Response of Hydrological Processes to Input Data in High Alpine Catchment

15 Pages·2016·3.55 MB·English

water
Article
Response of Hydrological Processes to Input Data in
High Alpine Catchment: An Assessment of the
Yarkant River basin in China
JiaoLiu1,2,3,TieLiu1,AnmingBao1,*,PhilippeDeMaeyer3,AlishirKurban1andXiChen1
1 StateKeyLaboratoryofDesertandOasisEcology,XinjiangInstituteofEcologyandGeography,
ChineseAcademyofSciences,Urumqi830011,China;[email protected](J.L.);
[email protected](T.L.);[email protected](A.K.);[email protected](X.C.)
2 UniversityofChineseAcademyofSciences,Beijing100049,China
3 DepartmentofGeography,GhentUniversity,Gent9000,Belgium;[email protected]
* Correspondence:[email protected];Tel.:[phone]
AcademicEditor:AthanasiosLoukas
Received:22February2016;Accepted:25April2016;Published:5May2016
Abstract: Moststudiesofinputdatausedinhydrologicalmodelshavefocusedonflow;however,
point discharge data negligibly reflect deviations in spatial input data. To study the effects of
different input data sources on hydrological processes at the catchment scale, eight MIKE SHE
models driven by station-based data (SBD) and remote sensing data (RSD) were implemented.
Thesignificantinfluencesofinputvariablesonwatercomponentswereexaminedusingananalysisof
thevariancemodel(ANOVA)withthehydrologiccatchmentresponsequantifiedbasedondifferent
water components. The results suggest that compared with SBD, RSD precipitation resulted in
greaterdifferencesinsnowstorageinthedifferentelevationbandsandRSDtemperaturesledto
moresnowpackareaswiththinnerdepths. Thesechangesinsnowpackprovidedanappropriate
interpretationofprecipitationandtemperaturedistinctionsbetweenRSDandSBD.Forpotential
evapotranspiration(PET),thelargerRSDvaluecausedlessplanttranspirationbecauseparameters
were adjusted to satisfy the outflow. At the catchment scale, the spatiotemporal distributions of
sensitivewatercomponents,whichcanbedefinedbytheANOVAmodel,indicatethatthisapproach
isrationalforassessingtheimpactsofinputdataonhydrologicalprocesses.
Keywords: inputdata;hydrologicalprocesses;statistichypothesistest;spatiotemporaldistribution;
YarkantRiver;MIKESHE
1. Introduction
Modelsimulationisaprincipalapproachforstudyinghydrologicalprocessesatthecatchment
scale; however, the accuracy of modelling results is dwarfed owing to the uncertainties of model
parameterizations, model structures and input data [1]. Most studies [2–5] have analysed the
uncertaintiesofparametersusingdifferentmethodsandcalibrationfocusedonmodel’sparameters
hasdeterminedthecontributionsfromothersourcesofuncertainties[6]. Ajamietal.[7]confirmed
the importance of input data and model structure to model output using the Integrated Bayesian
Uncertainty Estimator (IBUNE). Kavetski et al. [8] demonstrated a multitude of distinctions in
the predicted hydrographs and calibrated parameters, with or without consideration of the input
uncertaintyofprecipitationdatawithsimilarconclusionsdrawnbyXuetal.[9]. Inadditiontothe
precipitationdata,Thompson’s[10]researchrevealedthatPETrelateduncertaintyexistedinhighand
lowdischarges. Furthermore,anumberofstudieshaveillustratedthatinputdatamayprofoundly
influencepredictedriverrunoff[11–14].However,littleattentionhasbeenpaidtotheeffectofdifferent
inputdataontheentirehydrologicalcycleatthecatchmentscale.
Water2016,8,181;doi:10.3390/w8050181 www.mdpi.com/journal/water
Water 2016, 8, 181  2 of 16 
Water2016,8, 181 2of15
In mountainous catchments with scarce gauges, input data uncertainties are amplified due to 
the lack of observed data [15]. With the development of satellite technology, RSD provides a lot of 
Inmountainouscatchmentswithscarcegauges,inputdatauncertaintiesareamplifieddueto
information to drive, calibrate and validate hydrological models [16]. Some applications of RSD in 
thelackofobserveddata[15]. Withthedevelopmentofsatellitetechnology,RSDprovidesalotof
hydrological models, even in semi‐arid/arid watersheds, have attained promising performance [17–
informationtodrive,calibrateandvalidatehydrologicalmodels[16]. SomeapplicationsofRSDin
19]. RSD also supply a wealth of new observation types that can be applied to assess model 
hydrologicalmodels,eveninsemi-arid/aridwatersheds,haveattainedpromisingperformance[17–19].
uncertainty [20]. According to McMichael et al. [21], the prediction uncertainties of discharge among 
RSDalsosupplyawealthofnewobservationtypesthatcanbeappliedtoassessmodeluncertainty[20].
seven leaf area index (LAI) scenarios were less than 10% using the MIKE SHE simulation. Sun et al. 
AccordingtoMcMichaeletal.[21],thepredictionuncertaintiesofdischargeamongsevenleafarea
[18]  demonstrated  that  the  uncertainties  derived  from  RSD  are  smaller  than  those  of  model 
index(LAI)scenarioswerelessthan10%usingtheMIKESHEsimulation. Sunetal.[18]demonstrated
parameters. However, Knoche et al. [22] argued that high‐resolution land surface temperature (LST) 
thattheuncertaintiesderivedfromRSDaresmallerthanthoseofmodelparameters. However,Knoche
data did not yield a better result and that the simulated hydrographs could not explain the differences 
etal.[22]arguedthathigh-resolutionlandsurfacetemperature(LST)datadidnotyieldabetterresult
in input data. Due to satellite data limitations associated with the physical sensors, space‐time 
andthatthesimulatedhydrographscouldnotexplainthedifferencesininputdata. Duetosatellite
coverage and spatial resolution [23], the application of raw RSD has been disputed and has resulted 
datalimitationsassociatedwiththephysicalsensors,space-timecoverageandspatialresolution[23],
in some controversial results. 
theapplicationofrawRSDhasbeendisputedandhasresultedinsomecontroversialresults.
This paper characterizes the influence of different input data sources based on variations in 
This paper characterizes the influence of different input data sources based on variations in
distributed simulation outputs. Eight MIKE SHE models forced by SBD and RSD were calibrated 
distributedsimulationoutputs. EightMIKESHEmodelsforcedbySBDandRSDwerecalibrated
separately to obtain optimal performance. The significant impact of input data on different water 
separatelytoobtainoptimalperformance. The significantimpactofinputdataondifferentwater
components was determined via the ANOVA model. Then the effects of the different input data on 
componentswasdeterminedviatheANOVAmodel. Thentheeffectsofthedifferentinputdataon
hydrological processes was studied based on the water components. 
hydrologicalprocesseswasstudiedbasedonthewatercomponents.
22.. SSttuuddyy AArreeaa 
YYaarrkkaanntt RRiivveerr bbaassiinn ((FFiigguurree 11)) iiss llooccaatteedd iinn tthhee XXiinnjijiaanngg UUyygghhuurr aauuttoonnoommoouuss rreeggiioonn,, aann aarriidd 
ddiissttrriicctt ininN NorothrtwhewsetsCt hCinhai.nIat.i sItt hies lothneg elsotnsgoeusrtc estoruibruceta rtyribouftthareyT aorfi mthRei vTear,ritmhe lRoinvgeers, ttchoen tlionnengteaslt 
rciovnetrininenthtael wriovrelrd i.nT thheet owpoorgldra. pThhye otofpthoigsrsatpuhdyy ocfa ttchhism setnutdiys cvaetrcyhcmoemnpt liesx v.eGryen ceormalplyl,etxh. eGseonuetrhaelrlyn, 
htheea dsowuattheerrnre hgeioandwhaatsera rmeguiocnh hhaigs hae mrueclehv haitgiohnert ehlaenvatthioant othfatnh ethnaot rotfh tehren nmorotuhenrtna imnoouuntlteatinr eoguitolnet. 
Mreoguionnt.a Minos,ugnotarginess, agnodrgbeass ainnsd abraessintasg agreer setdagtgheroreudg hthoruotutghheowuta ttheres wheadte.rTshheede.l eTvhaet ieolenvraatniogne srafnrogmes 
1f4ro5m0m 14t5o08 m61 t1om 86,1a1v emra, gaivnegra4g4i5n0gm 4.45T0h ema. vTehrea gaeveslroapgee issloappep irso xaipmparotexliym3a0t.e8l%y .30.8%. 

FFiigguurree 11.. TThhee llooccaattiioonn,, ttooppooggrraapphhyy,, ggaauuggiinngg ssttaattiioonnss aanndd rriivveerr nneettwwoorrkk ooff YYaarrkkaannttR Riivveerrb baassinin.. 
This catchment is situated on the northern slope of Karakoram, the largest area of mountain 
This catchment is situated on the northern slope of Karakoram, the largest area of mountain
glaciers in the world. Based on the Landsat8 TM, land use data were generated and shows that the 
glaciersintheworld. BasedontheLandsat8TM,landusedataweregeneratedandshowsthatthe
snowpack and glaciers cover 26.14% of this watershed (Figure 2a), situated mainly above 5000 m 
snowpack and glaciers cover 26.14% of this watershed (Figure 2a), situated mainly above 5000 m
(Figure 2b). The total number of glaciers is 2698, including Insukati Valley, the largest glacier in China
Water2016,8, 181 3of15
Water 2016, 8, 181  3 of 16 
(Figure2b). Thetotalnumberofglaciersis2698,includingInsukatiValley,thelargestglacierinChina
wwiitthh aa lleennggtthh ooff 4400 kkmm [[2244]].. TThhee ootthheerr mmaajjoorr llaanndd ccoovveerr ttyyppeess aarree cclloosseedd‐-ooppeenn hheerrbbaacceeoouuss,, bbaarree llaanndd 
aanndd ssppaarrssee hheerrbbaacceeoouuss aarreeaass ((FFiigguurree 22aa)) llooccaatteedd bbeellooww 55000000 mm ((FFiigguurree 22bb)),, ccoonnttrriibbuuttiinngg ttoo 2299%%,, 2200%% 
aanndd 1199..11%% ooff tthhee ttoottaall aarreeaa,, rreessppeeccttiivveellyy.. TThhee aabbuunnddaanntt ssnnoowwppaacckk aanndd ggllaacciieerrss pprroovviiddee aammppllee wwaatteerr 
rreessoouurrcceess ttoo tthheeY YarakraknatnRt iRvievreirr riirgraigtiavteivoea soiassdiso wdonwstnresatrmeabmel obweloKwaq Kuanq,uwnit, hwaintha raena aorfe2a. 5oˆf 21.054 ×k m1024. 
kItmis2.t Ihte ilsa rtghees tlairrgriegsatt iiorrnigreagtiioonn rinegXioinnj iainn gXainnjdiaangm aajnodr par omdaujocerr porfogdruacinera nodf gcorattionn a[n2d5 ].coBtetcoanu s[2e5o]f. 
Btheecasueasseo onfa tlhsen osewasaonndailc senmowel ta,nthde ivceo lmumelet, othfrei vveorlurumneo offf irnivtheer rfluonoodffs einas tohne (fflrooomd Mseaaysotno S(ferpotmem Mbaery) 
taoc cSoeupntetsmfboerr8)0 a%ccoofutnhtes afonrn 8u0a%lr oufn tohfef. annual runoff. 

FFiigguurree 22.. TThhee llaanndd uussee ttyyppeess.. ((aa)) IInn 22000055 aanndd eelleevvaattiioonn bbaannddss;; ((bb)) iinn YYaarrkkaanntt RRiivveerr bbaassiinn.. 
The study area is a rarely studied catchment with only one internal meteorological station 
The study area is a rarely studied catchment with only one internal meteorological station
(Tashkurgan) and two adjacent stations (Shache and Pishan) (Figure 1). Based on the records from 
(Tashkurgan)andtwoadjacentstations(ShacheandPishan)(Figure1). Basedontherecordsfrom
Tashkurgan station in 2000–2009, pan evaporation was much higher than precipitation, averaging 
Tashkurganstationin2000–2009, panevaporationwasmuchhigherthanprecipitation, averaging
1500 mm compared to annual mean precipitation of 98 mm. Furthermore, because of the extreme 
1500mmcomparedtoannualmeanprecipitationof98mm. Furthermore, becauseoftheextreme
topography variations, there are strong heterogeneities of precipitation and temperature in spatial 
topographyvariations,therearestrongheterogeneitiesofprecipitationandtemperatureinspatial
distributions [26]. Annual precipitation is approximately 450 mm in high elevation areas (higher than 
distributions[26]. Annualprecipitationisapproximately450mminhighelevationareas(higherthan
5000 m) and only 100 mm in lower regions (approximately 3000 m). The mean temperature around 
5000m)andonly100mminlowerregions(approximately3000m). Themeantemperaturearound
the snowline (approximately 5500 m) is approximately –10.5 °C. Therefore, the elevation lapse rate 
thesnowline(approximately5500m)isapproximately–10.5˝C.Therefore,theelevationlapserateof
of precipitation (PCG) and air temperature (TCG) at different elevations can be determined, as shown 
precipitation(PCG)andairtemperature(TCG)atdifferentelevationscanbedetermined,asshown
in Table 1. 
inTable1.
TTaabbllee 11.. PPCCGG aanndd TTCCGG aatt ddiiffffeerreenntt eelleevvaattiioonnss iinn YYaarrkkaanntt RRiivveerr bbaassiinn.. 
Altitude Group (m) PCG (mm/km/year) TCG (°/km) 
AltitudeGroup(m) PCG(mm/km/year) TCG(˝/km)
<3000  0.0  −6.5 
<3000 0.0 ´6.5
3000–5000  −70.0  −6.8 
3000–5000 ´70.0 ´6.8
5000–7000  100.0  −7.0 
5000–7000 100.0 ´7.0
>>77000000  7700.0.0  −´66.8.8 
3. Forcing Data 
3. ForcingData
33..11.. SSttaattiioonn DDaattaa 
TThhee ssttaattiioonnm meteetoeroorloolgoigciacladl adtaatwa awsoabs taoibnteadinferdom frtohme Cthhien aCMhientae oMroelotegoicraolloDgaitcaalS hDaartina gSShearrvinicge 
SSeyrsvteicme (Shyttspt:e/m/ d(ahtatt.pcm://ad.acnta/.c)manad.cnd/a)i lyanpdre cdipaiiltya tiponreacnipditmateioann aairndte mmpeearna tuarier stwemerpeecroaltluecrteesd  wfroerme 
c2o0l0l0ecttoed20 f0r9o.mE le2v0a0t0i otno i2s0a0c9e. nEtlreavlafaticotnor itsh aa tcseignntriafilc afancttloyra ftfheactt ssdigisntirfiibcauntitolyn saofffemctest edoisrotrliobguictiaolndsa toaf, 
meteorological data, therefore the whole catchment was divided into 10 elevation bands at 700 m 
intervals (Figure 2b) to spatially distribute the SBD. The interpolated SBD are calculated as follows:
Water2016,8, 181 4of15
thereforethewholecatchmentwasdividedinto10elevationbandsat700mintervals(Figure2b)to
spatiallydistributetheSBD.TheinterpolatedSBDarecalculatedasfollows:
` ˘
PCG
R “ R ` EL ´EL ˆ ˆR ą0.01 (1)
band day band gage daysˆ1000 day
` ˘
TCG
T “ T ` EL ´EL ˆ (2)
band day band gage
1000
where R (T ) and R (T ) are the precipitation in mm (average temperature in ˝C) for
band band day day
the calculated elevation band and at the gauged station, the precipitation (mm H O) and average
2
temperature(˝C)recordedatagaugingstation,respectively;EL andEL arethemeanelevations
band gage
(m)ofthecalculatedelevationbandandgaugingstationanddaysistheaveragenumberofrainydays
inayearatthestation. Basedonapreviousstudy[27],thesameinterpolationsofprecipitationand
temperaturewereusedinSWATandMIKESHE.SWATcalculatesPETatasub-basinscaleusingthe
Penman-Monteithequationbasedonstation-interpolateddata. Thissemi-distributedPEToutputfrom
SWATischosenasinputSBDPETinMIKESHE.
3.2. TRMM
Remotely sensed precipitation data was obtained from the daily productions of the Tropical
RainfallMeasuringMission(TRMM)3B42V6(http://trmm.gsfc.nasa.gov)ataspatialresolutionof
0.25˝, althoughtheTRMMV6datawascalibratedonaglobalscaleusingraingaugestations[28].
In mountainous regions such as the Yarkant River basin with arid climate conditions, extreme
topographyandfewraingauges,theaccuracyisstillunfavourable,especiallyatthedailytemporal
scale[29]. Thus,ananalysiswasconductedtodeterminewhethertheTRMMsatellitedetectedthe
correctprecipitationevents(Equations(3)and(4)). Inthisanalysis,TRMM isdefinedasthetotal
dec
daysinwhichprecipitationeventsweredetectedbyTRMMbutnotrecordedbySBD.SBD isdefined
dec
asthetotalnumberofdaysinwhichprecipitationeventswererecordedbySBDbutnotrecorded
by TRMM and TRSB is defined as the total number of days in which precipitation events were
dec
detectedbybothTRMMandSBD.Inaddition,D andD arethepercentagesofcorrectandincorrect
r w
precipitationeventsdetectedbyTRMMforthesatellitegridsthatthestationsoccupy.
TRSB
D “ dec (3)
r
TRSB `SBD
dec dec
TRMM
D “ dec (4)
w
TRMM `TRSB
dec dec
The evaluation results are listed in Table 2. For the three stations, the D values are much
w
larger than D , indicating that many precipitation events detected by TRMM are redundant.
r
Anin-depthanalysisestimatedthedifferentintensityclassesofprecipitationusingthesameapproach
(Equations(3)and(4)). TheresultssuggestthathighvaluesofD mainlycorrespondtolow-intensity
w
precipitation events (<0.3 mm) with a high probability of incorrect precipitation events (D )
w_0.3
(Table2).
Table2.ComparisonbetweenrawandcorrectedTRMMprecipitation.
Station Dr Dw Dw_0.3 rraw rcor
Tashkurgan 0.54 0.75 0.95 0.11 0.45
Shache 0.21 0.87 0.96 0.67 0.77
Pishan 0.15 0.88 0.99 0.36 0.67
AscanbeconcludedfromTable2,directapplicationofraw3B42V6dataintheYarkantRiver
basinisinappropriateformodelsimulation. Therefore,acorrectionwasperformedbasedonthelocal
Water2016,8, 181 5of15
intensityscaling(LOCI)approach[30],whichcancorrectthewet-dayfrequenciesandintensitiesand
effectivelyimprovetheoverestimateofrainydaysintherawdata. First,awet-daythresholdP
thres
wasdeterminedfromtherawTRMMdatatoensurethatthefrequencymatchedtheSBD.Thisvalue
Water 2016, 8, 181  5 of 16 
wassetas0.3becauseTRMMdetectedtoomanyredundantrainydayswithprecipitationamounts
swmaasll esertt haas n0.03. 3bmecmau(sTea TbRleM2)M. A dsecteaclitnegd ftaocot omrasnyw raesdtuhnednacnatl crualiantye ddaaynsd wuistehd ptroedciepteitramtiionne athmaotuthnets 
m
msmeaanlloerf tthheanc o0r.r3e mctemd (pTraebcliep i2t)a. tAio nscwalainsge qfaucatlotro smth weaosb tsheervne cdalpcruelcaitpeidta atniodn u,assedfo tlolo dweste:rmine that the 
mean of the corrected precipitation was equ`al to thˇe observed ˘precipitation, as follows: 
ˇ
µ P P ą0
SBD,m SBD,m
sm “ (cid:2020)(cid:3435)(cid:1842) (cid:3627)(cid:1842) (cid:3408) 0(cid:3439) (5)
µpP (cid:3020)(cid:3003)(cid:3005)|,P(cid:3040) (cid:3020)(cid:3003)(cid:3005),(cid:3040) ą P q
(cid:1871) (cid:3404) TRMM.raw,m TRMM.raw,m thres   (5) 
(cid:3040)
(cid:2020)(cid:3435)(cid:1842) # (cid:3627)(cid:1842) (cid:3408) (cid:1842) (cid:3439)
(cid:3021)(cid:3019)(cid:3014)(cid:3014).(cid:3045)(cid:3028)(cid:3050),(cid:3040) (cid:3021)(cid:3019)(cid:3014)(cid:3014).(cid:3045)(cid:3028)(cid:3050),(cid:3040) (cid:3047)(cid:3035)(cid:3045)(cid:3032)(cid:3046)
0, if P ă P
P “ 0,(cid:1861)(cid:1858) (cid:1842)TRMM.raw,m (cid:3407)th(cid:1842)res (6)
(cid:1842)TRMM.cor,m (cid:3404) (cid:3420) PTRMM.r(cid:3021)aw(cid:3019),(cid:3014)m(cid:3014)ˆ.(cid:3045)s(cid:3028)m(cid:3050),,(cid:3040)otherwi(cid:3047)s(cid:3035)e(cid:3045)(cid:3032)(cid:3046)   (6) 
(cid:3021)(cid:3019)(cid:3014)(cid:3014).(cid:3030)(cid:3042)(cid:3045),(cid:3040) (cid:1842) (cid:3400)(cid:1871) ,(cid:1867)(cid:1872)(cid:1860)(cid:1857)(cid:1870)(cid:1875)(cid:1861)(cid:1871)(cid:1857)
(cid:3021)(cid:3019)(cid:3014)(cid:3014).(cid:3045)(cid:3028)(cid:3050),(cid:3040) (cid:3040)
whereP andP aretheSBDandrawTRMMprecipitationrespectivelyinthemthmonth,
SBD,m TRMM.raw,m
awndhePre PSBD,m anids tPhTeRMcMo.rrarwe,mc taerde TthReM SMBDp raencdip ritaawti oTnRiMntMhe pmrethcimpiotanttiho.nµ r(e.)spreepctrievseelnyt sint htehee xmpteh cmtaotniotnh, 
TRMM.cor,m
oapnedra PtoTRrM(Me..cgor.,m, µis(P the correcte)da TnRdMisMth eprmeceiapnitvaatiloune ionf trhaew mTtRh MmoMntphr.e c(cid:2020)i(p.)i traetpiornesienngtsiv tehne mexopnetchtamti)o.n 
TRMM,.raw,m
operAatfoterr (ceo.gr.r,e c(cid:2020)t(iPonTR,MtMh,e.rarwa,mw) aTnRdM isM thdea mtaeqauna vliatlyuwe aosf rmawuc hTRimMpMro pveredc.iCpiotmatipoanr eidn gtoivtehne mraownTthR mM)M.  ,
After correction, the raw TRMM data quality was much improved. Compared to the raw TRMM, 
the corrected TRMM was much closer to the SBD in the terms of the mean annual precipitation.
the corrected TRMM was much closer to the SBD in the terms of the mean annual precipitation. In 
Inaddition,themonthlycorrelationcoefficientalsodisplayedaremarkableimprovementrefertor
raw
aandddritionin, tThaeb mleo2n.tEhvlye nctourarelllya,ttiohnis ccooerfrfeicctieionnt aalpspo rdoiascphlacyaendb ae raepmpalirekdabtolet himepwrhoovleembeansti nrebfaesre tdo orrnaw 
cor
thanedo rrdcoir ninar TyaKbrleig 2in. gEvmenetthuoaldly,w, tihtihs cciorcrruelcatriomno adpeplrionatecrhp coalnat ebde asppulsieindg toA rthcGe IwS,hporleo vbiadsiinng braesveidse odn 
m
TtRhMe oMrddinaatar.y Kriging method, with circular model interpolated sm using ArcGIS, providing revised 
TRMM data. 
3.3. LST
3.3. LST 
As for temperature, Moderate Resolution Imaging Spectroradiometer (MODIS
http:/A/st rmmfor.g sfcte.nmaspae.rgaotuv/re), MOMDod11eCra1ted  aiRlyesLoSluTtiodna ta Iwmaasgiunsge d wSpitehctraorsapdaitoimalerteers olu(tMioOnDoIfS 
0h.0tt5p˝:.//Vtremrsmio.ngs4f.c1.nwasaas.gcohvo/s)e nMbOeDca1u1sCe1o dfaitilsye LffSeTct idvaetnae swsaisn usesemdi -wariitdh aan sdpaartiiadl rreegsioolnustio[3n1 ]o.f D0a.0il5y°. 
mVeearnsioanir 4t.e1m wpaesr achtuorseend abteacawuasse noef eitdse edffaescttihveenMesIsK iEn SseHmEi‐ianrpidu ta.nPdr eavriido uresgrieosnesa r[3ch1].c oDnafiilrym medeatnh aaitr 
thteemrepiesraatguoroed dlaintae awrarse lnateieodnesdh ipasb tehtew MeeInKLES TSHanEd inapirutte. mPpreevraiotuurse r[e3s2e,a3r3c]h. BceocnafuirsmeeSdh atchhaet stthaetrioen isi sa 
mgiososdin lginLeSaTr rdealatati,oonnslhyipT absehtwkueregna LnSaTn adnPdi sahira ntesmtapteiorantudraet a[3w2,e3r3e].p Bloecttaeudseto Svhearcihfye sthtaitsiorenl aist imonissshiinpg 
(FLiSgTu rdea3ta)., only Tashkurgan and Pishan station data were plotted to verify this relationship (Figure 3). 

Figure 3. Regression relationship between daily air temperature and LST at the Tashkurgan and 
Figure3.RegressionrelationshipbetweendailyairtemperatureandLSTattheTashkurganandPishan
Pishan stations in 2000–2009. 
stationsin2000–2009.
The mean values of the regression coefficients at Tashkurgan station and Pishan station were 
ThemeanvaluesoftheregressioncoefficientsatTashkurganstationandPishanstationwereused
used in a simple transform Equation (7) to calculate air temperature for the Yarkant River basin. Daily 
inasimpletransformEquation(7)tocalculateairtemperaturefortheYarkantRiverbasin. Dailymean
mean air temperatures were acquired as input data for the MIKE SHE model. 
airtemperatureswereacquiredasinputdatafortheMIKESHEmodel.
(cid:1846)(cid:3031)T(cid:3028)(cid:3052) (cid:3404)“00..77559922ˆ(cid:3400)LS(cid:1838)T(cid:1845)´(cid:1846)5(cid:3398).5655.565  ((77)) 
day
3.4. PET 
Daily  global  PET  (GPET)  data  derived  from  the  FEWS  NET  Data  Portal 
(http://earlywarning.usgs.gov/fews/) was used in this study at a spatial resolution of 1°. This daily
Water2016,8, 181 6of15
3.4. PET
Daily global PET (GPET) data derived from the FEWS NET Data Portal
(http://earlywarning.usgs.gov/fews/) was used in this study at a spatial resolution of 1˝.
This daily PET was calculated on a spatial basis using the Penman-Monteith equation and the
formulationforreferencecropevaporation[34]. GPETwasdirectlyappliedinthisresearch.
4. Methodology
4.1. Models
Becauseofthestrongspatialheterogeneityassociatedwithextremetopographicalconditions,
the fully distributed hydrological MIKE SHE model (European Hydrological System) [35] was
employedinthisstudytorevealthespatialheterogeneityoftheYarkantRiverbasin. MIKESHEisa
deterministic,dynamic,physicallybasedhydrologicalmodel. IntheMIKESHEmodel,catchmentsare
splitintoanumberofsquaregridstorevealthespatialheterogeneityofthecatchment. Theinputdata
andparametersineachgridareindependent. Basedonbilinearinterpolationmethod,allinputand
outputdatapossessesthesamespatialresolution,assetinthemodel.
Inthisstudy,aspatialresolutionof2kmwassetintheMIKESHEmodel. SBDwasappliedasa
benchmarkandthreetypesofRSDandtheircombinationswereappliedindividuallytoreplacethe
SBD.Finally,eightmodelswereutilizedbasedondifferentinputdata(Table3). Theentiresimulated
periodwasuniformfrom2000to2009,includingawarm-upperiodfrom2000–2002,calibrationperiod
from2003–2007andverificationperiodfrom2008–2009.
Table3.Theeightmodelsinthisstudybasedondifferentinputdata.
DataSource
ModelAbbreviation
Rainfall Temperature Evapotranspiration
station station station STA
TRMM station station TRMM
station LST station LST
station station GPET GPET
TRMM LST station TRLS
TRMM station GPET TRGP
station LST GPET LSGP
TRMM LST GPET RSD
4.2. Calibration
To obtain the optimal performance of each model, the same calibration procedures were
executedseparately. TheautocalibrationtoolbasedontheShuffledComplexEvolution(SCE)global
optimizationalgorithm[36],whichisapartoftheMIKESHEpackage,waschosenforcalibration.
Themainsignificantparametersofsnowmeltchosenforcalibrationincludedegree-dayfactor(DDF)
andthresholdmeltingtemperature(TMT).ThoseoflandsurfaceflowincludeManningvalues(MAN).
Thoseofinterflowincludehorizontalhydraulicconductivity(HHC)andverticalhydraulicconductivity
(VHC).ThoseofevapotranspirationincludeLAIandrootdepth(RD).Duringthecalibration,standard
deviation(STD)waschosenasastatisticaloutputmeasureandtheweightedsumofSTDwastaken
as the objective function. Additionally, three statistical coefficients were used as the performance
judgementcriteria: Nash-Sutcliffe[37]efficiencycoefficientNSC,Pearsoncorrelationcoefficientrand
rootmeansquareerrorRMSE.Theirexpressionsarewrittenasfollows:
ř ` ˘
n Q ´Q 2
NSC“1´ ři“1` obs,i sim,i˘ (8)
n Q ´Q 2
i“1 obs,i obs,i
Water2016,8, 181 7of15
ř ` ˘` ˘
n Q ´Q Q ´Q
r“ bř `i“1 obs,i ˘obsb,i ř sim`,i sim,i ˘ (9)
n Q ´Q 2 n Q ´Q 2
i“1 obs,i obs,i i“1 sim,i sim,i
d
ř
n pQ ´Q q2
RMSE“ i“1 sim obs (10)
n
where Q and Q are the measured and simulated discharges respectively, on day i (m3/s);
obs,i sim,i
Q andQ aretheaveragemeasuredandsimulateddischargesrespectively,duringthesimulated
obs sim
period(m3/s);andnisthetimestep.
4.3. HypothesisTest
Thenaturalprocessesofhydrologicalcyclesconstitutecomplexsystemswithhighlynonlinear
relationshipsbetweenthe“affected”andthe“caused”. Toevaluatetheeffectofinputdatauncertainty
onthehydrologicalprocess,itisnecessarytodefinethesensitivitiesofthewatercomponentstothe
differentinputdata. Inthisstudy,astatisticalhypothesistestbasedontheANOVAmodel,whichis
usefulincomparingthestatisticalsignificanceofthreeormoregroups,waschosentotestdifferent
significanteffectoftheinputdataonwatercomponents.
Threetypesofinputdata(precipitation,temperatureandPET)wereusedaseffectfactorsA,B
andC.Theywereinvestigatedattwolevels(SBDandRSD).WhenAisatleveli(I=1,2),Bisatlevelj
(j=1,2)andCisatlevelk(k=1,2),theresultingwatercomponentcanbedenotedbyu . TheANOVA
i,j,k
model[38]forthreefactorswithfixedeffectsisasfollows:
y “u...`α `β `γ `pαβq `pαγq `pβγq `pαβγq `ε (11)
ijkn i j k ij ik jk ijk ijkn
wherey istheobservationforthenthcaseortrialforthescenariobasedontheithlevelofA,jthlevel
ijkn
ofBandthekth levelofC;u...isthegrandmeanofalldependentvariablesu ; α, β andγ are
i,j,k i j k
themaineffectsoffactorsA,BandC,respectively; pαβq , pαγq andpβγq arethemaineffectsof
ij ik jk
thetwo-factorinteraction;pαβγq isthemaineffectofthethree-factorinteraction; and ε isthe
ijk ` ˘ ijkm
independentrandomerrorwhichfollowsanormaldistributionN 0,δ2 .
WhenfactorAistested,weassumethatthenullhypothesisH ,meaningthattheinfluenceofthe
0
factorontheresultisnotsignificant,isacceptable. Thus,α “ α “ 0. Inthiscase,theprobability
1 2
p=P(F >F )canbecalculated. Whenthecalculatedp-valueislargerthanthesignificance
[(a-1),abc(n-1)] A
level(α),setas0.05inthisstudy,suggestingthattheassumptioniscorrect,thenullhypothesisH will
0
beaccepted. Otherwise,thealternativehypothesisH shouldbeaccepted. Otherhypothesescouldbe
a
testedinthesameway.
5. ResultsandDiscussion
5.1. SimulatedDischarges
Afterseparatecalibrations,thefinalvaluesofchosenparameterswereobtained. Amongtheeight
models, MAN rangedfrom25to60basedondifferentlanduse. Forthedifferentsoiltypes, HHC
andVHCrangedfrom0.0003to0.001and0.0035to0.01,respectively. MAN,HHCandVHCshowed
insignificantchangesbetweendifferentmodelsforthesamelanduseorsoiltype.Thecalibratedvalues
ofotherparametersarelistedinTable4. Whentemperaturewasreplaced,theparametersofsnow
(DDFandTMT)wereadjusteddrastically. WhenPETwasreplaced,parametersofevapotranspiration
(LAIandRD)changedremarkably. Intheothercases,thevariationswereinsignificant. Thesewere
revealedbycomparingparametervaluesunderSTAtothethreemodelssettingwithonlyoneremote
sensinginput(i.e.,theTRMM,LSTandGPET)intable. Becauseoftheperiodicgrowth,LAIandRDin
herbaceousareasexhibitdistincttemporalvariations[39]andnotlistedinTable4.
Water2016,8, 181 8of15
Table4.Thefinalvaluesofcalibrationparametersindifferentmodels.
Water 2016, 8, 181  8 of 16 
Models STA TRMM LST GPET TRLS TRGP LSGP RSD
TMT (°C)  −0.98  −1.00  −0.56  −1.01  −0.57  −1.02  −0.56  −0.55 
DDF(mm/day/˝C) 2.01 2.03 1.25 1.98 1.25 2.00 1.23 1.25
LAI_NLT *  3.81  3.82  3.78  2.65  3.82  2.64  2.63  2.66 
TMT(˝C) ´0.98 ´1.00 ´0.56 ´1.01 ´0.57 ´1.02 ´0.56 ´0.55
RD_NLT (mm) *  4500  4500  4500  4000  4500  4000  4000  4000 
LAI_NLT* 3.81 3.82 3.78 2.65 3.82 2.64 2.63 2.66
RD_N*L TLA(mI_mN)L*T and R45D0_0NLT a4r5e0 t0he LA4I 5a0n0d RD 4v0a0lu0es of e4v5e0r0green n4e0e0d0le‐leav4e0d0 t0rees.  4000
*LAI_NLTandRD_NLTaretheLAIandRDvaluesofevergreenneedle-leavedtrees.
Based on the multiple evaluation coefficients (Table 5) in the verification period, the 
performances of most models were acceptable, except for the TRGP and RSD models. The 
Basedonthemultipleevaluationcoefficients(Table5)intheverificationperiod,theperformances
NSC values of the other six models were higher than 0.5 and all r values exceeded 0.7. The 
ofmostmodelswereacceptable,exceptfortheTRGPandRSDmodels. TheNSCvaluesoftheother
STA model exhibited the best result, with the highest NSC and r value and the smallest 
six models were higher than 0.5 and all r values exceeded 0.7. The STA model exhibited the best
RMSE, suggesting that the interpolation of meteorological data based on elevation intervals 
result,withthehighestNSCandrvalueandthesmallestRMSE,suggestingthattheinterpolationof
was feasible. The LOCI approach can correct only the magnitude and frequency of TRMM 
meteorologicaldatabasedonelevationintervalswasfeasible. TheLOCIapproachcancorrectonly
and not the probability distribution, possibly causing the relatively poor performance of 
themagnitudeandfrequencyofTRMMandnottheprobabilitydistribution, possiblycausingthe
the TRMM application. In general, the application of the corrected RSD in the Yarkant River 
relativelypoorperformanceoftheTRMMapplication. Ingeneral,theapplicationofthecorrectedRSD
basin was acceptable, although it could be improved. 
intheYarkantRiverbasinwasacceptable,althoughitcouldbeimproved.
Table 5. Statistical coefficients of the performances of different models. 
Table5.Statisticalcoefficientsoftheperformancesofdifferentmodels.
Model  STA  TRMM  LST  GPET  TRLS  TRGP  LSGP  RSD 
Model STA TRMM LST GPET TRLS TRGP LSGP RSD
NSC  0.65  0.50  0.52  0.61  0.55  0.42  0.52  0.46 
NSC 0.65 0.50 0.52 0.61 0.55 0.42 0.52 0.46
R R  0.804.84  0.703.73  0.705.75  00.8.811  00..7744  00.6.688  0.07.47 4 0.700.7 0
RMSE 172.10 207.15 204.49 183.52 197.35 222.81 204.11 214.66
RMSE  172.10  207.15  204.49  183.52  197.35  222.81  204.11  214.66 
TThhee bbooxxpplloottss ooff mmoonntthhllyy mmeeaann ssiimmuullaatteedd ddiisscchhaarrggee ((FFiigguurree 44)) aatt tthhee oouuttlleett ssttaattiioonn ddeerriivveedd ffrroomm 
tthhee eeigighhttm mododeleslsin idnidcaicteatsei gsniigfinciafinctavnat rivaatiroiantsiofnrosm frAompr ilAtporSile ptote mSebpetre.mThbeerla. rTgheed ilfafergreen cdeisffseuregngceesst 
tshuagtgeevste nthfaotr esvimenu lfaotre dsimdiusclahtaerdg edaistcthhaergoeu talet tthstea toiountl,ewt shtiacthioanr,e wadhijuchst eadreb aydtjhuestaedu tob-yc atlhibe raatuitoon‐
sccahliebmraet,iothne secihgehmtme, othdee lesiegxhht imbiotddeelvsi aetxihonibsi,ta dsecvleiaatriloynisll, uasst rcaleteadrlyb yilltuhsetpraetaekdfl boyw t,hwe hpiecahki sflionwfl,u wenhciecdh 
bisy inmfliuxeendcpedre bciyp mitaitxieodn parnedcispnitoawtiomne altn.dI fsnthoewimnpeultt. Idfa tthaei isnepvuatl udaattead isb eavsaelduaotnedo nblaysetdh eosni monullya ttehde 
rsuimnouflfa,ttehdi sruevnaolfuf,a tthioisn ewviallluaaftfieocnt twheillu anfdfeecrts ttahned uinngdeorfsotatnhderinwg aotef robthaelar nwceatceorm bpaloannecne tcsonmegpaotniveenltys. 
Aneltgeartniavteilvye.l yA,lvtearrniaabtiilviteileys, ivnatrhieabdiilsittireibs uinte tdhoe udtipsutrtibaruetemdo oreutapvuatil aarbele m[4o0r]e. aAvcaciolardbilne g[4ly0,]t.h Aecdciosrtrdibinugtleyd, 
wthaet edrisctormibpuotende nwtsarteerp rceosmenptoanbenetttse rrewpareysteonte vaa lbueattteert hweaeyff etcot seovfaliunaptue ttdhaet aefufnecctesr toafi nitniepsuot ndtahtea 
suinmcuerlatatiionntieosf ohny dthroel soigmicuallaptiroonc eosfs hesy.drological processes. 

FFiigguurree 44.. BBooxxpplloottss ooff mmoonntthhllyy mmeeaann ddiisscchhaarrggeess aatt tthhee oouuttlleett ssttaattiioonn ddeerriivveedd ffrroomm eeiigghhtt mmooddeellss ffrroomm 
22000033 ttoo 22000099..
Water2016,8, 181 9of15
5.2. SensitivitiesofWaterComponents
Water 2016, 8, 181  9 of 16 
Inthisstudy,thehypothesistestswereaimedatdifferentiatingwatercomponentsbasedontheir
annualmeanvaluesandeachgrouptestdatasetwasfirstadoptedbythenormaldistributionand
5.2. Sensitivities of Water Components 
variance homogeneity test. In the MIKE SHE model, ET was divided into different categories:
a
In this study, the hypothesis tests were aimed at differentiating water components based on their 
snow sublimation (SNOWS), canopy interception (CI), river and pond water evaporation (WE),
annual mean values and each group test dataset was first adopted by the normal distribution and 
soilevaporation(SOILE)andplanttranspiration(PT).ThesefiveET sources,overlandflow(OLF),
variance homogeneity test. In the MIKE SHE model, ETa was divided inato different categories: snow 
baseflow(BF)andsnowstorage(SS)wereemployedforthehypothesistest. Theprobabilitypvalue
sublimation (SNOWS), canopy interception (CI), river and pond water evaporation (WE), soil 
resultsareprovidedinTable6.
evaporation (SOILE) and plant transpiration (PT). These five ETa sources, overland flow (OLF), base 
flow (BF) and snow storage (SS) were employed for the hypothesis test. The probability p value 
Table6.TheprobabilitypvaluesofthehypothesisH test.
results are provided in Table 6.  0
Groups A B C A*B A*C B*C A*B*C
Table 6. The probability p values of the hypothesis H0 test. 
OLF 0.000 0.221 0.691 0.040 0.714 0.399 0.336
Groups  A  B C A*B A*C B*C A*B*C 
BF 0.003 0.436 0.008 0.087 0.619 0.450 0.553
OLF  0.000  0.221  0.691  0.040 0.714  0.399  0.336 
SS 0.003 0.000 0.945 0.001 0.828 0.415 0.430
BF  0.003  0.436  0.008 0.087  0.619  0.450  0.553 
SNOWS 0.271 0.000 0.224 0.789 0.546 0.120 0.282
CI 0.00S0S  00.0.00030  0.0000.000.9045  0.000.10070.828  00..048195  0.4300. 028 0.431
WE 0S.0N0O0WS  00.2.07010  0.0000.000.2024  0.708.99 370.546  00..018250  0.2802. 072 0.573
SOILE 0.13C8I  00.0.20308  0.0000.000.0000 0.000.77470.089  00..204218 0.4301. 739 0.932
PT 0.54W1E  00.0.50305  0.0000.000.0000 0.903.77 510.085  00..603772  0.5703. 809 0.905
SOILE  0.138  0.238  0.000 0.747  0.241  0.739  0.932 
PT  0.541  0.535  0.000 0.751  0.637  0.809  0.905 
Table 6 indicates that factor A, precipitation, had a significant effect on most hydrological
componeTnatbsl,ee x6c einpdticSaNteOs WthaSt, SfaOctIoLrE Aa,n pdrePcTip.iFtaatciotonr, Bh,adte ma psiegrnaitfuicraen,t pelfafyecst aocnr umcoiaslt rhoylderionlotghiecasln ow,
PIancdomWpEopneronctse,s esxecse.pAt lSlNevOaWpoSt, rSaOnIsLpEir aantido nPTs.o Fuarccteosr eBx,c teepmtpSeNraOtuWreS, pwlaeyres ast crrouncgialyl rsoelne siint itvhee tsonfoawct, orC,
PET.PTI harnodu WghE tphreocsetsasteiss.t Aiclal lehvayppoottrhanesspisirtaetsiotsn, stohuersceesn esxitcievpitt iSeNsOofWwSa wteerrec sotmropnoglnye snetnssittoivdei tfofe fraecntotrd  ata
typesCw, PeEreT.i lTluhsrotruagthed thine tsutaittiiostnicaalll yh.yBpaostehdesoisn ttehsets,r ethsue lstse,ntshiteivditiisetsr iobfu wteadteor uctopmuptoSnSencatsn tbo edsifpfeerceifinte dto
analydsaetap tryepceips itwaetiroe nilalunsdtratetemdp ienrtautituiorenadlleyv. iBataisoends ofnro tmheR rSeDsulatns,d thSeB Ddi.sPtrTibuwteads eomutppluoty SeSd cinant hbee PET
specified to analyse precipitation and temperature deviations from RSD and SBD. PT was employed 
deviationanalysis.
in the PET deviation analysis. 
5.3. ResponsesofHydrologicalProcesses
5.3. Responses of Hydrological Processes 
5.3.1. SnowStorage
5.3.1. Snow Storage 
Figure5illustratesthespatialdistributionofthesimulatedannualmeansnowstoragedepths
Figure 5 illustrates the spatial distribution of the simulated annual mean snow storage depths 
fromtheSTA,TRMM,LSTandGPETmodels. Whenthesignificantimpactfactors,precipitationand
from the STA, TRMM, LST and GPET models. When the significant impact factors, precipitation and 
temperature associated with snow storage (Table 5) were replaced by RSD in the TRMM and LST
temperature associated with snow storage (Table 5) were replaced by RSD in the TRMM and LST 
modemlso,dtehles, sthpea tsipaaltdiails dtriisbtruibtiuotnioonf otfh tehes nsnoowwppaacckk uunnddeerrwweenntt aa hhuugge echcahnagneg. eIn. Ianddaidtidointi, othne, ttehmepteomrapl oral
distridbiusttriiobnustiSonSsi nSSd iifnf edriefnfetreenletv ealteivoantibonan bdasn(dFsi g(Fuirgeu6re) a6l)s aolseox hexibhiitbeidtedsi gsniginfiicfaicnatnvt avrairaitaitoionnssa ammoonngg STA,
TRMSMTAa,n TdRLMSMT. and LST. 

FiguFreig5u.reS 5p. aStpiaatliadli dstirstirbiubutitoionnss ooff ssiimmuulalatetedd anannunaul amlemane sannowsn sotowrasgteo braagseedb oans tehde SoTnAt,h TeRSMTMA,, LTSRTM  M,
and GPET models in the Yarkant River basin from 2003 to 2009. 
LSTandGPETmodelsintheYarkantRiverbasinfrom2003to2009.
Water2016,8, 181 10of15
Water 2016, 8, 181  10 of 16 

FigFuirgeu6re. T6.h Tehtee mtempoproarlald disistrtribibuuttiioonnss ooff ssnnooww ssttoorraagge eini ndidffieffreernetn etleevleavtiaotni obnanbdasn. ds.
Compared to the STA model, the TRMM model resulted in significant variations in annual

mean snow storage depth, with change proportions of ´9.4% and 12.3% in the 5000 m–5700 m