Research Article Estimating the Surface Air Temperature by

Research ArticleEstimating the Surface Air Temperature byRemote Sensing in Northwest China Using an ImprovedAdvection-Energy Balance for Air Temperature Model

Suhua Liu12 Hongbo Su3 Renhua Zhang1 Jing Tian14 and Weizhen Wang5

1Key Laboratory of Water Cycle and Related Land Surface Processes Institute of Geographic Sciences and Natural Resources ResearchChinese Academy of Sciences Beijing 100101 China2University of Chinese Academy of Sciences Beijing 100049 China3Department of Civil Environmental and Geomatics Engineering Florida Atlantic University Florida FL 33431 USA4State Key Laboratory of Remote Sensing Science Institute of Remote Sensing and Digital Earth Chinese Academy of SciencesBeijing 100101 China5Heihe Remote Sensing Experimental Research Station Cold and Arid Regions Environmental and Engineering Research InstituteChinese Academy of Sciences Lanzhou 730000 China

Correspondence should be addressed to Hongbo Su hongboieeeorg

Received 2 February 2016 Accepted 29 May 2016

Academic Editor Philippe Ricaud

Copyright copy 2016 Suhua Liu et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

To estimate the surface air temperature by remote sensing the advection-energy balance for the surface air temperature (ADEBAT)model is developed which assumes the surface air temperature is driven by the local driving force and the advective drivingforce The local driving force produces a local surface air temperature whereas the advective driving force changes it by addingan exotic air temperature An advection factor 119891 is defined to measure the quantity of the exotic air brought by the advection Sincethe 119891 is determined by the advection this paper improves it to a regional scale by using the Inverse Distance Weighting (IDW)method whereas the original ADEBAT model uses a constant of 119891 for a block of area Results retrieved by the improved ADEBAT(IADEBAT) model are evaluated and comparison was made with the in situ measurements with an 1198772 (correlation coefficient) of077 an RMSE (Root Mean Square Error) of 031 K and a MAE (Mean Absolute Error) of 024K The evaluation shows that theIADEBAT model has higher accuracy than the original ADEBAT model Evaluations together with a 119905-test of the MAD (MeanAbsolute Deviation) reveal that the IADEBAT model has a significant improvement

1 Introduction

Air temperature is a basic variable that is normally measuredat the height of 2 or 15m by meteorological stations Itis not only the primary item of weather forecast but alsouseful for describing the climate change and energy exchangebetween the earth surface and the atmosphere [1 2] Airtemperature also plays an essential role in physical processesof evapotranspiration photosynthesis and heat transfer [3]As a result many land surface process models includingthose in climatology hydrology and ecology require airtemperature as an input variable [4]

Up to now air temperature has traditionally but mainlybeen obtained by meteorological stations However thesemeteorological stations are deployed in a limited number ofplaces and thus can only provide information representingthe specific local area and often lack a broad enough represen-tation for the regional areas [5] In addition meteorologicalstations are with a low spatial density that usually cannotsatisfy the needs either in scientific research or in practicalapplications [6] In order to extend the air temperature froma point scale to a regional scale many spatial interpola-tion methods [7] for example inverse distance weighting(IDW) spline functionmethod and the kriging interpolation

Hindawi Publishing CorporationAdvances in MeteorologyVolume 2016 Article ID 4294219 11 pageshttpdxdoiorg10115520164294219

2 Advances in Meteorology

method have been used However interpolation resultsusually cannot reflect the detailed spatial variability andthese spatial interpolation methods would produce largeerrors especially for the heterogeneous underlying surfaces[8 9] Hence the lack of sufficient spatial density and lackof enough spatial representativeness in the ground-basedair temperature would lead to an inaccurate estimation at aregional scale [1] Benefiting from the fast development ofremote sensing techniques spatially distributed informationon the underlying surface can be obtained Remote sensingtechniques provide a straightforward and consistent way toestimate air temperature at a regional scale with more detailsthan meteorological data Many studies attempted to retrievenear surface air temperature by thermal infrared remotesensing data [10 11] Methods for acquiring the surfaceair temperature include the temperature-vegetation indexapproaches (TVX) the statistical approaches the neuralnetwork approaches and the energy balance approaches [310 12 13]

The TVX method is a widely used approach based onthe correlation between the normalized difference vegetationindex (NDVI) and land surface temperature (LST) whichform a trapezoid space [14] This method assumes that thenear surface air temperature is approximately equal to thecanopy temperature when the vegetation is dense [1 14]Actually vegetation is with an uneven spatial distributionand then the air temperature under the condition of partialvegetation can be estimated by extrapolating the air temper-ature of full coverage [15] A unique advantage of the TVXmethod is that it only needs the LST and NDVI data ratherthan ground-based observations As a result the specialfeature makes it possible for researchers to use numeroustypes of data such as NOAAAVHRR [16] EOSMODIS [17]and LandsatETM+ [18] Although the TVXmethod is widelyused it has limitations on sparse vegetation regions wheregreat uncertainty is found [19]

According to many studies there is a high correla-tion between the air temperature and LST and statisticalapproaches are based upon this relationship [20] Throughlinear regression analysis the method first establishes anexperimental equation between air temperatures observed bymeteorological stations and LST or brightness temperatureof corresponding pixels then it applies the experimentalequation to estimate air temperature in the whole studyregion Using data from geostationary satellites Chen et aldeveloped a model for simulating surface air temperature atnight [21] Validation results showed a high correlation coef-ficient of 087 and a standard deviation of 157 K Jones et alusedMODIS data to build quantitative relationships betweenLST and surface air temperature for different geomorphologictypes and then applied the quantitative relationships to derivesurface air temperature [22] Results showed that correlationcoefficients were from 057 to 081 and the RMSE were below074K Kawashima et al estimated surface air temperatureon the basis of LST derived from Landsat TM and resultsshowed that the standard error was within 185 K [12]Statistical approaches are simple in principle and convenientto use however themodeling highly depends on the time andlocation of the data caption

Neural network approaches use plenty of neurons linkedto each other to simulate any complicated and nonlinearrelationships Without knowing the physical mechanismamong surface air temperature brightness temperature andland surface properties neural network approaches establishrelationship between surface air temperature and the inputvariables by training data alone Using the AVHRR imagesurface elevation and solar zenith angle together with Julianday as neurons Jang et al applied neural network approachesto estimate surface air temperature and the RMSE of esti-mates was 179∘C [10] Zhao et al used the albedo NDVIDEM and LST as inputs to the neural network approachesto calculate the daily averaged surface air temperature andthe daily maximum and the daily minimum air temperatureand the results indicated that the RMSEwas about 09∘C [23]Although the method has obvious advantages in expressingthe nonlinear relationship between surface air temperatureand other variables it is limited for inputting a large amountof data [11 24]

Energy and mass exchanging is happening between theland surface and the atmosphere Without considering theenergy transported by horizontal advection and consumedby photosynthesis the exchange of energy can be describedby the energy balance equation

119877119899= 119867 + 119871119864 + 119866 (1)

where 119877119899is the net radiation 119867 is the sensible heat flux 119871119864

is the latent heat flux and 119866 is the soil heat flux units of thefour items areWm2 Both the sensible heat flux and the latentheat flux can be regarded as functions of LST and surfaceair temperature and soil heat flux can be expressed by thenet radiation Through formula derivation the expression ofsurface air temperature which contains LST and other landsurface variables can be obtained [13] Using observationsat weather stations surface vegetation and geomorphologyinformation as inputs to the energy balance equation Papeand Loffler estimated surface air temperature in the alpineareas [25] The RMSE of the estimates in their study wasbetween 037 and 102∘C Zaksek and Schroedter-Homscheidtapplied the energy balance equation to estimate surface airtemperature [11] It turned out that 88 of the estimates hadabsolute deviations within 3∘C The energy balance equationis a physical method with a good portability However itneeds a large number of variables and some of the variablessuch as aerodynamic resistance and surface roughness aredifficult to acquire by remote sensing In addition thepresence of advection always characterizes exchanges in nearsurface layer and contributes to the uncertainty of energyinfluxes especially heat storage change that is to say theadvection would affect the energy balance equation andshould be discussed as one of the reasons of energy balanceimbalance That means item (119877

119899) on the left side of (1) is not

equal to items (119867 + 119871119864 + 119866) on the right side of (1) [26 27]In conclusion approaches to estimate surface air temper-

ature from remote sensing are usually based on statisticalempirical and energy balance models Su et al proposedan algorithm to expand the surface air temperature toa larger spatial domain which is based on the fact that

Advances in Meteorology 3

the variation of surface air temperature is controlled jointlyby the local turbulence of radiant driving force and thehorizontal advection of the advective driving force [2] Zhanget al developed a model called advection-energy balance forair temperature (ADEBAT) model which is based on thetheory put forward by Su et al [2] The ADEBAT modelassumes that surface air temperature is composed of twoparts the local air temperature and the exotic air temperature[28] The former is dominated by the radiation effect andmainly indicates the heating effect of longwave radiationon near-surface atmosphere and this part can be describedby the energy balance equation The latter is caused by theturbulent flow exchange including the turbulent diffusion andexchange of the horizontal advection Regarding the latterpart an advection factor 119891 is defined to measure the quantityof the exotic air According to the ADEBAT model Zhang etal retrieved the surface air temperature in north China withan 1198772 higher than 077 and an RMSE lower than 042K [28]

The objective of the paper is to improve the ADEBATmodel in expanding the advection factor 119891 In the originalADEBAT model 119891 would be shown as blocks with constantvalues which is not correct because the advection also hasspatial heterogeneity like other geographical elements Theadvection factor 119891 is affected by the intensity of the advectivedriving force and will make influence on other places bydiffusion Hence the spatial distribution of 119891 is interpolatedby the IDWmethod in the IADEBAT model

The outline of this paper is as follows Section 2 givesa brief description of the ADEBAT model and presentsthe improvement on 119891 of the IADEBAT model Section 3describes the study area and the data Results of the surfaceair temperature estimation are presented and analyzed inSection 4 The conclusion and discussion are displayed inSections 5 and 6 respectively

2 Methodology

21 The ADEBAT Model In this section the advection-energy balance for air temperature (ADEBAT) model isbriefly introduced The basic idea of the ADEBAT model[28] is as follows The surface air temperature is determinedby two physical processes the heating effect produced bythe longwave radiation from land surface and the advectiveeffect resulting from the turbulent flow exchange The twoeffects are also called the local driving force and the advec-tive driving force respectively For the former surface airtemperature increases with the accumulation of absorbingsurface longwave radiation For the latter the advectiondisturbs the surface air temperature by the turbulent diffusionexchange which develops between land surface and near-surface atmosphere

Assuming there is one cubic meter of air the temperatureof it is dominated only by the heating effect from land surfacewhich corresponds to a namely surface energy balanceclosure In this case the surface air temperature 119879

119886is equal

to the local air temperature 119879loc However the energy isimbalanced for the advective effect in most situations [29]The exotic air comes into the one cubic meter and mixesup with the local air in it If the air temperature driven by

the advective effect due to the exotic energy is defined as119879exothen the actual air temperature 119879

119886is a mixture of 119879loc and

119879exo In order to retrieve the surface air temperature variablesrelated to the two physical processes need to be determined

211 Obtaining the Surface Air Temperature 119879119886 Two situ-

ations are treated differently when calculating 119879119886 energy

balance closure and energy balance misclosureIn the case of energy balance closure namely the surface

air temperature is controlled by local driving force onlyThe exchange of energy is described by the energy balanceequation (1)The sensible heat flux (see (2)) and the definitionof the Bowen ratio 120573 (see (3)) are as follows

119867 =120588119862119901

119903119886

(1198790minus 119879119886) (2)

120573 =119867

119871119864 (3)

where 120588119862119901is the volumetric heat capacity of air 119903

119886is the air

aerodynamic resistance under the condition of no wind andno advection and 119879

0is aerodynamic temperature which is

usually substituted by LST in applications [30 31]By combining (1) (2) and (3) the surface air temperature119879119886can be derived as

119879119886= 1198790minus120573 (119877119899minus 119866)

120573 + 1

119903119886

120588119862119901

(4)

Because119879119886is equal to the local air temperature119879loc under

the condition of energy balance closure hence 119879119886can be

replaced by 119879loc in (4) and we attain 119879loc as follows

119879loc = 1198790 minus120573 (119877119899minus 119866)

120573 + 1

119903119886

120588119862119901

(5)

In most cases the surface energy balance of the nearsurface air is not closed due to the advective driving forceLike windy days the effect of local driving force on the airtemperature becomes less dominant at the same time theadvective driving force would be more dominant Both thelocal driving force and the advective driving force have animpact on the 119879

119886 and 119879

119886is mixed up by 119879loc and 119879exo

Provided that the mixing satisfies the linear mixed theory 119879119886

can be calculated as

119879119886= 119891119879exo + (1 minus 119891)119879loc (6)

119891 =119881exo119881119886

(7)

1 minus 119891 =119881loc119881119886

(8)

where 119879119886is the real air temperature and 119891 is the volume ratio

that is also known as the advection factor119881119886is the air volume

that equals the sum of 119881exo and 119881loc 119881exo and 119881loc are thevolumes of the exotic air and the local air after the mixingrespectively Because 119891 is the proportion of mixture it rangesfrom 0 to 1 When 119891 = 0 it means there is no exotic air that


corresponds to the situation of energy balance closure and119879119886equals 119879loc If 119891 = 1 it indicates that the exotic air entirely

replaces the local air In practice the vast majority of cases is0 lt 119891 lt 1

212 Obtaining the Proportion of Mixture 119891 In order tocalculate 119879

119886 two other variables 119891 and 119879exo need to be

acquired (as in (6)) with the aid of air temperature windspeed and wind direction observed at meteorological sta-tions Assuming the subscript 119894 represents any pixel in thestudy area for 119891

(119894)and 119879exo(119894) to be calculated we firstly

identify two nearest meteorological stations which have thesimilar wind speed andwind direction because it is supposedthat similar wind speed and wind direction supply similaradvection The two pixels covering the two meteorologicalstations are expressed as Pixel 1 and Pixel 2 thenwe obtain therelationship below 119891

(2)= 119891(1)= 119891(119894) 119879exo(2) = 119879exo(1) = 119879exo(119894)

[28] Based on (6) two new relationships can be derived

119879119886(1)= 119891(1)119879exo(1) + (1 minus 119891(1)) 119879loc(1)

119879119886(2)= 119891(2)119879exo(2) + (1 minus 119891(2)) 119879loc(2)

(9)

where 119879119886(1)

and 119879119886(2)

are air temperatures of the two weatherstations respectively 119879loc(1) and 119879loc(2) are local air tempera-tures driven by the local driving force respectively and canbe calculated by (4) and (5) then 119891 and 119879exo of Pixel 119894 can beobtained by solving (9)

119891(119894)= 119891(1)= 119891(2)= 1 minus119879119886(1)minus 119879119886(2)

119879loc(1) minus 119879loc(2)

119879exo(119894) = 119879exo(1) = 119879exo(2)

=(119879119886(1)+ 119879119886(2)) minus (1 minus 119891

(119894)) (119879loc(1) + 119879loc(2))

119891(119894)

(10)

After solving 119891(119894)

and 119879exo(119894) by (10) at every pixel 119879119886at

every pixel can be obtained by (6)

22 Improvement on the ADEBAT Model According toformula derivation in Section 212 we know that everyinputtedmeteorological stationwould get an advection factorby solving (10) For two nearest weather stations with similarwind speed and similar wind direction that is with similaradvection pixels around them would use the same advectionfactor so the result of the advection factor would be shown asblocks with constant value which are stations-centered Thisis how to get 119891 in the ADEBATmodel However the result of119891 acquired by the ADEBATmodel is unreasonable for spatialheterogeneity Hence it is necessary to make improvementon obtaining the spatial distribution of 119891 The IADEBATmodel uses the IDW method to expand the ground-based119891 from a point scale to a regional scale to obtain the spatialdistribution of 119891 The reason for applying the IDW methodcan be described as follows

The advection factor 119891 is affected by the intensity of theadvective driving force which goes to other places by diffu-sion As a result to the advective driving force of a certain

Site-based f f shown asgradient

f shown asblocks Equation (10) IDW

ADEBAT IADEBAT

Figure 1The sketch on difference between the ADEBATmodel andthe IADEBAT model

station with increasing the distance from the station theadvection factor is becoming less affected by it Since theIDWmethod interpolates according to the distance betweentwo inputted objectives it is appropriate to obtain the spatialdistribution of the advection factor by the IDW methodFormulas can be written as

119891(119894)=119899

sum119909=1

119908(119909)119891(119909)

119908(119909)=1119889119909

119896

sum119899

119909=11119889119909

119896

(11)

where 119891(119894)is the advection factor of any pixel to be calculated

119909 is the number of the inputted meteorological station119891(119909)

is the advection factor of the station number 119909 119908(119909)

is the weight 119889119909is the distance between Pixel 119894 and the

station number 119909 and 119896 is the specified power exponentDifferent from the ADEBAT model the IADEBAT modelwould get gradual distribution results of 119891 Figure 1 gives thesketch on difference betweenmodels of the ADEBAT and theIADEBAT

3 Study Area and Data

31 Study Area The study area is located in Zhangye GansuChina and ranges from 3883∘N to 3893∘N in latitude andfrom 10032∘E to 10042∘E in longitude (Figure 2) which isin the core oasis of middle reaches of the Heihe River Majorfeatures of the climate here are dry and rainless The annualmean temperature is 70∘C and the annual mean precipitationis 1249mmThe potential evaporation more than 2000mmis much higher than the precipitation The land cover in thestudy area mainly consists of cropland cultivated with seedcorn vegetables apple orchards and windbreak forests aswell as with some built-up areas (villages) scattered through-out From May to September seed corn is the staple crop inthe study area and there are also small areas of vegetablesand fruit trees Since it is arid almost all the farmlands areirrigated with the water from the Heihe River and the studyarea belongs to irrigation districts with alternating irrigationregimen

32 Remote Sensing Data The ASTER images and HJ-1ABimages (the small satellite constellation for environment anddisaster monitoring and forecasting) were utilized in thestudy for acquiring variables including surface albedo 120572surface emissivity 120576

119904 vegetation fraction119891V and LSTThe four

variables were aiming to calculate the net radiation 119877119899 soil

heat flux 119866 and the local air temperature 119879loc The ASTERis a sensor collecting multispectral visible near-infraredand thermal infrared images and it is a special remotesensing tool intended tomonitor land surface energy balance


1

17

12

11

1098

7

6

54

3

2

16

1514

13

N

0 920 1840460(Meters)

38∘ 54998400 0

998400998400N

38∘ 53998400 0

998400998400N

38∘ 52998400 0

998400998400N

38∘ 51998400 0

998400998400N

38∘ 53998400 30998400998400

N38∘ 52998400 30998400998400

N38∘ 51998400 30998400998400

N38∘ 50998400 30998400998400

N10

0∘ 24998400 30998400998400

E

100∘ 23998400 30998400998400

E

100∘ 22998400 30998400998400

E

100∘ 21998400 30998400998400

E

100∘ 20998400 30998400998400

E

100∘ 19998400 30998400998400

E

100∘249984000998400998400E100∘239984000998400998400E100∘229984000998400998400E100∘219984000998400998400E100∘209984000998400998400E

MaizeVillageOrchard

VegetableWoodlandMeteorological stations

Gansu Province

Figure 2 Land use status in the study area and spatial distribution of meteorological stations

hydrological processes and climate [32] From the ASTERimages the vegetation fraction 119891V surface emissivity 120576

119904 and

LST (based on method presented by [33]) were obtainedThesurface albedo 120572 was from products provided by [34] andthe albedo products were calculated from images acquiredfrom HJ-1AB which was launched by China in 2008 withtwo CCDs carried by each satellite and every 3 or 4 days thesatellite could revisit the same place

33Meteorological Data Meteorological data is supported bythe experiment namedHiWATER which is a watershed scaleeco-hydrological experiment designed from an interdisci-plinary perspective to address problems that include hetero-geneity scaling uncertainty and closing of the water cycle atthe watershed scale [35 36] The experiment was performedin the Heihe River Basin which is in the arid region of north-west China The data now is shared online and users canget data by submitting application to Cold and Arid RegionsScience Data at Lanzhou (httpwestdcwestgisaccn)

There are 17 Automatic Weather Stations (AWS) in thestudy area as shown in Figure 2Theywere all with automaticdata acquisition systems and sample intervals of the datawere 1 and 10 minutes Air temperature humidity windspeed wind direction air pressure land surface temperature

Table 1 General descriptions of the variables and sources

Variables SourcesLST at satellite passing time ASTERAlbedo HJ-1AB119891V ASTER120576119904

ASTERSolar radiation Meteorological stations

and solar radiation were observed In order to estimate 119891and 119879exo six meteorological stations selected randomly wereutilized as inputs and the other stations were used to validatethe estimations of air temperature Variables and sourcesapplied in the study are listed in Table 1

Two clear sky days chosen to conduct the study are June24th and July 10th 2012 To evaluate the results retrieved bythe IADEBATmodel andmake an objective assessment on itestimates by the other twomethods the ADEBATmodel andthe IDWmethod were provided for comparing

4 Results

By using remote sensing data and meteorological data asinputs to the three methods the spatial distributions of


Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302302-303

(a)

Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302Above 302

(b)

Air temperature (K)

N

0 1050 2100(Meters)

2975ndash298298ndash29852985ndash299

(c)

Figure 3 Air temperature retrieved by the IADEBAT model (a) the ADEBAT model (b) and the IDWmethod (c) on June 24th 2012

surface air temperature were finally retrieved Figure 3 showsthe inverted results based on the IADEBAT model (a) theADEBAT model (b) and the IDW method (c) on June 24th2012 Similarly estimates on July 10th 2012 are shown inFigure 4

41 Analysis on Results Interpolated by the IDW MethodInterpolations by the IDWmethodwere without detailed tex-tures and the estimated values are between weather stationswith high air temperature and those with low temperatureOtherwise the high temperature and low temperature occur


N

0 1050 2100(Meters)

Air temperature (K)296-297297-298298-299299-300

300-301301-302302-303303-304

(a)

N

Air temperature (K)

0 1050 2100(Meters)

296-297295-296

297-298298-299

299-300300-301301-302Above 302

(b)

N

Air temperature (K)

0 1050 2100(Meters)

2985ndash299299ndash29952995ndash300300ndash3005

(c)

Figure 4 Air temperature retrieved by the IADEBAT model (a) the ADEBAT model (b) and the IDWmethod (c) on July 10th 2012

in the form of circular high temperature area and lowtemperature area and this spatial pattern displayed by theIDW method does not exist actually The IDW method justconsiders the advective driving force but ignores the localdriving force so results interpolated by the IDWmethod failto describe the true distribution of surface air temperature

Taking the red circle area in Figure 2 as an example it wasirrigated in the way of traditional constant flooding from1200 pm June 22nd to 1200 pm June 24th 2012The irrigationwater was snowmelt water about 15∘C fromQilianMountainand transferred by the Heihe River Because the wind speedin the study area on June 24th 2012 was rather low about


Table 2 Air temperatures measured bymeteorological stations and retrieved by the IADEBAT and the ADEBATmodels from three differentland use types on June 24th 2012

Land use Observations IADEBAT (K) ADEBAT (K)Estimations 119879

119886-est minus 119879119886-obs Estimations 119879119886-est minus 119879119886-obs

Village 2997 3020 23 3023 26Maize 2985 2985 0 2970 minus15Orchard 2989 2998 08 3000 11Note 119879119886-est means air temperature is estimated by the model 119879119886-obs means air temperature is observed by meteorological stations

Table 3 Air temperatures measured bymeteorological stations and retrieved by the IADEBAT and the ADEBATmodels from three differentland use types on July 10th 2012



Village 2993 2995 02 3021 28Maize 2989 2989 0 3001 12Orchard 2999 2998 minus01 3012 13

14ms which means the advective driving force was weakbut the local driving force was in dominant position therewould be a low temperature area surrounding the red circle(as shown in Figure 3(a)) However the IDWmethod cannotrepresent this information because the area is between twointerpolatedweather stationswith relative high temperatures

42 Analysis on Results Retrieved by the IADEBAT andthe ADEBAT Models Comparing to the IDW method theIADEBAT and the ADEBATmodels take into account effectsincluding not only the advection driving force but also thelocal driving force on near surface air It is obvious thatsurface air temperature retrieved by both models providesa wealth of information in detailed features In additionthe IADEBAT and the ADEBAT models can express lowtemperature regions such as water area and can displayhigh temperature regions such as built-up areas (villages)exactly Referring to low air temperature region the irrigatedarea on June 24th 2012 can be used as an example Theair temperature of the irrigated area must be lower thansurrounding regions (as shown in Figures 3(a) and 3(b))because the area had been irrigated for more than two daysby low temperature water and was mainly controlled by localdriving force About high temperature regions built-up areas(villages) were taken as examples On both two days built-upvillages showed high air temperature

43 Comparison between the IADEBAT and the ADE-BAT Models The comparisons were made between resultsacquired from the IADEBAT and the ADEBAT modelsIn order to assess the accuracy of the two methods threedifferent land use types were chosen to analyze Tables 2 and3 give the observations and estimations of air temperatureacquired by meteorological stations and the two methodsFromTables 2 and 3 we know that the estimation errors of theIADEBATmodel are much lower than those of the ADEBATmodel Hence results obtained by the IADEBAT model aremore accurate

As stated above we know that the red circle was irrigatedfrom 1200 pm June 22nd to 1200 pm June 24th 2012 sothe area around the red circle must have low air temperatureAccording to alternating irrigation schedule provided by theWater Authority the irrigation area was largely in line withthe low air temperature area obtained by the IADEBATmodel(as shown in Figure 3(a)) However the low air temperaturearea shown in Figure 3(b) was greatly overestimated by theADEBAT model this is because the advection factor 119891 wascontrolled by the intensity of the advective driving force andthe 119891 in the study area should be gradual However constant119891 was used by the ADEBAT model which is displayed asblocks of the same 119891 hence the 119891 in the ADEBAT modelmay be overestimated or underestimated for the reason thatthe 119891 has spatial heterogeneity and cannot remain the same

44 Validation In order to validate the results comparisonswere made among results acquired by the three methods asshown in Figure 5 The correlation coefficient 1198772 betweenestimations by the IADEBATmodel and in situ observations077 is higher than that between estimations by the ADE-BAT model and corresponding measurements 067 whichillustrates that results of the IADEBAT model have bettercorrelation with observations that is to say estimates bythe IADEBAT model are closer to the 1 1 line The RootMean Square Error (RMSE) andMeanAbsolute Error (MAE)of the IADEBAT model 031 K and 027K respectively arealso lower than those of the ADEBAT model 043 K and034K respectively which indicates that the accuracy ofcalculations from the IADEBAT model is better than thatfrom the ADEBAT model Obviously results derived by theIDW method with a correlation coefficient 1198772 of 017 anRMSE of 056K and a MAE of 051 K are much worse thanby the IADEBAT model

Because statistical indicators the RMSE of the IADEBATand of the ADEBAT models are both below 05 K we donot know whether there is a significant difference betweenthe two models Hence it is necessary to use a 119905-test (Paired


Table 4 Paired samples test between MADs of the IADEBAT and of the ADEBAT models

Paired difference

119905 df Sig (2-tailed)Mean Stddeviation

Std errormean

95 confidence intervalof the difference

Lower UpperMAD IADEBAT ndashMAD ADEBAT minus013947 032259 006585 minus027568 minus000325 minus2118 23 0043

299 300298Observations (K)

298

299

300

Estim

atio

ns (K

)

Y = 116X minus 474

R2 = 077MAE = 024 KRMSE = 031KPoints 24

(a)


298

299

300

Estim

atio

ns (K

)

Y = 109X minus 2996R2 = 067MAE = 034KRMSE = 043KPoints 24

(b)

298

299

Estim

atio

ns (K

)


Y = 032X + 2015

R2 = 017MAE = 051KRMSE = 056KPoints 24

(c)

Figure 5 Comparison of air temperature derived from the IADEBAT model (a) the ADEBAT model (b) and the IDWmethod (c)

Samples Test) Table 4 gives the 119905-test result between MADsof the two models A 5 significance level is set as criteria tojudge if MAD of the IADEBAT model is significantly lowerthan that of the ADEBAT model The Paired Samples Test inTable 4 shows that 119901 = 0043 which is lower than 005 As

a result we can draw a conclusion that results obtained bythe IADEBAT model have significant lower MAD than thatcalculated by the ADEBAT model Because the IADEBATmodel also has better correlation indicators (1198772 RMSE andMAE) than the ADEBAT model the improved IADEBAT


model can acquire better estimation of air temperature andthe improvement on 119891 is feasible

5 Discussion

The ADEBAT model divides the surface air temperature intotwo parts that can be obtained by energy balance equationand by meteorological observations respectively The idearelies on a clear physical mechanism and is completely differ-ent from traditionally interpolation methods such as IDWwhich just interpolates according to geostatistics principleand is different from other remote sensing methods such asthe TVX and the statistical approach which always neglectthe formation mechanism of air temperature and considerthe air temperature as a whole and are based on empiricalrelationship As a result benefiting from the clear physicalmeaning the IADEBAT model has good portability andgeneral applicability and also offers thought for acquiringthe spatial distribution of air temperature by remote sensingFurthermore the IADEBAT model can acquire the spatialdistribution of surface air temperature with a more wealth ofdetailed textures than traditional geostatistics interpolationmethods Comparing the IADEBAT model with the energybalance method it also has its merit on the specific airaerodynamic resistance under the condition of no wind andno advection which is an exact value and can be calculatedthrough the laboratory experiment However the air aero-dynamic resistance used in the energy balance method is inpractice difficult to be determined Hence the exact valueof air aerodynamic resistance used in the IADEBAT modelwould avoid complex computing process

However the IADEBAT model has limitations when anobjective evaluation is made because it is highly dependingon numerous ground-based observations If a place has noadequate observations then the model cannot be appliedsuccessfully and therefore the application of the IADEBATmodel is selectiveThe amount of ground-based observationsused as input or as validation is depending on the size ofthe study area the requested accuracy and the resolutionof the remote sensed images Furthermore the IADEBATmodel is based on the remote sensing data and hence ithas the generic defects of remote sensing methods TheIADEBAT model cannot be applied on cloudy or rainy daysIn the future it is looking forward to develop interpolationtechniques which can take into account both the temporalvariation and the spatial variation of surface air temperatureto acquire continuous results

6 Conclusion

Remote sensing is a promising tool to get information onthe underlying surface and the IADEBAT model used in thestudy is a remote sensing method which is based on the for-mation mechanism of air temperature It takes into accountthe local driving force that produces local air temperatureand the advective driving force that brings the exotic airtemperatureThe advection factor119891 is defined tomeasure thequantity of the advective driving force In order to increasethe accuracy of the retrieved air temperature the study

makes improvement on the advection factor 119891 of the IADE-BAT model which expanded the advection factor 119891 to aregional scale by the IDWmethodThe IADEBATmodel wasvalidated by using observed data and was compared with theADEBAT model and the IDW method Verification showedestimations acquired by the IADEBAT model with an 1198772of 077 an RMSE of 031 K and a MAE of 024K whichhad the highest accuracy than estimations acquired by theADEBAT model and the IDW method that had an 1198772 of067 an RMSE of 043 K and a MAE of 034K and an 1198772 of017 an RMSE of 056K and a MAE of 051 K respectivelyAs for the estimation errors we made analyses on threedifferent land use types Results indicated that the IADEBATmodel had lower estimation errors than the ADEBATmodelthat is to say the IADEBAT model has better accuracy thanthe ADEBAT model on heterogeneous underlying surfacesA 119905-test was made between MADs of results obtained bythe IADEBAT and the ADEBAT models The 119901 (0043)proved that the improvement on the advection factor 119891 wassignificant As a result the improvement proposed in thestudy is feasible and reasonable

Competing Interests

The authors declare no conflict of interests

Acknowledgments

This work is supported by the National Natural ScienceFoundation of China (Grant nos 41571356 41271380 and41171286) and the National Basic Research Program of China(Grant no 2013CB733406)

References

[1] L Prihodko and S N Goward ldquoEstimation of air temperaturefrom remotely sensed surface observationsrdquo Remote Sensing ofEnvironment vol 60 no 3 pp 335ndash346 1997

[2] H Su J Tian R Zhang et al ldquoA physically based spatialexpansion algorithm for surface air temperature and humidityrdquoAdvances in Meteorology vol 2013 Article ID 727546 8 pages2013

[3] V Lakshmi K Czajkowski R Dubayah and J Susskind ldquoLandsurface air temperature mapping using TOVS and AVHRRrdquoInternational Journal of Remote Sensing vol 22 no 4 pp 643ndash662 2001

[4] T A Huld M Suri E D Dunlop and F Micale ldquoEstimatingaverage daytime and daily temperature profiles within EuroperdquoEnvironmental Modelling and Software vol 21 no 12 pp 1650ndash1661 2006

[5] A D Hartkamp K De Beurs A Stein and J W White Inter-polation Techniques for Climate Variables CIMMYT 1999

[6] J V Vogt A A Viau and F Paquet ldquoMapping regional airtemperature fields using satellite-derived surface skin temper-aturesrdquo International Journal of Climatology vol 17 no 14 pp1559ndash1579 1997

[7] M Ninyerola X Pons and J M Roure ldquoA methodologicalapproach of climatological modelling of air temperature andprecipitation through GIS techniquesrdquo International Journal ofClimatology vol 20 no 14 pp 1823ndash1841 2000


[8] J Li and A D Heap ldquoA review of comparative studies of spatialinterpolation methods in environmental sciences performanceand impact factorsrdquo Ecological Informatics vol 6 no 3-4 pp228ndash241 2011

[9] C-D Xu J-F Wang M-G Hu and Q-X Li ldquoInterpolationof missing temperature data at meteorological stations usingP-BSHADErdquo Journal of Climate vol 26 no 19 pp 7452ndash74632013

[10] J D Jang A A Viau and F Anctil ldquoNeural network estimationof air temperatures from AVHRR datardquo International Journal ofRemote Sensing vol 25 no 21 pp 4541ndash4554 2004

[11] K Zaksek and M Schroedter-Homscheidt ldquoParameterizationof air temperature in high temporal and spatial resolution froma combination of the SEVIRI and MODIS instrumentsrdquo ISPRSJournal of Photogrammetry and Remote Sensing vol 64 no 4pp 414ndash421 2009

[12] S Kawashima T IshidaMMinomura and TMiwa ldquoRelationsbetween surface temperature and air temperature on a localscale during winter nightsrdquo Journal of Applied Meteorology vol39 no 9 pp 1570ndash1579 2000

[13] Y-J Sun J-FWang R-H Zhang R R Gillies Y Xue andY-CBo ldquoAir temperature retrieval from remote sensing data basedon thermodynamicsrdquoTheoretical and Applied Climatology vol80 no 1 pp 37ndash48 2005

[14] R R Nemani and S W Running ldquoEstimation of regional sur-face resistance to evapotranspiration from NDVI and thermal-IR AVHRR datardquo Journal of Applied Meteorology vol 28 no 4pp 276ndash284 1989

[15] S D Prince S J Goetz RODubayah K P Czajkowski andMThawley ldquoInference of surface and air temperature atmosphericprecipitable water and vapor pressure deficit using advancedvery high-resolution radiometer satellite observations compar-ison with field observationsrdquo Journal of Hydrology vol 212-213no 1ndash4 pp 230ndash249 1998

[16] S J Goetz ldquoMulti-sensor analysis ofNDVI surface temperatureand biophysical variables at a mixed grassland siterdquo Interna-tional Journal of Remote Sensing vol 18 no 1 pp 71ndash94 1997

[17] H Yan J Zhang Y Hou and Y He ldquoEstimation of air temp-erature from MODIS data in east Chinardquo International Journalof Remote Sensing vol 30 no 23 pp 6261ndash6275 2009

[18] C Wloczyk E Borg R Richter and K Miegel ldquoEstimationof instantaneous air temperature above vegetation and soilsurfaces from Landsat 7 ETM+ data in northern GermanyrdquoInternational Journal of Remote Sensing vol 32 no 24 pp 9119ndash9136 2011

[19] K P Czajkowski T Mulhern S N Goward J Cihlar RO Dubayah and S D Prince ldquoBiospheric environmentalmonitoring at BOREAS with AVHRR observationsrdquo Journal ofGeophysical Research Atmospheres vol 102 no 24 pp 29651ndash29662 1997

[20] A Benali A C Carvalho J P Nunes N Carvalhais and ASantos ldquoEstimating air surface temperature in Portugal usingMODIS LST datardquo Remote Sensing of Environment vol 124 pp108ndash121 2012

[21] E Chen L H Allen Jr J F Bartholic and J F GerberldquoComparison of winter-nocturnal geostationary satellite infra-red-surface temperature with shelter-height temperature inFloridardquo Remote Sensing of Environment vol 13 no 4 pp 313ndash327 1983

[22] P Jones G Jedlovec R Suggs and S Haines ldquoUsing MODISLST to estimate minimum air temperatures at nightrdquo in Pro-ceedings of the 13th Conference on Satellite Meteorology andOceanography pp 13ndash18 Norfolk Va USA September 2004

[23] D Zhao W Zhang and X Shijin ldquoA neural network algorithmto retrieve near-surface air temperature from landsat ETM+imagery over the Hanjiang River Basin Chinardquo in Proceedingsof the IEEE International Geoscience and Remote Sensing Sympo-sium (IGARSS rsquo07) pp 1705ndash1708 Barcelona Spain June 2007

[24] S Stisen I Sandholt A Noslashrgaard R Fensholt and L EklundhldquoEstimation of diurnal air temperature usingMSG SEVIRI datain West Africardquo Remote Sensing of Environment vol 110 no 2pp 262ndash274 2007

[25] R Pape and J Loffler ldquoModelling spatio-temporal near-surfacetemperature variation in high mountain landscapesrdquo EcologicalModelling vol 178 no 3-4 pp 483ndash501 2004

[26] R A Spronken-Smith T R Oke and W P Lowry ldquoAdvectionand the surface energy balance across an irrigated urban parkrdquoInternational Journal of Climatology vol 20 no 9 pp 1033ndash1047 2000

[27] VMasson C S B Grimmond and T R Oke ldquoEvaluation of thetown energy balance (TEB) scheme with direct measurementsfrom dry districts in two citiesrdquo American MeteorologicalSociety no 2000 pp 1011ndash1026 2002

[28] R Zhang Y Rong J Tian H Su Z-L Li and S Liu ldquoAremote sensing method for estimating surface air temperatureand surface vapor pressure on a regional Scalerdquo Remote Sensingvol 7 no 5 pp 6005ndash6025 2015

[29] K Wilson A Goldstein E Falge et al ldquoEnergy balance closureat FLUXNET sitesrdquoAgricultural and ForestMeteorology vol 113no 1ndash4 pp 223ndash243 2002

[30] W G M Bastiaanssen M Menenti R A Feddes and A A MHoltslag ldquoA remote sensing surface energy balance algorithmfor land (SEBAL) 1 Formulationrdquo Journal of Hydrology vol212-213 no 1ndash4 pp 198ndash212 1998

[31] Z Su ldquoThe Surface Energy Balance System (SEBS) for esti-mation of turbulent heat fluxesrdquo Hydrology and Earth SystemSciences vol 6 no 1 pp 85ndash100 2002

[32] A N French F Jacob M C Anderson et al ldquoSurface energyfluxes with the advanced spaceborne thermal emission andreflection radiometer (ASTER) at the Iowa 2002 SMACEX site(USA)rdquo Remote Sensing of Environment vol 99 no 1-2 pp 55ndash65 2005

[33] Z Qin A Karnieli and P Berliner ldquoAmono-window algorithmfor retrieving land surface temperature from Landsat TMdata and its application to the Israel-Egypt border regionrdquoInternational Journal of Remote Sensing vol 22 no 18 pp 3719ndash3746 2001

[34] L Qiang W Jianguang W Shengbiao and Q Ying AlbedoDataset in 30m-Resolution in the Heihe River Basin in 2012Heihe Plan Science Data Center 2014

[35] X Li G Cheng S Liu et al ldquoHeihe watershed allied teleme-try experimental research (HiWater) scientific objectives andexperimental designrdquo Bulletin of the American MeteorologicalSociety vol 94 no 8 pp 1145ndash1160 2013

[36] H Li D Sun Y Yu et al ldquoEvaluation of the VIIRS andMODISLST products in an arid area of Northwest Chinardquo RemoteSensing of Environment vol 142 pp 111ndash121 2014

Submit your manuscripts athttpwwwhindawicom

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ClimatologyJournal of

EcologyInternational Journal of


EarthquakesJournal of


Hindawi Publishing Corporationhttpwwwhindawicom

Applied ampEnvironmentalSoil Science

Volume 2014

Mining


Journal of

Hindawi Publishing Corporation httpwwwhindawicom Volume 2014

International Journal of

Geophysics

OceanographyInternational Journal of


Journal of Computational Environmental SciencesHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal ofPetroleum Engineering


GeochemistryHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Journal of

Atmospheric SciencesInternational Journal of


OceanographyHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

Advances in


MineralogyInternational Journal of


MeteorologyAdvances in

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

Paleontology JournalHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

ScientificaHindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Geological ResearchJournal of


Geology Advances in


method have been used However interpolation resultsusually cannot reflect the detailed spatial variability andthese spatial interpolation methods would produce largeerrors especially for the heterogeneous underlying surfaces[8 9] Hence the lack of sufficient spatial density and lackof enough spatial representativeness in the ground-basedair temperature would lead to an inaccurate estimation at aregional scale [1] Benefiting from the fast development ofremote sensing techniques spatially distributed informationon the underlying surface can be obtained Remote sensingtechniques provide a straightforward and consistent way toestimate air temperature at a regional scale with more detailsthan meteorological data Many studies attempted to retrievenear surface air temperature by thermal infrared remotesensing data [10 11] Methods for acquiring the surfaceair temperature include the temperature-vegetation indexapproaches (TVX) the statistical approaches the neuralnetwork approaches and the energy balance approaches [310 12 13]

The TVX method is a widely used approach based onthe correlation between the normalized difference vegetationindex (NDVI) and land surface temperature (LST) whichform a trapezoid space [14] This method assumes that thenear surface air temperature is approximately equal to thecanopy temperature when the vegetation is dense [1 14]Actually vegetation is with an uneven spatial distributionand then the air temperature under the condition of partialvegetation can be estimated by extrapolating the air temper-ature of full coverage [15] A unique advantage of the TVXmethod is that it only needs the LST and NDVI data ratherthan ground-based observations As a result the specialfeature makes it possible for researchers to use numeroustypes of data such as NOAAAVHRR [16] EOSMODIS [17]and LandsatETM+ [18] Although the TVXmethod is widelyused it has limitations on sparse vegetation regions wheregreat uncertainty is found [19]

According to many studies there is a high correla-tion between the air temperature and LST and statisticalapproaches are based upon this relationship [20] Throughlinear regression analysis the method first establishes anexperimental equation between air temperatures observed bymeteorological stations and LST or brightness temperatureof corresponding pixels then it applies the experimentalequation to estimate air temperature in the whole studyregion Using data from geostationary satellites Chen et aldeveloped a model for simulating surface air temperature atnight [21] Validation results showed a high correlation coef-ficient of 087 and a standard deviation of 157 K Jones et alusedMODIS data to build quantitative relationships betweenLST and surface air temperature for different geomorphologictypes and then applied the quantitative relationships to derivesurface air temperature [22] Results showed that correlationcoefficients were from 057 to 081 and the RMSE were below074K Kawashima et al estimated surface air temperatureon the basis of LST derived from Landsat TM and resultsshowed that the standard error was within 185 K [12]Statistical approaches are simple in principle and convenientto use however themodeling highly depends on the time andlocation of the data caption

Neural network approaches use plenty of neurons linkedto each other to simulate any complicated and nonlinearrelationships Without knowing the physical mechanismamong surface air temperature brightness temperature andland surface properties neural network approaches establishrelationship between surface air temperature and the inputvariables by training data alone Using the AVHRR imagesurface elevation and solar zenith angle together with Julianday as neurons Jang et al applied neural network approachesto estimate surface air temperature and the RMSE of esti-mates was 179∘C [10] Zhao et al used the albedo NDVIDEM and LST as inputs to the neural network approachesto calculate the daily averaged surface air temperature andthe daily maximum and the daily minimum air temperatureand the results indicated that the RMSEwas about 09∘C [23]Although the method has obvious advantages in expressingthe nonlinear relationship between surface air temperatureand other variables it is limited for inputting a large amountof data [11 24]

Energy and mass exchanging is happening between theland surface and the atmosphere Without considering theenergy transported by horizontal advection and consumedby photosynthesis the exchange of energy can be describedby the energy balance equation

119877119899= 119867 + 119871119864 + 119866 (1)

where 119877119899is the net radiation 119867 is the sensible heat flux 119871119864

is the latent heat flux and 119866 is the soil heat flux units of thefour items areWm2 Both the sensible heat flux and the latentheat flux can be regarded as functions of LST and surfaceair temperature and soil heat flux can be expressed by thenet radiation Through formula derivation the expression ofsurface air temperature which contains LST and other landsurface variables can be obtained [13] Using observationsat weather stations surface vegetation and geomorphologyinformation as inputs to the energy balance equation Papeand Loffler estimated surface air temperature in the alpineareas [25] The RMSE of the estimates in their study wasbetween 037 and 102∘C Zaksek and Schroedter-Homscheidtapplied the energy balance equation to estimate surface airtemperature [11] It turned out that 88 of the estimates hadabsolute deviations within 3∘C The energy balance equationis a physical method with a good portability However itneeds a large number of variables and some of the variablessuch as aerodynamic resistance and surface roughness aredifficult to acquire by remote sensing In addition thepresence of advection always characterizes exchanges in nearsurface layer and contributes to the uncertainty of energyinfluxes especially heat storage change that is to say theadvection would affect the energy balance equation andshould be discussed as one of the reasons of energy balanceimbalance That means item (119877

119899) on the left side of (1) is not

equal to items (119867 + 119871119864 + 119866) on the right side of (1) [26 27]In conclusion approaches to estimate surface air temper-

ature from remote sensing are usually based on statisticalempirical and energy balance models Su et al proposedan algorithm to expand the surface air temperature toa larger spatial domain which is based on the fact that





2 Methodology



119886is equal









119867 =120588119862119901

119903119886

(1198790minus 119879119886) (2)

120573 =119867

119871119864 (3)


119886is the air




119879119886= 1198790minus120573 (119877119899minus 119866)

120573 + 1

119903119886

120588119862119901

(4)




119879loc = 1198790 minus120573 (119877119899minus 119866)

120573 + 1

119903119886

120588119862119901

(5)


119886 and 119879




119879119886= 119891119879exo + (1 minus 119891)119879loc (6)

119891 =119881exo119881119886

(7)

1 minus 119891 =119881loc119881119886

(8)












(2)= 119891(1)= 119891(119894) 119879exo(2) = 119879exo(1) = 119879exo(119894)


119879119886(1)= 119891(1)119879exo(1) + (1 minus 119891(1)) 119879loc(1)

119879119886(2)= 119891(2)119879exo(2) + (1 minus 119891(2)) 119879loc(2)

(9)

where 119879119886(1)

and 119879119886(2)


119891(119894)= 119891(1)= 119891(2)= 1 minus119879119886(1)minus 119879119886(2)

119879loc(1) minus 119879loc(2)

119879exo(119894) = 119879exo(1) = 119879exo(2)

=(119879119886(1)+ 119879119886(2)) minus (1 minus 119891

(119894)) (119879loc(1) + 119879loc(2))

119891(119894)

(10)








ADEBAT IADEBAT



119891(119894)=119899

sum119909=1

119908(119909)119891(119909)

119908(119909)=1119889119909

119896

sum119899

119909=11119889119909

119896

(11)













1

17

12

11

1098

7

6

54

3

2

16

1514

13

N

0 920 1840460(Meters)

38∘ 54998400 0

998400998400N

38∘ 53998400 0

998400998400N

38∘ 52998400 0

998400998400N

38∘ 51998400 0

998400998400N

38∘ 53998400 30998400998400

N38∘ 52998400 30998400998400

N38∘ 51998400 30998400998400

N38∘ 50998400 30998400998400

N10

0∘ 24998400 30998400998400

E

100∘ 23998400 30998400998400

E

100∘ 22998400 30998400998400

E

100∘ 21998400 30998400998400

E

100∘ 20998400 30998400998400

E

100∘ 19998400 30998400998400

E

100∘249984000998400998400E100∘239984000998400998400E100∘229984000998400998400E100∘219984000998400998400E100∘209984000998400998400E

MaizeVillageOrchard


Gansu Province



119904 and









4 Results



Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302302-303

(a)

Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302Above 302

(b)

Air temperature (K)

N

0 1050 2100(Meters)


(c)





N

0 1050 2100(Meters)


300-301301-302302-303303-304

(a)

N

Air temperature (K)

0 1050 2100(Meters)

296-297295-296

297-298298-299

299-300300-301301-302Above 302

(b)

N

Air temperature (K)

0 1050 2100(Meters)


(c)





















Paired difference


Std errormean




298

299

300

Estim

atio

ns (K

)

Y = 116X minus 474


(a)


298

299

300

Estim

atio

ns (K

)


(b)

298

299

Estim

atio

ns (K

)


Y = 032X + 2015


(c)






5 Discussion



6 Conclusion



Competing Interests


Acknowledgments


References















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in





2 Methodology



119886is equal









119867 =120588119862119901

119903119886

(1198790minus 119879119886) (2)

120573 =119867

119871119864 (3)


119886is the air




119879119886= 1198790minus120573 (119877119899minus 119866)

120573 + 1

119903119886

120588119862119901

(4)




119879loc = 1198790 minus120573 (119877119899minus 119866)

120573 + 1

119903119886

120588119862119901

(5)


119886 and 119879




119879119886= 119891119879exo + (1 minus 119891)119879loc (6)

119891 =119881exo119881119886

(7)

1 minus 119891 =119881loc119881119886

(8)












(2)= 119891(1)= 119891(119894) 119879exo(2) = 119879exo(1) = 119879exo(119894)


119879119886(1)= 119891(1)119879exo(1) + (1 minus 119891(1)) 119879loc(1)

119879119886(2)= 119891(2)119879exo(2) + (1 minus 119891(2)) 119879loc(2)

(9)

where 119879119886(1)

and 119879119886(2)


119891(119894)= 119891(1)= 119891(2)= 1 minus119879119886(1)minus 119879119886(2)

119879loc(1) minus 119879loc(2)

119879exo(119894) = 119879exo(1) = 119879exo(2)

=(119879119886(1)+ 119879119886(2)) minus (1 minus 119891

(119894)) (119879loc(1) + 119879loc(2))

119891(119894)

(10)








ADEBAT IADEBAT



119891(119894)=119899

sum119909=1

119908(119909)119891(119909)

119908(119909)=1119889119909

119896

sum119899

119909=11119889119909

119896

(11)













1

17

12

11

1098

7

6

54

3

2

16

1514

13

N

0 920 1840460(Meters)

38∘ 54998400 0

998400998400N

38∘ 53998400 0

998400998400N

38∘ 52998400 0

998400998400N

38∘ 51998400 0

998400998400N

38∘ 53998400 30998400998400

N38∘ 52998400 30998400998400

N38∘ 51998400 30998400998400

N38∘ 50998400 30998400998400

N10

0∘ 24998400 30998400998400

E

100∘ 23998400 30998400998400

E

100∘ 22998400 30998400998400

E

100∘ 21998400 30998400998400

E

100∘ 20998400 30998400998400

E

100∘ 19998400 30998400998400

E

100∘249984000998400998400E100∘239984000998400998400E100∘229984000998400998400E100∘219984000998400998400E100∘209984000998400998400E

MaizeVillageOrchard


Gansu Province



119904 and









4 Results



Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302302-303

(a)

Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302Above 302

(b)

Air temperature (K)

N

0 1050 2100(Meters)


(c)





N

0 1050 2100(Meters)


300-301301-302302-303303-304

(a)

N

Air temperature (K)

0 1050 2100(Meters)

296-297295-296

297-298298-299

299-300300-301301-302Above 302

(b)

N

Air temperature (K)

0 1050 2100(Meters)


(c)





















Paired difference


Std errormean




298

299

300

Estim

atio

ns (K

)

Y = 116X minus 474


(a)


298

299

300

Estim

atio

ns (K

)


(b)

298

299

Estim

atio

ns (K

)


Y = 032X + 2015


(c)






5 Discussion



6 Conclusion



Competing Interests


Acknowledgments


References















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in









(2)= 119891(1)= 119891(119894) 119879exo(2) = 119879exo(1) = 119879exo(119894)


119879119886(1)= 119891(1)119879exo(1) + (1 minus 119891(1)) 119879loc(1)

119879119886(2)= 119891(2)119879exo(2) + (1 minus 119891(2)) 119879loc(2)

(9)

where 119879119886(1)

and 119879119886(2)


119891(119894)= 119891(1)= 119891(2)= 1 minus119879119886(1)minus 119879119886(2)

119879loc(1) minus 119879loc(2)

119879exo(119894) = 119879exo(1) = 119879exo(2)

=(119879119886(1)+ 119879119886(2)) minus (1 minus 119891

(119894)) (119879loc(1) + 119879loc(2))

119891(119894)

(10)








ADEBAT IADEBAT



119891(119894)=119899

sum119909=1

119908(119909)119891(119909)

119908(119909)=1119889119909

119896

sum119899

119909=11119889119909

119896

(11)













1

17

12

11

1098

7

6

54

3

2

16

1514

13

N

0 920 1840460(Meters)

38∘ 54998400 0

998400998400N

38∘ 53998400 0

998400998400N

38∘ 52998400 0

998400998400N

38∘ 51998400 0

998400998400N

38∘ 53998400 30998400998400

N38∘ 52998400 30998400998400

N38∘ 51998400 30998400998400

N38∘ 50998400 30998400998400

N10

0∘ 24998400 30998400998400

E

100∘ 23998400 30998400998400

E

100∘ 22998400 30998400998400

E

100∘ 21998400 30998400998400

E

100∘ 20998400 30998400998400

E

100∘ 19998400 30998400998400

E

100∘249984000998400998400E100∘239984000998400998400E100∘229984000998400998400E100∘219984000998400998400E100∘209984000998400998400E

MaizeVillageOrchard


Gansu Province



119904 and









4 Results



Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302302-303

(a)

Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302Above 302

(b)

Air temperature (K)

N

0 1050 2100(Meters)


(c)





N

0 1050 2100(Meters)


300-301301-302302-303303-304

(a)

N

Air temperature (K)

0 1050 2100(Meters)

296-297295-296

297-298298-299

299-300300-301301-302Above 302

(b)

N

Air temperature (K)

0 1050 2100(Meters)


(c)





















Paired difference


Std errormean




298

299

300

Estim

atio

ns (K

)

Y = 116X minus 474


(a)


298

299

300

Estim

atio

ns (K

)


(b)

298

299

Estim

atio

ns (K

)


Y = 032X + 2015


(c)






5 Discussion



6 Conclusion



Competing Interests


Acknowledgments


References















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


1

17

12

11

1098

7

6

54

3

2

16

1514

13

N

0 920 1840460(Meters)

38∘ 54998400 0

998400998400N

38∘ 53998400 0

998400998400N

38∘ 52998400 0

998400998400N

38∘ 51998400 0

998400998400N

38∘ 53998400 30998400998400

N38∘ 52998400 30998400998400

N38∘ 51998400 30998400998400

N38∘ 50998400 30998400998400

N10

0∘ 24998400 30998400998400

E

100∘ 23998400 30998400998400

E

100∘ 22998400 30998400998400

E

100∘ 21998400 30998400998400

E

100∘ 20998400 30998400998400

E

100∘ 19998400 30998400998400

E

100∘249984000998400998400E100∘239984000998400998400E100∘229984000998400998400E100∘219984000998400998400E100∘209984000998400998400E

MaizeVillageOrchard


Gansu Province



119904 and









4 Results



Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302302-303

(a)

Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302Above 302

(b)

Air temperature (K)

N

0 1050 2100(Meters)


(c)





N

0 1050 2100(Meters)


300-301301-302302-303303-304

(a)

N

Air temperature (K)

0 1050 2100(Meters)

296-297295-296

297-298298-299

299-300300-301301-302Above 302

(b)

N

Air temperature (K)

0 1050 2100(Meters)


(c)





















Paired difference


Std errormean




298

299

300

Estim

atio

ns (K

)

Y = 116X minus 474


(a)


298

299

300

Estim

atio

ns (K

)


(b)

298

299

Estim

atio

ns (K

)


Y = 032X + 2015


(c)






5 Discussion



6 Conclusion



Competing Interests


Acknowledgments


References















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302302-303

(a)

Air temperature (K)

N

0 1050 2100(Meters)

292-293293-294294-295295-296296-297297-298

298-299299-300300-301301-302Above 302

(b)

Air temperature (K)

N

0 1050 2100(Meters)


(c)





N

0 1050 2100(Meters)


300-301301-302302-303303-304

(a)

N

Air temperature (K)

0 1050 2100(Meters)

296-297295-296

297-298298-299

299-300300-301301-302Above 302

(b)

N

Air temperature (K)

0 1050 2100(Meters)


(c)





















Paired difference


Std errormean




298

299

300

Estim

atio

ns (K

)

Y = 116X minus 474


(a)


298

299

300

Estim

atio

ns (K

)


(b)

298

299

Estim

atio

ns (K

)


Y = 032X + 2015


(c)






5 Discussion



6 Conclusion



Competing Interests


Acknowledgments


References















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


N

0 1050 2100(Meters)


300-301301-302302-303303-304

(a)

N

Air temperature (K)

0 1050 2100(Meters)

296-297295-296

297-298298-299

299-300300-301301-302Above 302

(b)

N

Air temperature (K)

0 1050 2100(Meters)


(c)





















Paired difference


Std errormean




298

299

300

Estim

atio

ns (K

)

Y = 116X minus 474


(a)


298

299

300

Estim

atio

ns (K

)


(b)

298

299

Estim

atio

ns (K

)


Y = 032X + 2015


(c)






5 Discussion



6 Conclusion



Competing Interests


Acknowledgments


References















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in


















Paired difference


Std errormean




298

299

300

Estim

atio

ns (K

)

Y = 116X minus 474


(a)


298

299

300

Estim

atio

ns (K

)


(b)

298

299

Estim

atio

ns (K

)


Y = 032X + 2015


(c)






5 Discussion



6 Conclusion



Competing Interests


Acknowledgments


References















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in



Paired difference


Std errormean




298

299

300

Estim

atio

ns (K

)

Y = 116X minus 474


(a)


298

299

300

Estim

atio

ns (K

)


(b)

298

299

Estim

atio

ns (K

)


Y = 032X + 2015


(c)






5 Discussion



6 Conclusion



Competing Interests


Acknowledgments


References















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in



5 Discussion



6 Conclusion



Competing Interests


Acknowledgments


References















































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in








































Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in










Volume 2014

Mining


Journal of



Geophysics







Journal of




Advances in











Geology Advances in

Documents

Research Article Estimating the Surface Air Temperature by