed a strong sensitivity to LAI The more thevegetation the more the solar radiation intercepted When itcame to1198850119898 the values of 119877net and119876ℎ were increased but119876leand 119879soil were decreased with the increase in 1198850119898 Figure 6indicated that 119876le 119876ℎ and 119879soil were highly sensitive to 1198850119898Taking the KZ-Ara site as an example 1198772 values for themodelled 119876le improved from 04 to 07 when 1198850119898 decreasedfrom 05 (S4) to 0005 (S5) Figure 6 also showed a strongsensitivity of albedo to 119877net 119876ℎ and 119879soil The imprecisesettings of albedomay enlarge the errors for sensible heat fluxin CoLMGround sensible heat flux was negatively correlatedto albedo The increase in the surface albedo decreasedthe solar radiation absorbed by soil and soil temperaturewas decreased However 119876le was slightly sensitive to albedo(Figure 6)
These sensitivity analysis results demonstrated that theimprovement in model performance observed in S2ndashS7 wassignificantly affected by the values of these three parametersin CoLM further justifying the significance of these keyparameters (LAI 1198850119898 and albedo) to the Common LandModel
4 Discussion
Arid and semiarid areas cover approximately one-third ofthe global terrestrial land surfaces [15] Central Asia has
Advances in Meteorology 7
KZ-AraSi
mul
atio
ns (W
mminus2)
S0
minus200
0
200
400
600
800
400 600 8000 200minus200
Observations (W mminus2)
(a) 119877net
KZ-AraS1
minus200
0
200
400
600
800
200 400 600 800minus200 0
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(b) 119877net
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400minus100
minus100
0
100
200
300
400
KZ-AraS0
Observations (W mminus2)
(c) 119876le
100
minus100
0
100
200
300
400
200 300 400minus100 0
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(d) 119876le
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
minus100
0
100
200
300
400
500
KZ-AraS0
Observations (W mminus2)
(e) 119876ℎ
minus100
0
100
200
300
400
500
minus100 0 100 200 300 400 500
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(f) 119876ℎKZ-Bal
Sim
ulat
ions
(W m
minus2)
S0
0 200 400 600 800minus200
minus200
0
200
400
600
800
Observations (W mminus2)
(g) 119877net
KZ-BalS1
minus200
0
200
400
600
800
minus200 0 200 400 600 800
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(h) 119877net
Figure 3 Continued
8 Advances in Meteorology
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
0
200
400
Observations (W mminus2)
KZ-BalS0
(i) 119876le
KZ-BalS1
0
200
400
minus100 0 100 200 300 400 500
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(j) 119876le
Sim
ulat
ions
(W m
minus2)
minus100
0
100
200
300
0 100 300minus100
Observations (W mminus2)
KZ-BalS0
200
(k) 119876ℎ
minus100
0
100
200
300
0 100 200 300minus100
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
KZ-BalS1
(l) 119876ℎ
Figure 3 Comparison between the measured half-hourly net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the valuessimulated by the reference model (S0) and the model with refined roughness length (S1) at the KZ-Ara and KZ-Bal sites The solid red linerepresents the linear regression between the simulation and the observed data and the dashed line represents a 1 1 relationship between thedatasets
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(a) KZ-Ara
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(b) KZ-Bal
Figure 4 Comparison between the measured net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the values simulatedby the reference model (S0) and themodel with refined roughness length (S1) on a diurnal course at the KZ-Ara and KZ-Bal sitesThe diurnalflux values were calculated as the mean values of all data at same measurement time in a day for the entire time period
Advances in Meteorology 9
20
30
40
50
60
70
80
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
Soil
dept
h (c
m)
DOY in 2012
10
20
30
40
(a)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
40555555555555555555555555555555555 1
(b)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
405555555555555555555 1
(c)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(d)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(e)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(f)
Figure 5 Panels (a) and (b) are the isograms changed with time and depth variation for the reference model (S0) and the model with refinedroughness length (S1) and (c) shows the corresponding observed data at the depth of 20ndash80 cm below ground at the KZ-Ara site panels (d)and (e) are the isograms for S0 and S1 and (f) shows the observations at the KZ-Bal site
vast proportion of dryland ecosystems where climate wasfeatured as hot and dry during growing season [19] and thevegetation is sparseThe characteristic of dryland ecosystemsis significantly different from those in humid areas in termsof surface energy budget Many researchers have focused onecosystem functioning and structure in response to climatechange Kariyeva et al [38] examined spatiotemporal changepatterns and interactions between terrestrial phenology andclimate change in Central Asia during the period 1981ndash2008Lioubimtseva et al [19] have found that local and regionalhuman impacts in arid zones can significantly modify landsurface albedo as well as water exchange and nutrient cyclesthat could have essential impacts on the climate at both theregional and global scales Such kind of analyses advancedour understanding on the response of dryland ecosystembehaviour and functioning to climate change Howeverapplying LSM into dryland ecosystems was far more limitedRecently the CoLM has been validated at one desert shrubecosystem in Xinjiang China and the results found that rootfraction and root water uptake have important effects on theperformance of CoLM in simulating energy and water fluxes[13 21] In this study a refined parameterization of CoLMhasbeen evaluated at two newly built EC sites in Kazakhstan
The most commonly used technique to obtain landsurface turbulent fluxes is measurement of eddy covariancesystem and the analyzer was based on flux footprint modelsThe footprint concept is the probability that a scalar comingfrom a given elemental source reaches the measurementpoint Footprint models describe the relationship between
the spatial distribution of surface sources and the measuredsignal using footprint functions Several flux footprintmodelshave been designed [39ndash42] But most of them cannotaccount for inhomogeneous turbulence or require largercomputational resources Gockede et al [43] improved anEulerian footprint model use of satellite maps for explicitassignment of surface type Gockede et al [44] and Rebmannet al [45] applied this newmodel at the EC sites and obtainedsatisfactory results At present footprint models are used toestimate the source areas contributing to the flux observa-tions In addition they provide a tool for quality control ofthe flux measurements and provide guidance in designingexperiments [46] Thus the footprint models have consider-able potential in microclimatology investigations especiallyin studies which include nonhomogeneous surfaces
The momentum roughness length (1198850119898) thermal rough-ness length (1198850ℎ) and the water vapor roughness length(1198850119908) are crucial parameters for calculating momentum andheat fluxes in bulk transfer equations which is one of theessential components in LSMs It has been widely observedthat 1198850119898 differs from 1198850ℎ and 1198850119908 [4 47] Unfortunately1198850119898 1198850ℎ and 1198850119908 up to date are still treated as constants inmost LSMs Inaccurate estimates of roughness length wouldenlarge the bias of simulated energy andwater fluxes in LSMsMany researchers have found that roughness length stronglydepended on surface heterogeneity vegetation height andcoverage [48 49] Therefore the values of roughness lengthvary considerably in different geographical context or veg-etation types [50ndash54] Dryland ecosystems were sparsely
10 Advances in Meteorology
20
260
200
140
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
80
140
80
60
40
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
70
50
30
(a) (b)
200
150
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
40
80
120
8
6
4
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
4
2
(c) (d)
KZ-Ara
2
03
06
08
09
095
099
280
0
160
220
Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs20
60
100
150
120
90
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
60
90
80
60
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
50
70
ObsS0S2S3
S4S5S6S7
0
2
25
302
0405
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
3
25
15
ObsS0S2S3
S4S5S6S7
(a) (b)
(c) (d)
KZ-Bal
Figure 6 Taylor diagramplot of the simulated119877net (a)119876le (b)119876ℎ (c) and119879soil (d) computed from a referencemodel (S0) and six independentsensitivity tests (S2ndashS7 see Table 3) from the CoLM against EC observations at the two EC sites Standard deviation (STD Wmminus2) iscalculated as the simulated variables divided by the observed data ldquoObsrdquo refers to observed data points Root mean square error (RMSEWmminus2) is represented by green lines 119877 is the correlation coefficient The higher the 119877 and the smaller the STD and RMSE the better theagreement between model and data When comparing two simulations with different parameter values the longer the distance between thetwo simulation points the greater the sensitivity to that parameter
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
Advances in Meteorology 7
KZ-AraSi
mul
atio
ns (W
mminus2)
S0
minus200
0
200
400
600
800
400 600 8000 200minus200
Observations (W mminus2)
(a) 119877net
KZ-AraS1
minus200
0
200
400
600
800
200 400 600 800minus200 0
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(b) 119877net
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400minus100
minus100
0
100
200
300
400
KZ-AraS0
Observations (W mminus2)
(c) 119876le
100
minus100
0
100
200
300
400
200 300 400minus100 0
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(d) 119876le
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
minus100
0
100
200
300
400
500
KZ-AraS0
Observations (W mminus2)
(e) 119876ℎ
minus100
0
100
200
300
400
500
minus100 0 100 200 300 400 500
KZ-AraS1
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(f) 119876ℎKZ-Bal
Sim
ulat
ions
(W m
minus2)
S0
0 200 400 600 800minus200
minus200
0
200
400
600
800
Observations (W mminus2)
(g) 119877net
KZ-BalS1
minus200
0
200
400
600
800
minus200 0 200 400 600 800
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(h) 119877net
Figure 3 Continued
8 Advances in Meteorology
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
0
200
400
Observations (W mminus2)
KZ-BalS0
(i) 119876le
KZ-BalS1
0
200
400
minus100 0 100 200 300 400 500
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(j) 119876le
Sim
ulat
ions
(W m
minus2)
minus100
0
100
200
300
0 100 300minus100
Observations (W mminus2)
KZ-BalS0
200
(k) 119876ℎ
minus100
0
100
200
300
0 100 200 300minus100
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
KZ-BalS1
(l) 119876ℎ
Figure 3 Comparison between the measured half-hourly net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the valuessimulated by the reference model (S0) and the model with refined roughness length (S1) at the KZ-Ara and KZ-Bal sites The solid red linerepresents the linear regression between the simulation and the observed data and the dashed line represents a 1 1 relationship between thedatasets
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(a) KZ-Ara
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(b) KZ-Bal
Figure 4 Comparison between the measured net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the values simulatedby the reference model (S0) and themodel with refined roughness length (S1) on a diurnal course at the KZ-Ara and KZ-Bal sitesThe diurnalflux values were calculated as the mean values of all data at same measurement time in a day for the entire time period
Advances in Meteorology 9
20
30
40
50
60
70
80
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
Soil
dept
h (c
m)
DOY in 2012
10
20
30
40
(a)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
40555555555555555555555555555555555 1
(b)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
405555555555555555555 1
(c)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(d)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(e)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(f)
Figure 5 Panels (a) and (b) are the isograms changed with time and depth variation for the reference model (S0) and the model with refinedroughness length (S1) and (c) shows the corresponding observed data at the depth of 20ndash80 cm below ground at the KZ-Ara site panels (d)and (e) are the isograms for S0 and S1 and (f) shows the observations at the KZ-Bal site
vast proportion of dryland ecosystems where climate wasfeatured as hot and dry during growing season [19] and thevegetation is sparseThe characteristic of dryland ecosystemsis significantly different from those in humid areas in termsof surface energy budget Many researchers have focused onecosystem functioning and structure in response to climatechange Kariyeva et al [38] examined spatiotemporal changepatterns and interactions between terrestrial phenology andclimate change in Central Asia during the period 1981ndash2008Lioubimtseva et al [19] have found that local and regionalhuman impacts in arid zones can significantly modify landsurface albedo as well as water exchange and nutrient cyclesthat could have essential impacts on the climate at both theregional and global scales Such kind of analyses advancedour understanding on the response of dryland ecosystembehaviour and functioning to climate change Howeverapplying LSM into dryland ecosystems was far more limitedRecently the CoLM has been validated at one desert shrubecosystem in Xinjiang China and the results found that rootfraction and root water uptake have important effects on theperformance of CoLM in simulating energy and water fluxes[13 21] In this study a refined parameterization of CoLMhasbeen evaluated at two newly built EC sites in Kazakhstan
The most commonly used technique to obtain landsurface turbulent fluxes is measurement of eddy covariancesystem and the analyzer was based on flux footprint modelsThe footprint concept is the probability that a scalar comingfrom a given elemental source reaches the measurementpoint Footprint models describe the relationship between
the spatial distribution of surface sources and the measuredsignal using footprint functions Several flux footprintmodelshave been designed [39ndash42] But most of them cannotaccount for inhomogeneous turbulence or require largercomputational resources Gockede et al [43] improved anEulerian footprint model use of satellite maps for explicitassignment of surface type Gockede et al [44] and Rebmannet al [45] applied this newmodel at the EC sites and obtainedsatisfactory results At present footprint models are used toestimate the source areas contributing to the flux observa-tions In addition they provide a tool for quality control ofthe flux measurements and provide guidance in designingexperiments [46] Thus the footprint models have consider-able potential in microclimatology investigations especiallyin studies which include nonhomogeneous surfaces
The momentum roughness length (1198850119898) thermal rough-ness length (1198850ℎ) and the water vapor roughness length(1198850119908) are crucial parameters for calculating momentum andheat fluxes in bulk transfer equations which is one of theessential components in LSMs It has been widely observedthat 1198850119898 differs from 1198850ℎ and 1198850119908 [4 47] Unfortunately1198850119898 1198850ℎ and 1198850119908 up to date are still treated as constants inmost LSMs Inaccurate estimates of roughness length wouldenlarge the bias of simulated energy andwater fluxes in LSMsMany researchers have found that roughness length stronglydepended on surface heterogeneity vegetation height andcoverage [48 49] Therefore the values of roughness lengthvary considerably in different geographical context or veg-etation types [50ndash54] Dryland ecosystems were sparsely
10 Advances in Meteorology
20
260
200
140
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
80
140
80
60
40
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
70
50
30
(a) (b)
200
150
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
40
80
120
8
6
4
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
4
2
(c) (d)
KZ-Ara
2
03
06
08
09
095
099
280
0
160
220
Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs20
60
100
150
120
90
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
60
90
80
60
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
50
70
ObsS0S2S3
S4S5S6S7
0
2
25
302
0405
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
3
25
15
ObsS0S2S3
S4S5S6S7
(a) (b)
(c) (d)
KZ-Bal
Figure 6 Taylor diagramplot of the simulated119877net (a)119876le (b)119876ℎ (c) and119879soil (d) computed from a referencemodel (S0) and six independentsensitivity tests (S2ndashS7 see Table 3) from the CoLM against EC observations at the two EC sites Standard deviation (STD Wmminus2) iscalculated as the simulated variables divided by the observed data ldquoObsrdquo refers to observed data points Root mean square error (RMSEWmminus2) is represented by green lines 119877 is the correlation coefficient The higher the 119877 and the smaller the STD and RMSE the better theagreement between model and data When comparing two simulations with different parameter values the longer the distance between thetwo simulation points the greater the sensitivity to that parameter
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
8 Advances in Meteorology
Sim
ulat
ions
(W m
minus2)
0 100 200 300 400 500minus100
0
200
400
Observations (W mminus2)
KZ-BalS0
(i) 119876le
KZ-BalS1
0
200
400
minus100 0 100 200 300 400 500
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
(j) 119876le
Sim
ulat
ions
(W m
minus2)
minus100
0
100
200
300
0 100 300minus100
Observations (W mminus2)
KZ-BalS0
200
(k) 119876ℎ
minus100
0
100
200
300
0 100 200 300minus100
Observations (W mminus2)
Sim
ulat
ions
(W m
minus2)
KZ-BalS1
(l) 119876ℎ
Figure 3 Comparison between the measured half-hourly net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the valuessimulated by the reference model (S0) and the model with refined roughness length (S1) at the KZ-Ara and KZ-Bal sites The solid red linerepresents the linear regression between the simulation and the observed data and the dashed line represents a 1 1 relationship between thedatasets
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(a) KZ-Ara
Ener
gy fl
uxes
(W m
minus2)
600
500
400
300
200
100
0
minus100
4 8 12 16 20 240
Hour
Rnet_obsRnet_S0Rnet_S1
Qh_obsQh_Qh_
_Qle
_Qle
_obsQleS0
S1
S0S1
(b) KZ-Bal
Figure 4 Comparison between the measured net radiation (119877net) latent heat flux (119876le) and sensible heat flux (119876ℎ) and the values simulatedby the reference model (S0) and themodel with refined roughness length (S1) on a diurnal course at the KZ-Ara and KZ-Bal sitesThe diurnalflux values were calculated as the mean values of all data at same measurement time in a day for the entire time period
Advances in Meteorology 9
20
30
40
50
60
70
80
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
Soil
dept
h (c
m)
DOY in 2012
10
20
30
40
(a)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
40555555555555555555555555555555555 1
(b)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
405555555555555555555 1
(c)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(d)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(e)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(f)
Figure 5 Panels (a) and (b) are the isograms changed with time and depth variation for the reference model (S0) and the model with refinedroughness length (S1) and (c) shows the corresponding observed data at the depth of 20ndash80 cm below ground at the KZ-Ara site panels (d)and (e) are the isograms for S0 and S1 and (f) shows the observations at the KZ-Bal site
vast proportion of dryland ecosystems where climate wasfeatured as hot and dry during growing season [19] and thevegetation is sparseThe characteristic of dryland ecosystemsis significantly different from those in humid areas in termsof surface energy budget Many researchers have focused onecosystem functioning and structure in response to climatechange Kariyeva et al [38] examined spatiotemporal changepatterns and interactions between terrestrial phenology andclimate change in Central Asia during the period 1981ndash2008Lioubimtseva et al [19] have found that local and regionalhuman impacts in arid zones can significantly modify landsurface albedo as well as water exchange and nutrient cyclesthat could have essential impacts on the climate at both theregional and global scales Such kind of analyses advancedour understanding on the response of dryland ecosystembehaviour and functioning to climate change Howeverapplying LSM into dryland ecosystems was far more limitedRecently the CoLM has been validated at one desert shrubecosystem in Xinjiang China and the results found that rootfraction and root water uptake have important effects on theperformance of CoLM in simulating energy and water fluxes[13 21] In this study a refined parameterization of CoLMhasbeen evaluated at two newly built EC sites in Kazakhstan
The most commonly used technique to obtain landsurface turbulent fluxes is measurement of eddy covariancesystem and the analyzer was based on flux footprint modelsThe footprint concept is the probability that a scalar comingfrom a given elemental source reaches the measurementpoint Footprint models describe the relationship between
the spatial distribution of surface sources and the measuredsignal using footprint functions Several flux footprintmodelshave been designed [39ndash42] But most of them cannotaccount for inhomogeneous turbulence or require largercomputational resources Gockede et al [43] improved anEulerian footprint model use of satellite maps for explicitassignment of surface type Gockede et al [44] and Rebmannet al [45] applied this newmodel at the EC sites and obtainedsatisfactory results At present footprint models are used toestimate the source areas contributing to the flux observa-tions In addition they provide a tool for quality control ofthe flux measurements and provide guidance in designingexperiments [46] Thus the footprint models have consider-able potential in microclimatology investigations especiallyin studies which include nonhomogeneous surfaces
The momentum roughness length (1198850119898) thermal rough-ness length (1198850ℎ) and the water vapor roughness length(1198850119908) are crucial parameters for calculating momentum andheat fluxes in bulk transfer equations which is one of theessential components in LSMs It has been widely observedthat 1198850119898 differs from 1198850ℎ and 1198850119908 [4 47] Unfortunately1198850119898 1198850ℎ and 1198850119908 up to date are still treated as constants inmost LSMs Inaccurate estimates of roughness length wouldenlarge the bias of simulated energy andwater fluxes in LSMsMany researchers have found that roughness length stronglydepended on surface heterogeneity vegetation height andcoverage [48 49] Therefore the values of roughness lengthvary considerably in different geographical context or veg-etation types [50ndash54] Dryland ecosystems were sparsely
10 Advances in Meteorology
20
260
200
140
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
80
140
80
60
40
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
70
50
30
(a) (b)
200
150
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
40
80
120
8
6
4
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
4
2
(c) (d)
KZ-Ara
2
03
06
08
09
095
099
280
0
160
220
Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs20
60
100
150
120
90
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
60
90
80
60
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
50
70
ObsS0S2S3
S4S5S6S7
0
2
25
302
0405
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
3
25
15
ObsS0S2S3
S4S5S6S7
(a) (b)
(c) (d)
KZ-Bal
Figure 6 Taylor diagramplot of the simulated119877net (a)119876le (b)119876ℎ (c) and119879soil (d) computed from a referencemodel (S0) and six independentsensitivity tests (S2ndashS7 see Table 3) from the CoLM against EC observations at the two EC sites Standard deviation (STD Wmminus2) iscalculated as the simulated variables divided by the observed data ldquoObsrdquo refers to observed data points Root mean square error (RMSEWmminus2) is represented by green lines 119877 is the correlation coefficient The higher the 119877 and the smaller the STD and RMSE the better theagreement between model and data When comparing two simulations with different parameter values the longer the distance between thetwo simulation points the greater the sensitivity to that parameter
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
Advances in Meteorology 9
20
30
40
50
60
70
80
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
Soil
dept
h (c
m)
DOY in 2012
10
20
30
40
(a)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
40555555555555555555555555555555555 1
(b)
15 1
KZ-Ara
151 161 171 181 191 201 211 221 231 151 161 171
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012
10
20
30
405555555555555555555 1
(c)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(d)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(e)
20
30
40
50
60
70
80Soil
dept
h (c
m)
DOY in 2012114 124 134 144 154 164 174 184 194 204 214
KZ-Bal
10
15
20
25
30
(f)
Figure 5 Panels (a) and (b) are the isograms changed with time and depth variation for the reference model (S0) and the model with refinedroughness length (S1) and (c) shows the corresponding observed data at the depth of 20ndash80 cm below ground at the KZ-Ara site panels (d)and (e) are the isograms for S0 and S1 and (f) shows the observations at the KZ-Bal site
vast proportion of dryland ecosystems where climate wasfeatured as hot and dry during growing season [19] and thevegetation is sparseThe characteristic of dryland ecosystemsis significantly different from those in humid areas in termsof surface energy budget Many researchers have focused onecosystem functioning and structure in response to climatechange Kariyeva et al [38] examined spatiotemporal changepatterns and interactions between terrestrial phenology andclimate change in Central Asia during the period 1981ndash2008Lioubimtseva et al [19] have found that local and regionalhuman impacts in arid zones can significantly modify landsurface albedo as well as water exchange and nutrient cyclesthat could have essential impacts on the climate at both theregional and global scales Such kind of analyses advancedour understanding on the response of dryland ecosystembehaviour and functioning to climate change Howeverapplying LSM into dryland ecosystems was far more limitedRecently the CoLM has been validated at one desert shrubecosystem in Xinjiang China and the results found that rootfraction and root water uptake have important effects on theperformance of CoLM in simulating energy and water fluxes[13 21] In this study a refined parameterization of CoLMhasbeen evaluated at two newly built EC sites in Kazakhstan
The most commonly used technique to obtain landsurface turbulent fluxes is measurement of eddy covariancesystem and the analyzer was based on flux footprint modelsThe footprint concept is the probability that a scalar comingfrom a given elemental source reaches the measurementpoint Footprint models describe the relationship between
the spatial distribution of surface sources and the measuredsignal using footprint functions Several flux footprintmodelshave been designed [39ndash42] But most of them cannotaccount for inhomogeneous turbulence or require largercomputational resources Gockede et al [43] improved anEulerian footprint model use of satellite maps for explicitassignment of surface type Gockede et al [44] and Rebmannet al [45] applied this newmodel at the EC sites and obtainedsatisfactory results At present footprint models are used toestimate the source areas contributing to the flux observa-tions In addition they provide a tool for quality control ofthe flux measurements and provide guidance in designingexperiments [46] Thus the footprint models have consider-able potential in microclimatology investigations especiallyin studies which include nonhomogeneous surfaces
The momentum roughness length (1198850119898) thermal rough-ness length (1198850ℎ) and the water vapor roughness length(1198850119908) are crucial parameters for calculating momentum andheat fluxes in bulk transfer equations which is one of theessential components in LSMs It has been widely observedthat 1198850119898 differs from 1198850ℎ and 1198850119908 [4 47] Unfortunately1198850119898 1198850ℎ and 1198850119908 up to date are still treated as constants inmost LSMs Inaccurate estimates of roughness length wouldenlarge the bias of simulated energy andwater fluxes in LSMsMany researchers have found that roughness length stronglydepended on surface heterogeneity vegetation height andcoverage [48 49] Therefore the values of roughness lengthvary considerably in different geographical context or veg-etation types [50ndash54] Dryland ecosystems were sparsely
10 Advances in Meteorology
20
260
200
140
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
80
140
80
60
40
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
70
50
30
(a) (b)
200
150
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
40
80
120
8
6
4
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
4
2
(c) (d)
KZ-Ara
2
03
06
08
09
095
099
280
0
160
220
Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs20
60
100
150
120
90
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
60
90
80
60
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
50
70
ObsS0S2S3
S4S5S6S7
0
2
25
302
0405
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
3
25
15
ObsS0S2S3
S4S5S6S7
(a) (b)
(c) (d)
KZ-Bal
Figure 6 Taylor diagramplot of the simulated119877net (a)119876le (b)119876ℎ (c) and119879soil (d) computed from a referencemodel (S0) and six independentsensitivity tests (S2ndashS7 see Table 3) from the CoLM against EC observations at the two EC sites Standard deviation (STD Wmminus2) iscalculated as the simulated variables divided by the observed data ldquoObsrdquo refers to observed data points Root mean square error (RMSEWmminus2) is represented by green lines 119877 is the correlation coefficient The higher the 119877 and the smaller the STD and RMSE the better theagreement between model and data When comparing two simulations with different parameter values the longer the distance between thetwo simulation points the greater the sensitivity to that parameter
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
10 Advances in Meteorology
20
260
200
140
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
80
140
80
60
40
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
70
50
30
(a) (b)
200
150
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
40
80
120
8
6
4
0
0204
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
4
2
(c) (d)
KZ-Ara
2
03
06
08
09
095
099
280
0
160
220
Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs20
60
100
150
120
90
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
60
90
80
60
100
0
03
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
30
50
70
ObsS0S2S3
S4S5S6S7
0
2
25
302
0405
06
08
09
095
099Stan
dard
dev
iatio
n
Correlation coefficient
Root mean square errorObs
3
25
15
ObsS0S2S3
S4S5S6S7
(a) (b)
(c) (d)
KZ-Bal
Figure 6 Taylor diagramplot of the simulated119877net (a)119876le (b)119876ℎ (c) and119879soil (d) computed from a referencemodel (S0) and six independentsensitivity tests (S2ndashS7 see Table 3) from the CoLM against EC observations at the two EC sites Standard deviation (STD Wmminus2) iscalculated as the simulated variables divided by the observed data ldquoObsrdquo refers to observed data points Root mean square error (RMSEWmminus2) is represented by green lines 119877 is the correlation coefficient The higher the 119877 and the smaller the STD and RMSE the better theagreement between model and data When comparing two simulations with different parameter values the longer the distance between thetwo simulation points the greater the sensitivity to that parameter
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
Advances in Meteorology 11
vegetated with vegetation fraction of 10ndash40 at CentralAsia desert ecosystems Bare soil has larger proportion inthe Central Asia desert ecosystems and its characteristicis entirely different from the high-vegetated land surfaceparticularly for the roughness lengthThe values of roughnesslength were replaced by empirical formula based on experi-mental observations in this study and the results found thatsuch treatment significantly improved the performance ofthe model The simulated turbulent heat fluxes with defaultroughness length showed very large variability during a dayespecially at daytime The most possible reason is that theroughness length was extremely overestimated and henceaerodynamic resistance was overestimated in the defaultversion of CoLM As a consequence the turbulent exchangewas strong and of high fluctuations However observeddiurnal dynamics of energy fluxes are quite harmoniousUsing an empirical approach to estimate roughness lengthinto CoLM significantly improved the performance in sim-ulating turbulent heat fluxes which indicated that accurateparameterization of roughness length is of crucial importancefor predicting energy and water fluxes in LSM when appliedin Central Asia desert ecosystems
Many researches have indicated that LSMs are very sensi-tive tomodel physics land characteristics (including leaf areaindex roughness length albedo and soil properties) andforcing [9 29 55 56] In Central Asia dryland ecosystemsleaf area index albedo and roughness length should bemuchmore important due to low fraction of vegetation and highheterogeneity of the land surface Compared with recentfindings at a Chinese desert shrub ecosystem [21] which isquite similar to the two sites used in this study that simulationof latent heat flux in CoLM was strongly dependent on thechoice of different root water uptake functions [21] Landsurface and vegetation parameters such as albedo roughnesslength and LAI also showed significant sensitivities in sim-ulating the energy and water fluxes in Central Asia drylandecosystems These researches implied that CoLM can bevery sensitive to both belowground ecological process (suchas root water uptake) and aboveground soil and vegetationproperties (LAI 1198850119898 and 120572) in Central Asia regions whichis different from the sensitivities of LSMs in humid regions[8 26 57]
Vegetation and soil parameters such as leaf area indexroughness length and albedo are closely related to vegetationcoverage on the land surface With recent findings at a Chi-nese desert shrub ecosystem [21] root water uptake processis more relevant to root distribution and root water uptakeefficiency However many of these crucial parameters aredifficult to observe in large areas Biases in land surface forc-ing data and parameterizations in representing soil moisturesoil temperature and other variables in numerical weatherforecast and climate models will enlarge the model errorsof water and energy fluxes Reinitialization of land surfacestates would mollify this problem if the land surface variablesand parameters were reliable and available in large areas andwith high spatial resolution Fortunately the improvement ofremote sensing technologies made these operable Remotesensing technologies also could obtain LAI canopy andground albedo vegetation height and other parameters in
large regions To make best use of the satellite-based andground-based observational data in land surface modellingfor investigating global climate change issue in regional areamany land data assimilation systems have been developed[58 59] Such treatment is to generate optimal fields forparameterizing and forcing LSMs [59] Broad use of land dataassimilation systemsrsquo results is valuable for predicting climatechange weather and biological and agricultural productivityand for performing a wide array of studies in the broaderbiogeosciences
5 Conclusions
In this study the CoLM for the first time has been eval-uated at two Central Asia desert ecosystems Additionallysensitivities of the model to LAI1198850119898 and 120572were conductedEvaluation of the CoLM and their sensitivities against theobserved energy fluxes using eddy covariance system and thesensitivity tests resulted in the following conclusions
(1) The reference simulations (S0) significantly under-estimated the latent heat flux and overestimated thesensible heat flux at two sites especially at KZ-AraHowever refined estimate of roughness length (S1)significantly improved the performance in simulat-ing turbulent heat fluxes The latent heat flux wasincreased but sensible heat flux was decreased whichwere in better agreement with the observations fromeddy covariance system
(2) Sensitivity analysis regarding leaf area index rough-ness length and albedo showed that net radiation isvery sensitive to albedo but latent and sensible heatfluxes and soil temperature are sensitively varyingwith the estimate of 1198850119898 at two EC sites over CentralAsia
Conflict of Interests
The authors declare that there is no conflict of interestsregarding the publication of this paper
Acknowledgments
This research was supported by the National Natural ScienceFoundation of China (Grant no 41561021) and NSFC-XJproject (Grant no U1403382)
References
[1] J Williams R G Barry andW M Washington ldquoSimulation ofthe atmospheric circulation using the NCAR global circulationmodel with ice age boundary conditionsrdquo Journal of AppliedMeteorology vol 13 no 3 pp 305ndash317 1974
[2] G B Bonan ldquoLand surface model (LSM version 10) for ecolog-ical hydrological and atmospheric studies technical descrip-tion and users guiderdquo Technical Note PBndash97-131494XABNCARTNndash417-STR Climate and Global Dynamics DivisionNational Center for Atmospheric Research Boulder ColoUSA 1996
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
12 Advances in Meteorology
[3] A J Pitman ldquoThe evolution of and revolution in land surfaceschemes designed for climate modelsrdquo International Journal ofClimatology vol 23 no 5 pp 479ndash510 2003
[4] Y Bao H Zuo S Lv and Y Zhang ldquoThe effect of improved landsurface process parameters in Arid area on climatic simulationin GCMrdquo PlateauMeteorology vol 23 no 2 pp 220ndash227 2004
[5] S Manabe ldquoClimate and the ocean circulation I The atmo-spheric circulation and the hydrology of the Earthrsquos surfacerdquoMonthly Weather Review vol 97 no 11 pp 739ndash774 1969
[6] F M Schurr O Bossdorf S J Milton and J Schumacher ldquoSpa-tial pattern formation in semi-arid shrubland a priori predictedversus observed pattern characteristicsrdquo Plant Ecology vol 173no 2 pp 271ndash282 2004
[7] MWilliams A D RichardsonM Reichstein et al ldquoImprovingland surface models with FLUXNET datardquo Biogeosciences vol6 no 7 pp 1341ndash1359 2009
[8] G Abramowitz R LeuningM Clark and A Pitman ldquoEvaluat-ing the performance of land surfacemodelsrdquo Journal of Climatevol 21 no 21 pp 5468ndash5481 2008
[9] M Choi S O Lee and H Kwon ldquoUnderstanding of thecommon land model performance for water and energy fluxesin a farmland during the growing season inKoreardquoHydrologicalProcesses vol 24 no 8 pp 1063ndash1071 2010
[10] M L Goulden J W Munger S-M Fan B C Daube and SCWofsy ldquoMeasurements of carbon sequestration by long-termeddy covariance methods and a critical evaluation of accuracyrdquoGlobal Change Biology vol 2 no 3 pp 169ndash182 1996
[11] J Mao L Dan BWang and Y Dai ldquoSimulation and evaluationof terrestrial ecosystem NPP with M-SDGVM over continentalChinardquoAdvances in Atmospheric Sciences vol 27 no 2 pp 427ndash442 2010
[12] T W Hudiburg B E Law and P E Thornton ldquoEvaluationand improvement of the Community Land Model (CLM4) inOregon forestsrdquo Biogeosciences vol 10 no 1 pp 453ndash470 2013
[13] L Li C van der Tol X Chen et al ldquoRepresenting the rootwater uptake process in the Common Land Model for bettersimulating the energy and water vapour fluxes in a CentralAsian desert ecosystemrdquo Journal of Hydrology vol 502 pp 145ndash155 2013
[14] L Li Y Wang Q Yu et al ldquoImproving the responses of theAustralian community land surfacemodel (CABLE) to seasonaldroughtrdquo Journal of Geophysical Research G Biogeosciences vol117 no 4 2012
[15] R Lal ldquoCarbon sequestration in dryland ecosystemsrdquo Environ-mental Management vol 33 no 4 pp 528ndash544 2004
[16] Y Chen K Yang D Zhou J Qin and X Guo ldquoImproving thenoah land surface model in arid regions with an appropriateparameterization of the thermal roughness lengthrdquo Journal ofHydrometeorology vol 11 no 4 pp 995ndash1006 2010
[17] X Zeng X Zeng and M Barlage ldquoGrowing temperate shrubsover arid and semiarid regions in the Community LandModel-Dynamic Global Vegetation Modelrdquo Global BiogeochemicalCycles vol 22 no 3 p 3003 2008
[18] Y Dai X Zeng R E Dickinson et al ldquoThe common landmodelrdquo Bulletin of the American Meteorological Society vol 84no 8 pp 1013ndash1023 2003
[19] E Lioubimtseva R Cole J M Adams and G KapustinldquoImpacts of climate and land-cover changes in arid lands ofCentral Asiardquo Journal of Arid Environments vol 62 no 2 pp285ndash308 2005
[20] L Li G Luo X Chen et al ldquoModelling evapotranspiration in aCentral Asian desert ecosystemrdquo Ecological Modelling vol 222no 20ndash22 pp 3680ndash3691 2011
[21] C-Q Jing L Li X Chen and G-P Luo ldquoComparison of rootwater uptake functions to simulate surface energy fluxes withina deep-rooted desert shrub ecosystemrdquo Hydrological Processesvol 28 no 21 pp 5436ndash5449 2014
[22] V M Starodubtsev and S R Truskavetskiy ldquoDesertificationprocesses in the Ili River delta under anthropogenic pressurerdquoWater Resources vol 38 no 2 pp 253ndash256 2011
[23] L Li X Chen C van der Tol G Luo and Z Su ldquoGrowingseason net ecosystem CO2 exchange of two desert ecosystemswith alkaline soils in Kazakhstanrdquo Ecology and Evolution vol 4no 1 pp 14ndash26 2014
[24] J T Kiehl J JHackG B Bonan BA BovilleD LWilliamsonand P J Rasch ldquoThe national center for atmospheric researchcommunity climate model CCM3lowastrdquo Journal of Climate vol 11no 6 pp 1131ndash1149 1998
[25] W J Riley Z M Subin D M Lawrence et al ldquoBarriers topredicting changes in global terrestrial methane fluxes analysesusing CLM4Me a methane biogeochemistry model integratedin CESMrdquo Biogeosciences vol 8 no 7 pp 1925ndash1953 2011
[26] S Bachner A Kapala and C Simmer ldquoEvaluation of dailyprecipitation characteristics in the CLM and their sensitivity toparameterizationsrdquoMeteorologische Zeitschrift vol 17 no 4 pp407ndash419 2008
[27] K W Oleson G Y Niu Z L Yang et al ldquoImprovements to thecommunity land model and their impact on the hydrologicalcyclerdquo Journal of Geophysical Research vol 113 no 1 2008
[28] G B Bonan P J Lawrence K W Oleson et al ldquoImprovingcanopy processes in the Community Land Model version 4(CLM4) using global flux fields empirically inferred fromFLUXNETdatardquo Journal of Geophysical Research vol 116 articleG2 2011
[29] A Henderson-Sellers Z-L Yang and R E Dickinson ldquoTheproject for intercomparison of land surface parameterisationschemesrdquo Bulletin of the American Meteorological Society vol74 no 7 pp 1335ndash1349 1993
[30] X Zeng M Shajkh Y Dai R E Dickinson and R MynenildquoCoupling of the common landmodel to theNCAR communityclimate modelrdquo Journal of Climate vol 15 no 14 pp 1832ndash18542002
[31] T Foken ldquo50 years of the Monin-Obukhov similarity theoryrdquoBoundary-Layer Meteorology vol 119 no 3 pp 431ndash447 2006
[32] X Zeng and R E Dickinson ldquoEffect of surface sublayer onsurface skin temperature and fluxesrdquo Journal of Climate vol 11no 4 pp 537ndash550 1998
[33] M KandaM Kanega T Kawai RMoriwaki andH SugawaraldquoRoughness lengths for momentum and heat derived fromoutdoor urban scale modelsrdquo Journal of Applied Meteorology ampClimatology vol 46 no 7 pp 1067ndash1079 2007
[34] K Yang T Koike H Ishikawa et al ldquoTurbulent flux transferover bare-soil surfaces characteristics and parameterizationrdquoJournal of Applied Meteorology amp Climatology vol 47 no 1 pp276ndash290 2008
[35] L Mahrt ldquoFlux sampling errors for aircraft and towersrdquo Journalof Atmospheric and Oceanic Technology vol 15 no 2 pp 416ndash429 1998
[36] Q Zhang X Cao G Wei and R Huang ldquoObservation andstudy of land surface parameters over Gobi in typical aridregionrdquoAdvances in Atmospheric Sciences vol 19 no 1 pp 120ndash135 2002
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
Advances in Meteorology 13
[37] K E Taylor ldquoSummarizing multiple aspects of model perfor-mance in a single diagramrdquo Journal of Geophysical ResearchAtmospheres vol 106 no 7 pp 7183ndash7192 2001
[38] J Kariyeva W J D van Leeuwen and C A WoodhouseldquoImpacts of climate gradients on the vegetation phenology ofmajor land use types in Central Asia (1981ndash2008)rdquo Frontiers ofEarth Science vol 6 no 2 pp 206ndash225 2012
[39] M Y Leclerc and G W Thurtell ldquoFootprint prediction ofscalar fluxes using a Markovian analysisrdquo Boundary-LayerMeteorology vol 52 no 3 pp 247ndash258 1990
[40] H P Schmid ldquoSource areas for scalars and scalar fluxesrdquoBoundary-Layer Meteorology vol 67 no 3 pp 293ndash318 1994
[41] T W Horst and J C Weil ldquoFootprint estimation for scalarfluxmeasurements in the atmospheric surface layerrdquoBoundary-Layer Meteorology vol 59 no 3 pp 279ndash296 1992
[42] A Sogachev M Y Leclerc A Karipot G Zhang and T VesalaldquoEffect of clearcuts on footprints and flux measurements abovea forest canopyrdquo Agricultural amp Forest Meteorology vol 133 no1ndash4 pp 182ndash196 2005
[43] M Gockede C Rebmann and T Foken ldquoA combination ofquality assessment tools for eddy covariance measurementswith footprint modelling for the characterisation of complexsitesrdquo Agricultural and Forest Meteorology vol 127 no 3-4 pp175ndash188 2004
[44] M Gockede T Markkanen M Mauder K Arnold J-P Lepsand T Foken ldquoValidation of footprint models using naturaltracer measurements from a field experimentrdquoAgricultural andForest Meteorology vol 135 no 1ndash4 pp 314ndash325 2005
[45] C Rebmann M Gockede T Foken et al ldquoQuality analysisapplied on eddy covariance measurements at complex forestsites using footprint modellingrdquoTheoretical and Applied Clima-tology vol 80 no 2ndash4 pp 121ndash141 2005
[46] G Peng X Cai H Zhang A Li F Hu and M Y LeclercldquoHeat flux apportionment to heterogeneous surfaces using fluxfootprint analysisrdquo Advances in Atmospheric Sciences vol 25no 1 pp 107ndash116 2008
[47] J R Garratt The Atmospheric Boundary Layer CambridgeUniversity Press 1994
[48] Y Zhou X Sun Z Zhu et al ldquoSurface roughness lengthdynamic over several different surfaces and its effects onmodeling fluxesrdquo Science in China Series D Earth Sciences vol49 no 2 pp 262ndash272 2006
[49] W Brutsaert ldquoHeat andmass transfer to and from surfaces withdense vegetation or similar permeable roughnessrdquo Boundary-Layer Meteorology vol 16 no 4 pp 365ndash388 1979
[50] B W Brock I C Willis and M J Sharp ldquoMeasurement andparameterization of aerodynamic roughness length variationsat Haut Glacier drsquoArolla Switzerlandrdquo Journal of Glaciology vol52 no 177 pp 281ndash297 2006
[51] Z Dong S Gao and D W Fryrear ldquoDrag coefficients rough-ness length and zero-plane displacement height as disturbed byartificial standing vegetationrdquo Journal of Arid Environments vol49 no 3 pp 485ndash505 2001
[52] N J Clifford A Robert and K S Richards ldquoEstimation offlow resistance in gravel-bedded rivers a physical explanationof the multiplier of roughness lengthrdquo Earth Surface Processesamp Landforms vol 17 no 2 pp 111ndash126 1992
[53] H A R De Bruin and C J Moore ldquoZero-plane displacementand roughness length for tall vegetation derived from a simplemass conservation hypothesisrdquo Boundary-Layer Meteorologyvol 31 no 1 pp 39ndash49 1985
[54] A C Chamberlain ldquoRoughness length of sea sand and snowrdquoBoundary-Layer Meteorology vol 25 no 4 pp 405ndash409 1983
[55] M B Ek K E Mitchell Y Lin et al ldquoImplementation ofNoah land surface model advances in the National Centers forEnvironmental Prediction operational mesoscale Eta modelrdquoJournal of Geophysical Research D Atmospheres vol 108 no 22pp 1ndash16 2003
[56] H Kato M Rodell F Beyrich et al ldquoSensitivity of land surfacesimulations tomodel physics land characteristics and forcingsat four CEOP sitesrdquo Journal of the Meteorological Society ofJapanmdashSeries II vol 85 pp 187ndash204 2007
[57] C K G Castillo and K R Gurney ldquoA sensitivity analysis ofsurface biophysical carbon and climate impacts of tropicaldeforestation rates in CCSM4-CNDVrdquo Journal of Climate vol26 no 3 pp 805ndash821 2013
[58] W T Crow and E F Wood ldquoThe assimilation of remotelysensed soil brightness temperature imagery into a land surfacemodel using Ensemble Kalman filtering a case study basedon ESTAR measurements during SGP97rdquo Advances in WaterResources vol 26 no 2 pp 137ndash149 2003
[59] M Rodell P R Houser U Jambor et al ldquoThe global land dataassimilation systemrdquo Bulletin of the American MeteorologicalSociety vol 85 no 3 pp 381ndash394 2004
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in
Submit your manuscripts athttpwwwhindawicom
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ClimatologyJournal of
EcologyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
EarthquakesJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom
Applied ampEnvironmentalSoil Science
Volume 2014
Mining
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Hindawi Publishing Corporation httpwwwhindawicom Volume 2014
International Journal of
Geophysics
OceanographyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal ofPetroleum Engineering
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Journal of
Atmospheric SciencesInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Advances in
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MineralogyInternational Journal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
MeteorologyAdvances in
The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014
Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geological ResearchJournal of
Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014
Geology Advances in