Hydrol.EarthSyst.Sci.,21,669–684,2017 www.hydrol-earth-syst-sci.net/21/669/2017/ doi:10.5194/hess-21-669-2017 ©Author(s)2017.CCAttribution3.0License. Review article: Hydrological modeling in glacierized catchments of central Asia – status and challenges YaningChen,WeihongLi,GonghuanFang,andZhiLi StateKeyLaboratoryofDesertandOasisEcology,XinjiangInstituteofEcologyandGeography, ChineseAcademyofSciences,Urumqi830011,China Correspondenceto:YaningChen([email protected]) Received:27June2016–PublishedinHydrol.EarthSyst.Sci.Discuss.:25July2016 Revised:17December2016–Accepted:22December2016–Published:2February2017 Abstract. Meltwater from glacierized catchments is one of et al., 2014). In fact, in the alpine river basins of the north- themostimportantwatersuppliesincentralAsia.Therefore, ernTienshans,glaciermeltwatercontributes10%ofannual theeffectsofclimatechangeonglaciersandsnowcoverwill runoffand20%ofrunoffduringthedroughtyears(Aizenet have increasingly significant consequences for runoff. Hy- al.,1997);therefore,climate-drivenchangesinglacier/snow- drological modeling has become an indispensable research fedrunoffregimeshavesignificanteffectsonwatersupplies approach to water resources management in large glacier- (Immerzeeletal.,2010;Kaseretal.,2010). ized river basins, but there is a lack of focus in the mod- According to a study conducted by the Eurasian Devel- eling of glacial discharge. This paper reviews the status of opment Bank, changes in temperature and precipitation in hydrological modeling in glacierized catchments of central central Asia have led to rapid regression in glaciers (Ibat- Asia, discussing the limitations of the available models and ullin et al., 2009). The overall decrease in total glacier area extrapolatingthesetofuturechallengesanddirections.After andmassfrom1961to2012tobe18±6%and27±15%, reviewing recent efforts, we conclude that the main sources respectively. These values correspond to a total area loss of ofuncertaintyinassessingtheregionalhydrologicalimpacts 2960±1030km2, and an average glacier mass change rate ofclimatechangearetheunreliableandincompletedatasets of −5.4±2.8Gtyr−1 (Farinotti et al., 2015). If the warm- and the lack of understanding of the hydrological regimes ingprojectionsdevelopedbytheIntergovernmentalPanelon of glacierized catchments of central Asia. Runoff trends in- ClimateChange(IPCC)provetobetrue,theglacierizedriver dicateacomplexresponsetochangesinclimate.Forfuture systems in central Asia will undergo unfavorable hydrolog- variationofwaterresources,itisessentialtoquantifythere- ical changes, e.g., altered seasonality, increased flood risk, sponsesofhydrologicprocessestobothclimatechangeand higherandintensespringdischargeandwaterdeficiency,in shrinking glaciers in glacierized catchments, and scientific hotanddrysummerperiods,especiallygiventhesharprise focusshouldbeonreducinguncertaintieslinkedtothesepro- in water demand (Hagg et al., 2006; Siegfried et al., 2012). cesses. The development of hydrological models on accounting for changes in current and future runoff is therefore crucial for water resources allocation in river basins, and includes un- derstandingclimatic variabilityas wellas theimpact ofhu- 1 Introduction manactivitiesonclimate(Bierkens,2015). Hydrological modeling is an indispensable approach to Climate change is widely anticipated to exacerbate water water resources research and management in large river stress in central Asia in the near future (Siegfried et al., basins. Such models help researchers understand past and 2012),asthevastmajorityofthearidlowlandsintheregion current changes and provide a way to explore the implica- are highly dependent on glacier meltwater supplied by the tions of management decisions and imposed changes. The TienshanMountains,whichareknownasthe“watertower” purposeofhydrologicalmodelingonbasinscaleisprimarily of central Asia (Hagg et al., 2007; Sorg et al., 2012; Lutz PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion. 670 Y.Chenetal.:HydrologicalmodelinginglacierizedcatchmentsofcentralAsia tosupportdecision-makingforwaterresourcesmanagement, thetotalrunoffgeneratedfromthemeltingofglaciers(snow which can be summarized as resource assessment, vulnera- and glacier), but can also include liquid precipitation on bilityassessment,impactassessment,floodriskassessment, glacierized areas (Unger-Shayesteh et al., 2013). A large prediction and early warning (World Meteorological Orga- numberofhydrologicalmodelsappliedinglacierizedcatch- nization, 2009). It is important to choose the most suitable mentsofcentralAsiaarebasin-scalemodels,whichcontain hydrological model for a particular watershed based on the empiricalhydrologicalmodelsaswellasphysicalhydrolog- area’sclimate,hydrologyandunderlyingsurfaceconditions. icalmodels(Table2).Theseglacio-hydrologicalmodelsare The Tienshan Mountains span several countries and sub- usefultoolsforanticipatingandevaluatingtheimpactsofcli- regions, creating a decentralized political entity of complex matechangesintheheadwatercatchmentsofthemainAsian multi-nationalandmulti-ethnicforms.Therearethreelarge rivers(Milleretal.,2012). transboundary international rivers that originate in the high mountains of central Asia. In an international river, hydro- 2.1 Currentandfuturerunoffchanges logicalchangesarerelatedtotheinterestsoftheabuttingri- pariancountries(StarodubtsevandTruskavetskiy,2011;Xie Riverrunoffrespondsinacomplexwaytovariationsincli- etal.,2011;Guoetal.,2015).However,asconflictsbetween mate and the cryosphere. At the same time, runoff changes political states may arise for any number of reasons (politi- alsodependondominantrunoffcomponents.Table2shows cal, cultural, etc.), transboundary issues may result in frag- that annual runoff anomalies have increased to some ex- mentedresearchandthuslimitthedevelopmentofhydrolog- tent (except in the western Tienshan Mountains) and in- icalmodeling. consistencies between changes in precipitation and runoff Amid this potential hindrance to robust research efforts, have occurred in heavily glacierized catchments. In rivers theeffectofclimatechangeonglaciers,permafrostandsnow fed by snow and glaciers, runoff has increased (e.g., in cover is having increasing impacts on runoff in glacierized northernTienshanMountains)andrisingtemperaturesdom- central Asian catchments. However, solid water is seldom inate the runoff changes by, for instance, increasing the explicitly considered within hydrological models due to the snowmelt/glacier melt and decreasing snowfall fraction (ra- lackofcompleteglacierdata.Ourknowledgeofsnow/glacier tio of solid precipitation to liquid precipitation) (Chen, changesandtheirresponsestoclimateforcingisstillmostly 2014). Khan and Holko (2009) compared runoff changes incomplete. Analysis of current and future water resources withvariationsinsnowcoverareaandsnowdepth.Theysug- variationsincentralAsiamaypromoteadaptationstrategies gestedthatthemismatchbetweendecreasingtrendsinsnow toalleviatethenegativeimpactsofexpectedincreasedvari- indicatorsandtheincreasingriverrunoffcouldbetheresult abilityinrunoffchangesresultingfromclimatechange. ofenhancedglaciermelting.Heavilyglacierizedriverbasins In this paper, we review hydrological modeling efforts in showedmainlypositiverunofftrendsinthepastfewdecades fivemajorriverbasinsoriginatingfromtheTienshanMoun- (simulatedunderdifferentscenariosintheheadriversofthe tainsincentralAsia,namely,theTarimRiverbasin,thewa- TarimRiverbasin),whilethosewithlessornoglacierization tersheds in the northern slope of the Tienshan Mountains exhibitedwidevariationsinrunoff(Duethmannetal.,2015; (which includes several small river basins), the Issyk Lake Kaldybayevetal.,2016). basin,theIliRiverbasin,andtheAmuDaryaandSyrDarya With further warming and the resulted acceleration of basins (Fig. 1). Their topographical characteristics, climate glacier retreat, glacier inflection points will or have already and vegetation together with the glacierized area are listed appeared.Theamountofsurfacewaterwillprobablydecline inTable1.Weexaminethetypes,purposeanduseofexist- orkeephighvolatilityduetoglacialretreatandreducedstor- ingmodelsandassesstheconstraintsandgapsinknowledge. age capacity of glaciers (Chen et al., 2015). For instance, Thecurrentlackofunderstandingofhigh-altitudehydrolog- near-futurerunoffsareprojectedtoincreasetosomeextent, icalregimesiscausinguncertaintyinassessingtheregional withincrementsof13–35%during2011–2050comparedto hydrologicalimpactsofclimatechange(Milleretal.,2012). 1960–2006fortheYarkandRiver,−1–18%inthe21stcen- Snowandglacialmeltassuppliesofsolidwaterisakeyel- tury compared to 1986–2020 under RCP4.5 for the Kaidu ementinstreamflowregimes(Lutzetal.,2014);therefore,it River, and 23% in 2020 for the Hotan River (Table 2). For isnecessarytoincludeglaciermassbalanceestimatesinthe thelong-term,however,totalrunoffisprojectedtobesmaller model calibration procedure (Schaefli et al., 2005; Stahl et than today. The hydrological responses to climate change al.,2008;KonzandSeibert,2010;Mayretal.,2013). aroundtheworldwerediscussedinSect.2.3. 2.2 Contributionofglaciermelt/snowmeltwaterin 2 Modelinghydrologicalresponsestoclimatechange riverrunoff Changes in the amount and seasonal distribution of river Kemmerikh(1972)estimatedthecontributionofgroundwa- runoff may have severe implications for water resources ter,snowmeltandglaciermelttothetotalrunoffofthealpine management in central Asia. “Glacier runoff” is defined as rivers in central Asia. Based on the hydrograph separation Hydrol.EarthSyst.Sci.,21,669–684,2017 www.hydrol-earth-syst-sci.net/21/669/2017/ Y.Chenetal.:HydrologicalmodelinginglacierizedcatchmentsofcentralAsia 671 Figure1.MapofcentralAsianheadwaterswithmainriverbasinsorhydrologicalregions,namely,theTarimRiverbasin,thewatershedsin thenorthernslopeoftheTienshanMountains,theIssykLakebasin,theIliRiverbasin,andtheAmuDaryaandtheSyrDaryabasins.Lake outlinesarefromNaturalEarth(http://www.naturalearthdata.com/).TheriversystemisderivedbasedonelevationsoftheSRTM(Shuttle RadarTopographyMission)90mdata.GlacierinformationwasobtainedfromRGI(RandolphGlacierInventory). Table1.Summaryofclimaticandunderlyingconditionsofthebasins.ThetopographyisbasedonSRTMdata,glacierdataarefromRGI (RandolphGlacierInventory)andclimateisbasedontheworldmapoftheKöppen–Geigerclimateclassification.Vegetationisfromthe landusedatafromXinjianginstituteofEcologyandGeography. Catchment Tarim Catchments Issyk Ili Amu Syr River innorthern Lake River Darya Darya basin TienshanMountains, basin basin Basin Basin China Location Surroundedbythe Northern Western Western Western Western TienshanMountainsand Tienshan Tienshan Tienshan Tienshanand Tienshan theKunlunMountains Valley Pamir Topography Basinarea(km2) 868811 126463 102396 429183 674848 442476 Percentageofelevation 28.00 13.80 14.50 4.60 20.50 9.50 >3000m(%) Glaciationarea(km2) 15789 1795 994 2170 9080 1850 Climate Dominantclimate arid arid arid arid arid arid cold cold cold; cold; cold; cold continental continental snow Vegetation Forestpercent(%) 0.7 10.4 6.4 4.1 10.9 2.5 Pasturepercent(%) 16.7 14.5 31.2 28.6 19.4 17.3 Percentofwater,snow,ice(%) 5.4 3.9 7.8 5.3 5.3 2.8 methodology,theglaciermeltcontributionrangedbetween5 catchments in the Tienshan Mountains based on hydrologi- and40%intheplainsandaround70%inupstreambasins. calmodeling(Y.Zhangetal.,2016). The ratio of glacier melt contribution to runoff varies be- Distributed hydrological models provide a more useful tween 3.5 and 67.5% with a mean of 24.0% for the 24 toolfortheinvestigationofchangesindifferentrunoffcom- ponents.Forexample,thevariableinfiltrationcapacity(VIC) www.hydrol-earth-syst-sci.net/21/669/2017/ Hydrol.EarthSyst.Sci.,21,669–684,2017 672 Y.Chenetal.:HydrologicalmodelinginglacierizedcatchmentsofcentralAsia Table2.SummaryofhydrologicalmodelinginglacierizedcentralAsiancatchments. Catchments Models Majorconclusions Innovationsandlimitations References TarimRiver TarimRiverbasin Modified two-parameter Improvedtheoriginaltwo-parametermonthlywaterbal- Lessinputdataarerequired;Lackof PengandXu(2010); semi-distributed water ancemodelbyincorporatingthetopographicindexesand glacierandsnowmeltprocesses. Chenetal.(2006) balancemodel couldgetcomparableresultstotheTOPMODELmodeland Xinanjiangmodel. TOPMODELmodel IntheAksuRiver,runoffwasmorecloselyrelatedtopre- cipitation,whereasintheHotanRiver,itwasmoreclosely relatedtotemperature. Xinanjiangmodel RunoffsoftheAksu,YarkandandHotanriversexhibited increasingtendenciesin2010and2020underdifferentsce- nariosgeneratedfromthereferenceyears,e.g.,23%in- creasefortheHotanRiver. PPR(Projectionpursuitre- Iftemperaturerises0.5–2.0◦C,runoffwillincreasewith Lackofphysicalbasis. Wuetal.(2003) gression)model temperaturefortheAksu,YarkandandHotanrivers. VIC(variableinfiltrationca- FortheTarimRiver,runoffwilldecreaseslightlyin2020– Lackofglaciermodule. Liuetal.(2010) pacity) 2025basedonVICforcedbyHadCM3underA2andB2 whennotconsideringglaciermelt. TailanRiver Modifieddegree-daymodel Glacier runoff increases linearly with temperature over Consideredtheeffectofsolarradiation Y.Zhangetal.(2007) includingpotentialclearsky theserangeswhetherornotthedebrislayeristakeninto andquantifiedthedebriseffect. direct solar radiation cou- consideration.Theglacierrunoffislesssensitivetotemper- pledwithalinearreservoir aturechangeinthedebris-coveredareathanthedebris-free model area. Aksu River includ- Xinanjiangmodel PrecipitationhasaweakrelationshipwithrunoffintheKu- Joinedthesnowmeltmodule. P.Wangetal.(2012) ing Kumalike and malikeRiver. Toxkanrivers Themodelcouldnotwellcapturethe snowmelt-/precipitation-induced peak streamflow. VIC-3Lmodel Glaciermelt,snowmeltandrainfallaccountedfor43.8,27.7 Themodelperformancewasobviously Zhaoetal.(2013,2015) and28.5%ofthedischargefortheKumalikeRiverand improved through coupling a degree- 23.0,26.1and50.9%fortheToxkanRiver. dayglaciermeltscheme,butaccurately estimatingarealprecipitationinalpine regionsstillremains. FortheKumalikeRiverandtheToxkanRiver,therunoff hasincreased13.6and44.9%during1970–2007,and94.5 and100%oftheincreaseswereattributedbyprecipitation increase. FortheKumalikeRiver,glacierareawillreduceby>30% resultingindecreasedmeltwaterinsummerandannualdis- charge(about2.8–19.4%inthe2050s). SWIMmodel Themodeliscapabletoreproducethemonthlydischargeat Investigated the glacier lake outburst Huangetal.(2015); thedownstreamgaugewell,usingthelocalirrigationinfor- floodsusingamodelingtool.Inclusion Wortmannetal.(2014) mationandtheobservedupstreaminflowdischarges. ofanirrigationmoduleandarivertrans- mission losses module of the SWIM model. About18%oftheincomingheadwaterresourcesconsumed Modeluncertaintiesarethelargestinthe uptothegaugeinXidaqiao,andabout30%additionalwa- snowmeltandglaciermeltperiods. terisconsumedbetweenXidaqiaoandAlar. Differentirrigationscenariosweredevelopedandshowed thattheimprovementofirrigationefficiencywasthemost effectivemeasureforreducingirrigationwaterconsumption andincreasingriverdischargedownstream. WASAmodel Glaciermeltcontributesto35–48and9–24%fortheKu- The model considered changes in Duethmannetal.(2015) malikeRiverandtheToxkanRiver. glaciergeometry(e.g.,glacierareaand surfaceelevation). FortheKumalikeRiver,glaciergeometrychangesleadto It used a multi-objective calibration areductionof14–23%ofstreamflowincreasecomparedto basedonglaciermassbalanceanddis- constantglaciergeometry. charge. Thetemperatureandprecip- TheAR(p)modeliscapableofpredictingthestreamflowin AR(p)needslesshydrologicalandme- Ouyangetal.(2007) itationrevisedAR(p)model; theAksuRiverbasinwhiletheNAMmodelisnotideal. teorological data. Both model fails to modelsuddenfloodssuchasicedam collapsefloods. NAM (NedborAfstromn- ingsModel)rainfallrunoff model Hydrol.EarthSyst.Sci.,21,669–684,2017 www.hydrol-earth-syst-sci.net/21/669/2017/ Y.Chenetal.:HydrologicalmodelinginglacierizedcatchmentsofcentralAsia 673 Table2.Continued. Catchments Models Majorconclusions Innovationsandlimitations References TarimRiver KaiduRiver MIKE-SHEmodel Comparedremotesensingdataandstation-baseddatain Missingglaciermelt;Lackofobserva- T.Liuetal.(2012); simulatingthehydrologicalprocesses.Remotesensingdata tiontoverifythemeteorologicalcondi- Liuetal.(2013) arecomparabletoconventionaldata.Remotesensingdata tioninthemountainousregions. couldpartlyovercomethelackofnecessaryhydrological modelinputdataindevelopingorremoteregions. HBV(HydrologiskaByrans Whenthebaserunoffis100m3s−1,thecriticalrainfallfor Itunderestimatedthepeakstreamflow Fanetal.(2014) Vattenbalansavdelning) primaryandsecondarywarningfloodsare50and30mm, whileoverestimatedthebaseflow. model respectively,fortheKaiduRiver. SRM (snowmelt runoff Spring streamflow is projected to increase in the future Limited observations resulted in low Y.C.Zhangetal.(2007); model) including potential basedonHadCM3. modelingprecision.TheAPHRODITE Ma et al. (2013); Li et clearskydirectsolarradia- precipitationperformedwellinhydro- al.(2014). tionandtheeffectiveactive logicalmodelingintheKaiduRiver. temperature. SWAT Precipitationandtemperaturelapseratesaccountfor64.0% Quantifieduncertaintyresultedfromthe Fangetal.(2015a,b) ofmodeluncertainty. meteorologicalinputs. Runoffincreases(−1)–18and4–20%inthe21stcentury underRCP4.5andRCP8.5comparedto1986–2005based onacascadeofregionalclimatemodelRegCM,biascor- rectionandSWATmodel. Modified system dynamics Simulationsoflowflowandnormalflowaremuchbetter Applied the effective cumulative tem- F.Y.Zhangetal.(2016) model thanthehighflow,andspringpeakflowisbetterthanthe peraturetocalculatesnowmeltprocess summerpecksintheKaiduRiver. andsoiltemperatureforeachlayerto describewatermovementinsoil. YarkandRiver MIKE-SHEmodel Simulatedsnowpackusingstationdatadifferssignificantly Lackofglaciermodule Liuetal.(2016b) fromthatusingremotesensingdata. Integrating Wavelet Anal- Runoffpresentedanincreasingtrendsimilarwithtempera- Interpreted the nonlinear characteris- Xuetal.(2014) ysis (WA) and back- tureandprecipitationatthetimescaleof32years.Butatthe ticsofthehydro-climaticprocessusing propagationartificialneural 2-,4-,8-,and16-yeartimescale,runoffpresentednonlinear statisticmethod. network(BPANN) variation. Degree-daymodel Decreasingrateofglaciermasswas4.39mma−1result- Theglacierdynamicsisconsideredand Xieetal.(2006);Zhang inginarunoffincreasingtrendof0.23×108m3a−1dur- thearea–volumescalingfactoriscali- etal.(2012a,b) ing1961–2006.Sensitivityofmassbalancetotemperature bratedusingremotesensingdata. is0.16mma−1◦C−1. Glacier runoff will increase 13–35% during 2011–2050 comparedto1960–2006withobviousincreaseinsummer. Tizinapu SRMincludingsnowalbedo ItcouldwellsimulatetherunoffoftheTizinapuRiver. Lackofglaciermodule. LiandWilliams(2008) Runoffisdominatedbyprecipitationandtemperaturelapse rates,andsnowalbedo. HotanRiver Integrating Wavelet Anal- Runoffcorrelateswellwiththe0◦Clevelheightinsummer Interpretedthenonlinearcharacteristics Xuetal.(2011) ysis (WA) and back- forthenorthernslopeofKunlunMountains. ofthehydro-climaticprocess. propagationartificialneural network(BPANN) CatchmentsinnorthernslopeofTienshanMountains,China ManasRiver (1)SWATmodel Glacier area decreased by 11% during 1961–1999 and Boththeglaciermeltmoduleandtwo- Yu et al. (2011); Luo glaciermeltcontributes25%ofdischarge. reservoirmethodwereincludedinthe et al. (2012); Luo et hydrologicalsimulations. al. (2013); Gan and Luo(2013) (2)SRMmodel Bettersimulationofsnowmeltrunoffthanrainfall–runoff Snow cover calculation algorithm is Yuetal.(2013) bytheSRM. addedtovalidatemodelperformance. (3)EasyDHMmodel EasyDHMmodelcouldreproducethestreamflow. Thevalidationisbasedonstreamflow alone. Xingetal.(2014) UrumqiRiver (1)Isotopehydrographsep- Glaciermeltwatercontributesto9%ofrunoff. TheIHSmethodhasoverwhelmingpo- KongandPang(2012) aration(IHS) tentialinanalyzinghydrologicalcom- ponentsforungaugedwatersheds. (2)Waterbalancemodel Thecumulativemassbalanceoftheglacierwas-13.69m Foused on runoff generation on the Sunetal.(2013) during1959–2008;proportionofglacierrunoffincreased glacierizedandablationarea. from62.8to72.1%. www.hydrol-earth-syst-sci.net/21/669/2017/ Hydrol.EarthSyst.Sci.,21,669–684,2017 674 Y.Chenetal.:HydrologicalmodelinginglacierizedcatchmentsofcentralAsia Table2.Continued. Catchments Models Majorconclusions Innovationsandlimitations References CatchmentsinnorthernslopeofTienshanMountains,China (3)HBVmodel Foraglacierizedcatchment(glacierizationratiois18%), Consideringfuturerunoffunderdiffer- Sunetal.(2015) thedischargewillincreaseby66±35%ordecreaseby entglacierchangescenarios. 40±13%iftheglaciersizekeepsunchangedorglacierdis- appearsin2041–2060. (4)Exponentialregression Glacierrunoffiscriticallyaffectedbythegroundtempera- Thisstudyshedlightonglacierrunoff Chenetal.(2012) ture. estimationbasedongroundtemperature fordata-scarceregions. (5)SRMmodel Thedegreedayfactorisnotconstantfordifferentelevation Calculated the curve of snow cover Huaietal.(2013) bands. shrinkagebasedonMODISdata. (6) THModel (Thermody- THModel can indeed simulate runoff processes in the Anenergybalancemodelisproposedto Mouetal.(2008) namicWatershedHydrolog- glacierandsnow-dominatedcatchmentreasonablywell. closethebalanceequationofsoilfreez- icalModel) ingandthawing. Ebinur Lake catch- (1)SWATmodelandthese- For the Jinghe River, 85.7% of the runoff reduction is Identifiedtheeffectsofhumanactivities Dongetal.(2014); mentincluding quentialclustermethod causedbyhumanactivityand14.3%byclimatechange. andclimatechangeonrunoff. Yaoetal.(2014) Jinghe River, Kuy- tun River and Bor- talaRiver (2)RunoffCAR(Controlled TheJingheRiverandKuytunRiverexhibitedaslightlyin- TheCARisbasedonpastandpresent AutoRegressive)model creasingtrend,butanadversetrendintheBortalaRiver. valueswithoutphysicalbasis. Inawarmhumidscenario,runoffintheJingheRiverand BortalaRiverwillincreasewhileitwilldecreaseinthe KuytunRiver. JuntanghuBasin DHSVM (Distributed Hy- The coupled WRF (Weather Research and Forecasting) MODIS snow cover and the calcu- Zhaoetal.(2009) drology Soil Vegetation modellingsystemandDHSVMmodelcouldpredict24h latedsnowdepthdataareusedinthe Model) snowmeltrunoffwithrelativeerrorwithin15%. snowmeltrunoffmodeling. IssykLakeBasin Small rivers around Degree-dayapproach Runoffcontributionisvaryinginabroadrangedepend- Theglaciermeltrunofffractionatthe DikichandHagg(2003) theIssykLake ingonthedegreeofglacierizationintheparticularsub- catchment outlet can be considerably catchment.Allriversshowedarelativeincreaseinannual overestimated. riverrunoffrangingbetween3.2and36%. ChuRiver SWAT-RSG (RSG: rain, Generaldecreasewasexpectedinglacierrunoff(−26.6 Usetheglacierdynamicsandassessed Maetal.(2015) snowandglacier)model to−1.0%),snowmelt(−21.4to+1.1%)andstreamflow themodelperformancebasedonboth (−27.7to−6.6%);Peakstreamflowwillbeputforward streamflowandglacierarea. for1month. IliRiverbasin GongnaisiRiver SRMmodel Runoffissensitivetosnowcoverareaandtemperature. SRMiscapabletomodelthesnowmelt MaandCheng(2003) runoff. Iftemperatureincreases4◦C,therunoffwilldecreaseby 9.7%withsnowcoverageandrunoffshiftingforward. TekesRiver SWATmodel Glaciershaveretreatedabout22%since1970s,whichwas Using two land use data and two Xuetal.(2015) considerablyhigherthantheTienshanaverage(4.7%)and Chineseglacierinventories,themodel Chinaaverage(11.5%),resultinginadecreaseofpropor- couldwellreproducestreamflow. tionofprecipitationrechargedrunofffrom9.8%in1966– 1975to7.8%in2000–2008. IliRiver DTVGM (multi-spatial Dailyrunoffcorrelatedcloselywithsnowmelt,suggesting This method has less dependence on Caietal.(2014) data-based Distributed asnowmeltmoduleisindispensable. conventionalobservation. Time-Variant Gain Model) model Waterbalancemodel Water decrease in 1911–1986 in the middle and lower – KezerandMatsuyama reachesoftheLakeBalkhashisduetodecreasedrainfall (2006);Guoetal.(2011) andreservoirsstorage. AmuDaryaandSyrDaryabasins AmuDaryaandSyr STREAM TherunoffoftheSyrDaryadeclinedconsiderablyoverthe Simulatedlong-termdischargeforthe Aertsetal.(2006) Darya last9000years,butshowmuchsmallerresponsestofuture Holoceneandfutureperiod. warming. For the Amu Darya and Syr Darya basins, the glacier- Themodelincludesthecalculationof Savoskul et al. (2003); coveredareashavedecrease15and22%in2001–2010 rainwater,snowmeltwaterandglacier Savoskul et al. (2004); comparedtothebaseline(1960–1990). runoff(basedontheglacieraltitudeand SavoskulandSmakhtin equilibriumlimealtitude). (2013) FortheAmuRiverBasin,20–25%oftheglacierswillre- tainunderatemperatureincrementbeing4–5◦Candpre- cipitationincreaseratebeing3%/◦C. FortheSyrDarya,runoffundertheA2andB2scenarios willincrease3–8%in2010–2039,withsharpenedspring peakandaslightloweredrunofffromlateJunetoAugust. Hydrol.EarthSyst.Sci.,21,669–684,2017 www.hydrol-earth-syst-sci.net/21/669/2017/ Y.Chenetal.:HydrologicalmodelinginglacierizedcatchmentsofcentralAsia 675 Table2.Continued. Catchments Models Majorconclusions Innovationsandlimitations References AmuDaryaandSyrDaryabasins AralMountainmodel FortheAmuDarya,glaciermeltandsnowmeltcontribute Fully simulated the hydrological pro- Immerzeeletal.(2012a) to38and26.9%ofrunoff,whilefortheSyrDaryathepro- cesses. portionsare10.7and35.2%. Glacierwillretreatby46.4–59.5%by2050dependingon selectedGCM(GeneralCirculationModel).FortheSyr Darya,averagewatersupplytothedownstreamwillde- creaseby15%for2021–2030and25%for2041–2050. FortheAmuDaryatheexpecteddecreasesare13%(2021– 2030)and31%(2041–2050). Testsites HBV-ETH(HBVmodelex- Overallgoodmodelperformanceswereachievedwiththe Itconsideredgeographical,topographi- Haggetal.(2007) “Abramov”in pandedattheSwissFederal maximumdiscrepancyofsimulatedandobservedmonthly calandhydrometeorologicalfeaturesof SyrDaryaand InstituteofTechnology)and runoffwithin20mm. testsites,andreducedmodelinguncer- “Oigaing”in OEZ(awaterbalanceequa- tainties. AmuDarya tionmodel) General enhanced snowmelt during spring and a higher Thisprocedurerequiresalotmeteoro- floodriskinsummerarepredictedunderadoublingatmo- logicalandlandsurfacedataandknowl- sphericCO2concentrationwithgreatestrunoffincreases edgeofthehydrologicalprocesses. occurringinAugustforthehighlyglaciatedcatchmentsand inJuneforthenivalcatchment. PanjRiver HBV-ETH FortheupperPanjcatchment,thecurrentglacierextentwill Applicationofglacierparameterization Haggetal.(2013) decreaseby36and45%,respectively,assumingtempera- scheme. tureincrementbeing2.2and3.1◦C. NarynRiver SWAT-RSGmodel Glacierareahasdecreased7.3%during1973–2002. Incorporatedglacierdynamicsandvali- Ganetal.(2015) datedthemodelusingtwoglacierinven- tories. Glacierswillrecedewithonly8%ofthesmallglaciersre- tainby2100underRCP8.5andnetglaciermeltrunoffwill reachpeakinabout2040anddecreaselater. SyrDarya NAMmodelwithaseparate Glaciervolumewilllose31%±4%underSRESA2until TheNAMmodelwasimprovedtobero- Siegfriedetal.(2012) land-icemodel 2050s,andtherunoffpeakwillshiftforwardby30–60days bustusingonlyfivefreelycalibratedpa- fromthecurrentspring/earlysummertowardsalatewin- rameters. ter/earlyspringrunoffregime. modelwasusedtocalculatethecomponentsofrunoffinthe tributed to about 11.4–28.7% of the lake level rise in the sourceriverfortheTarimRiver.Theresultsshowedthat,in threeglacier-fedlakes,namely,SilingCo,NamCoandPung termsofrunoff,glaciermeltwater,snowmeltwaterandrain- Co(Leietal.,2013).Analysisfromgroundwaterstoragein- fall accounted for 43.8, 27.7 and 28.5% of the Kumalike dicated that the groundwater for the major basins in the Ti- River, and 23.0, 26.1 and 50.9% of the Toxkan River, re- betan Plateauincreasedduring 2003–2009 with a trend rate spectively(Zhaoetal.,2013);thisresultiscomparabletothe of +1.86±1.69Gtyr−1 for the Yangtze River Source Re- conclusionthat glaciermelt accountsfor31–36% basedon gion and +1.14±1.39Gtyr−1 for the Yellow River Source isotopetracer(Sunetal.,2016).However,accuratelyquan- Region(Xiangetal.,2016). tifyingthecontributionsofglaciermelt,snowmeltandrain- FortheSouthAmericanAndes,meltingattheglaciersum- falltorunoffincentralAsianstreamsischallenging(Unger- mit has occurred. With the continually increase in tempera- Shayestehetal.,2013). ture, although glacier melt was dominated by maybe other processes in some regions, the probability seems high that 2.3 Glacio-hydrologicalresponsestoclimatechange:a the current glacier melting will continue. With the loss of comparison glacierwater,thecurrentdry-seasonwaterresourceswillbe heavilydepletedoncetheglaciershavedisappeared(Barnett To analyze the hydrological responses to climate change of etal.,2005). theglacierizedTienshanMountains,theresponsesofseveral For the Alps, many investigations have been imple- majorglacierizedmountainousregionsarediscussed.Forthe mented,rangingfromglacier-scalemodelingtolargebasin- Himalaya–HinduKushregion,investigationssuggestedthat scale or region-scale modeling (Finger et al., 2015; Ab- aregressionofthemaximumspringstreamflowperiodinthe baspour et al., 2015). Glacier meltwater provided about annual cycle of about 30 days, and annual runoff decreased 5.28±0.48km3a−1offreshwaterduring1980–2009.About by about 18% for the snow-fed basin, whereas it increased 75% of this volume occurred during July–September, pro- byabout33%fortheglacier-fedbasinusingtheSatlujBasin vidingwaterforlargelow-lyingriversincludingthePo,the asatypicalregion(SinghandBengtsson,2005).FortheTi- RhineandtheRhône(Farinottietal.,2016).Underthecon- betanPlateau,theglacierretreatcouldleadtoanexpansion textofclimatechange,decreasesofglaciermeltwaterinboth oflakes;e.g.,glaciermasslossbetween1999and2010con- annualandsummerrunoffcontributionsareanticipated.For www.hydrol-earth-syst-sci.net/21/669/2017/ Hydrol.EarthSyst.Sci.,21,669–684,2017 676 Y.Chenetal.:HydrologicalmodelinginglacierizedcatchmentsofcentralAsia example,annualrunoffcontributionsfrompresentlyglacier- cipitation data set (APPRODITE) to force the SRM model. izedsurfacesareexpectedtodecreaseby13%by2070–2099 Applyinginsituobservationalmeteorologicaldataisalsoas- comparedto1980–2009,despiteofnearlyunchangedcontri- sociatedwithotherchallenges,asdetailedbelow. butions from precipitation under RCP 4.5 (Farinotti et al., 2016). Lackofstations Thehydrologicalprocessesintheglacierizedregionshave Oneofthegreatestchallengesinherentinstation-scalemete- somethingincommon;i.e.,theannualrunoffislikelytore- orologicaldataisthelowdensityofmeteorologicalstations. duce in a warming climate with high spatial–temporal vari- AsthemountainousregionsofcentralAsiaarecharacterized ation at the middle or end of the 21st century. Seasonally, bycomplexterrain,itisinaccuratetorepresenttheclimatic increased snowmelt runoff and water shortage of summer conditions of basins using data from limited stations. Some runoff with the disappearing glaciers are expected. How- researchers (Liu et al., 2016b; Fang et al., 2015a) have ad- ever, there are also differences in the responses of hydro- dressed this challenge by attempting to interpolate temper- logical processes to climate change. For example, the con- ature/precipitation into a basin scale using elevation bands, trasting climate change impact on river flows from glacier- based on the assumption that climate variables increase or izedcatchmentsintheHimalayasandAndes(Ragettlietal., decrease with elevation. Temperature lapse rates could also 2016).IntheLangtangcatchmentinNepal,increasedrunoff bevalidatedusingtheIntegratedGlobalRadiosondeArchive isexpectedwithlimitedshiftsbetweenseasons,whereasfor (IGRA)dataset(LiandWilliams,2008).However,thismod- the Juncal catchment in Chile, the runoff has already been ificationcouldnottakeaccountofthesourceofwatervapor decreasing. These qualitative or quantitative differences are andmountainaspectforbasinswithcomplexlandform.Due mainly caused by glaciation ratio, regional weather pattern tothefactthatuniformprecipitationgradientscannotbede- andglacierproperty(HaggandBraun,2005). rivedandtemperaturelapseratesarenotconstantthroughout However,formanyglacierizedcatchmentintheTienshan theyear(Immerzeeletal.,2014),itisachallengetouseel- Mountains, currently or for the next several decades, the evationbandstointerpolatestation-scaleclimateintobasin- runoffappearstobenormalorevenanincreasingtrend,giv- scaleclimate. ing an illusion of better prospects. It is particularly worth mentioningthat,oncetheglacierstorage(fossilwater)melts Lackofhomogeneitytest away, the water system is likely to go from plenty to want, exacerbatingwaterstressgiventheincreasingwaterdemand. Many hydrological modeling studies do not factor in errors in observations, even though homogeneous climate records arerequiredinhydrologicaldesign.IncentralAsia,changes 3 Limitationsoftheavailablehydrologicalmodels in regulation protocols or relocation of stations also lead to 3.1 Meteorologicalinputsinhydrologicalmodelingand observational errors. Checking the input data should be the prediction firststepinhydrologicalmodelingduetotheruleof“garbage inyieldsgarbageout”. InmountainousregionsofcentralAsia,meteorologicalinput uncertaintycouldaccountforover60%ofmodeluncertainty 3.1.2 Remotesensingdataandreanalysisdata (Fang et al., 2015a). The greatest challenge in hydrological Remote sensing and reanalysis data are increasingly being modelinghasbeenlackofrobustandreliablecompleteme- usedinhydrologicalmodeling.T.Liuetal.(2012)andLiuet teorologicaldata,especiallysincethecollapseoftheSoviet al.(2016b)evaluatedremotesensingprecipitationdataofthe Union in the late 1980s. In this section, the value and limi- TropicalRainfallMeasuringMission(TRMM)andtempera- tations of different data sets used in hydrological modeling turedataofModerateResolutionImagingSpectroradiometer (e.g., station data, remote sensing data) and future predic- (MODIS).Theresultsindicatedthatsnowstorageandsnow- tions (e.g., outputs of GCMs (General Circulation Models) pack that were modeled using the remote sensing climate andRCMs(RegionalClimateModels)arediscussed. are different from those modeled using station-scale obser- 3.1.1 Observationaldata vational data. The model forced by the remote sensing data showedbetterperformanceinspringsnowmelt(T.Liuetal., Traditionally, hydrological models are forced by station- 2012). Huang et al. (2010a) analyzed the input uncertainty scale meteorological data in or near the studied watershed of remote sensing precipitation data interpreted from FY-2. (e.g., Fang et al., 2015a; Peng and Xu, 2010). However, Inadditiontometeorologicaldata,surfaceinformationinter- station-scaledatacanonlydescribetheclimateataspecific pretedfromsatelliteimages,e.g.,soilmoisture,landuseand pointinspace,andmostofthemlocatedatthefootofmoun- snowcover,canalsobeusedinhydrologicmodeling(Caiet tains. This limitation needs to be taken into consideration al.,2014). wheninterpolatingstationdataintobasin-scaleoverrugged As demonstrated in numerous research studies, data as- terrain.Lietal.(2014)appliedtheinterpolatedgriddedpre- similationholdsconsiderablepotentialforimprovinghydro- Hydrol.EarthSyst.Sci.,21,669–684,2017 www.hydrol-earth-syst-sci.net/21/669/2017/ Y.Chenetal.:HydrologicalmodelinginglacierizedcatchmentsofcentralAsia 677 logical predictions (Y. Liu et al., 2012). Cai et al. (2014) glacier module in Xinanjiang and TOPMODEL; and Fang used Global Land Data Assimilation System (GLDAS) 3h etal.(2015a)failedtoaccountforglacierprocesses,though airtemperaturedatatoforcetheMS-DTVGMmodel,while theglaciermeltcouldcontributeupto10%ofdischargeof Duethmannetal.(2015)usedtheWatchForcingDatabased the Kaidu River basin. Similarly, in their research on the onERA-40(WFD-E40)toforcethehydrologicalmodel. Yarkand River basin, Liu et al. (2016a) neglected the in- Remote sensing and reanalysis data are supposed for use fluence of glacier melt in the SWAT and MIKE-SHE mod- in large-scale hydrological modeling due to their low spa- els, even though the glacier covered an area of 5574km2. tial resolution. Another limitation in using remote sensing Themostwidelyusedhydrologicalmodels,suchasthedis- andreanalysisdataisthatthesedataarebiasedtosomeex- tributed SWAT, the MIKE-SHE model and the conceptual tent.Forexample,theTRMMdataaremostlyvaluableonly SRMmodel,asaruledonotcalculateglaciermeltprocesses, fortropicalregions,andreanalysisdata,includingERA-40, despite the fact that excluding the glacier processes could NCEP/NCARandGPCC(GlobalPrecipitationClimatology induce large errors in glacierized catchments. Glacier pro- Centre),failtorevealanysignificantcorrelationwithstation cesses are complex in that glacier melt will at first increase data(Sorgetal.,2012). duetotheriseinablationandloweringofglacierelevation, Given the advantages and disadvantages of observation and then, after reaching its peak, will decrease due to the data, remote sensing data and reanalysis data, a better ap- shrinkinginglacierarea(Xieetal.,2006).Moreover,simu- proachwouldbetocombineobservationsandotherdatasets lationerrorscanbere-categorizedasprecipitationorglacier inhydrologicalmodeling. meltwaterandconsequentlyresultinagreateruncertaintyin thewaterbalanceinhighmountainareas(Mayretal.,2013). 3.1.3 GCMorRCMoutputs Duringthelastfewdecades,alargevarietyofmeltmod- elshavebeendeveloped(Hock,2005).Previousstudieshave GCMs or RCMs provide climate variables for evaluating investigated glacier dynamics for the mountainous regions. future hydrological processes. However, the greatest chal- Among these studies, Hock (2005) reviewed glacier melt- lengesinapplyingthesedatasetsaretheirlowspatialresolu- related processes at the surface–atmosphere interface rang- tions(e.g.,thespatialresolutionofGCMsinCMIP5ranges ing from a simple temperature-index model to a sophisti- from0.75to3.25◦)andconsiderablebiases.Inaddition,dif- cated energy-balance model. Glacier models that are phys- ferentGCMsorRCMsgenerallygivedifferentclimatepro- ically based (e.g., mass-energy fluxes and glacier flow dy- jections. Therefore, when forcing a hydrological model us- namics) depend heavily on detailed knowledge of local to- ingtheoutputsofclimatemodels,theevaluationresultsde- pographyandhydrometeorologicaldata,whicharegenerally pendheavilyontheselectionofGCMsandconsequentlyre- limited in high mountain regions (Michlmayr et al., 2008). sultinhigheruncertaintyinGCMsthanthatinothersources Hence,theymostlyappliedtowell-documentedglaciersand (emissionscenarios,hydrologicalmodels,downscaling,etc.) havefewapplicationsinbasin-scalehydrologicalmodels. (Bosshardetal.,2013). The temperature-index method (or its variants), which Manydownscalingmethodshavebeendevelopedtoover- onlyrequirestemperatureformeteorologicalinput,iswidely comethesedrawbacks.Althoughsomestatisticaldownscal- usedtocalculateglaciermelt(KonzandSeibert,2010).Asis ingmethods,suchasSDSM(Wilbyetal.,2002),arewidely illustratedbyOerlemansandReichert(2000),glacierscanbe usedinclimatechangeimpactstudies,theiruseinthemoun- reconstructed from long-term meteorological records, e.g., tainous regions of central Asia is limited due to the lack of summer temperature is the dominant factor for glaciers in fineobservationaldatatodownscaleGCMoutputs.Toover- a dry climate (e.g., Abramov glacier). In recent years, hy- comethedatascarcityforthisregion,G.H.Fangetal.(2015) drologistshavebeentryingtoaddothermeteorologicalvari- evaluated different bias correction methods in downscaling ables into the calculations of glacier melt; e.g., Y. Zhang et the outputs of one RCM model and used the bias-corrected al. (2007) included potential clear sky direct solar radiation climate to force a hydrological model in the data-scarce inthedegree-daymodel,andYuetal.(2013)statedthatac- KaiduRiverbasin.Liuetal.(2011)usedperturbationfactors cumulated temperature is more effective than daily average to downscale the GCM outputs and force the hydrological temperatureforcalculatingthesnowmeltrunoffmodel.Us- model. ingdegree-daycalculationismuchsimplerthanusingenergy balance approaches and could actually produce comparable 3.2 Glaciermeltmodeling or better model performance when applied in mountainous basins(Ohmura,2001). Glaciermeltaccountsforalargepartofthedischargeforthe Morerecently,themeltmodulehasbeenincorporatedinto alpine basins in central Asia as discussed above. However, different kinds of hydrological models. Zhao et al. (2015) most hydrological modeling does not include glacier melt integrated a degree-day glacier melt algorithm into a and accumulation processes. For example, Liu et al. (2010) macroscale hydrologic model (VIC) and indicated that an- failed to account for the glacier processes in the VIC nualandsummerrunoffwoulddecreaseby9.3and10.4%, model in the Tarim River; Peng and Xu (2010) missed the respectively, for reductions in glacier areas of 13.2% in the www.hydrol-earth-syst-sci.net/21/669/2017/ Hydrol.EarthSyst.Sci.,21,669–684,2017 678 Y.Chenetal.:HydrologicalmodelinginglacierizedcatchmentsofcentralAsia Kumalike River basin. Hagg et al. (2013) analyzed antici- 3.3 Modelcalibrationandvalidation patedglacierandrunoffchangesintheRukhkcatchmentof theupperAmuDaryaBasin,usingtheHBV-ETHmodelby includingglaciermeltandsnowmeltprocesses.Theirresults showedthatwithtemperatureincreasesof2.2and3.1◦C,the currentglacierextentof431km2willreduceby36and45%, For model calibration, two important issues are discussed respectively.Luoetal.(2013),takingtheManasRiverbasin here:thelengthofthecalibrationperiodandobjectivefunc- asacasestudy,investigatedglaciermeltprocessesbyinclud- tions. ing the algorithm of glacier melt, sublimation/evaporation, Generally, hydrological modeling requires several years’ accumulation, mass balance and retreat in a SWAT model. calibration.Forexample,Yangetal.(2012)indicatedthata The results showed that glacier melt contributed 25% to 5-yearwarm-upissufficientbeforehydrologicalmodelcal- streamflow, although the glacier area makes up only 14% ibration and a 4-year calibration could obtain satisfactory ofthecatchmentdrainagearea. modelperformance.Moreventuresomely,a6-monthcalibra- tioncouldleadtogoodmodelperformanceforanaridwater- 3.2.1 Paucityofglaciervariationdata shed (Sun et al., 2016). Konz and Seibert (2010) stated that one year’s calibration of using glacier mass balances could The existing glacier data set, which includes the World effectivelyimprovethehydrologicalmodel.Selectingtheap- Glacier Inventory (WGI), the Randolph Glacier Inven- propriate calibration period is significant, as model perfor- tory (RGI) and global land-ice measurements from space mance could depend on calibration data. Refsgaard (1997) (GLIMS),hasbeendevelopedrapidly.Thesedata,however, usedasplit-sampleproceduretoobtainbettermodelcalibra- generally focus on glaciers in the present time or those ex- tionandvalidationeffectivelyandefficiently. isting in the former Soviet Union. For example, the source Moststudiesoncalibrationproceduresinhydrologyhave data of WGI were derived during 1940s–1960s, and the examined goodness-of-fit measures based on simulated and GLIMS for the Amu Darya Basin is from 1960 to 2004 observed runoff. However, as the hydrological sciences de- (Donald et al., 2015). These data can depict the charac- velop further, multi-objective calibration is emerging as the teristics of the glacier status, but fail to reproduce glacier preferredapproach.Itnotonlyincludesmulti-sitestreamflow variation. Only a few glaciers (Abramov, Tuyuksu, Urumqi (which has proved to be advantageous compared to single- no. 1 Glacier, etc.) have long-term variation measurements site calibration (S. Wang et al., 2012), and multi-metrics of (Savoskul and Smakhtin, 2013). CAWa (Central Asian Wa- streamflow(Yangetal.,2014),butalsoinvolvesmultipleex- ter; www.cawa-project.net/) are intended to contribute to a aminedhydrologicalcomponents(e.g.,soilmoisture).Most reliable regional data basis of central Asia from the moni- of the studies reviewed here use the discharge to calibrate toring stations, sampling and remote sensing. The missing andvalidatethehydrologicalmodel,yetGuptaetal.(1998) glaciervariationinformationleadstoamisrepresentationof arguedthatastrong“equifinalityeffect”mayexistduetothe glacierdynamics. compensation effect, where an underestimation of precipi- tation may be compensated by an overestimation of glacier 3.2.2 Lackofglaciermassbalancedata melt, and vice versa. Stahl et al. (2008) suggested that ob- servations on mass balances should be used for model cali- Glacier measurements reproduced by remote sensing data bration,aslargeuncertaintiesexistinthedata-scarcealpine usually give glacier area instead of glacier water equiva- regions. Therefore, multi-criteria calibration and validation lent; therefore, errors will occur when converting glacier isnecessary,especiallyforglacier/snowrechargedregions. area to glacier mass. Glaciologists normally use a speci- Many recent studies have attempted to include mass bal- fied relation (e.g., empirical) between glacier volume and ance data into model calibration (Stahl et al., 2008; Huss glacier area to estimate glacier mass balance (Stahl et al., et al., 2008; Konz and Seibert, 2010; Parajuli et al., 2009). 2008; Luo et al., 2013). Aizen et al. (2007) applied the Duethmannetal.(2015)usedamulti-objectiveoptimization radio-echo sounding approach to obtain glacier ice volume. algorithm that included objective functions of glacier mass Recently, ICESat (Ice, Cloud and land Elevation Satellite; balance and discharge to calibrate the hydrological model http://icesat.gsfc.nasa.gov/) could provide multi-year eleva- WASA(ModelofWaterAvailabilityinSemi-AridEnviron- tiondataneededtodetermineice-sheetmassbalance. ments).Anotherapproachforimprovingmodelefficiencyis This paper focuses primarily on glacier melt modules. It to calibrate the glacier melt processes and the precipitation does not discuss snowmelt processes, as hydrological mod- dominated processes separately (Immerzeel et al., 2012b). els generally include them either in a degree-day approach Further,inadditiontothemassbalancedatausedtocalibrate orenergybalancebasis.Furthermore,thispaperdoesnotan- thehydrologicalmodel,theglacier-area/glacier-volumescal- alyzewaterroutingprocessesorevapotranspirationbecause ing factor can also be calibrated with the observed glacier thereareseveralwaystosimulatesoilwaterstoragechange areachangemonitoredbyremotesensingdata(Zhangetal., andmodelevapotranspiration(Bierkens,2015). 2012b). Hydrol.EarthSyst.Sci.,21,669–684,2017 www.hydrol-earth-syst-sci.net/21/669/2017/
Description: