Hydrol Earth Syst Sci 15 2789ndash2803 2011wwwhydrol-earth-syst-scinet1527892011doi105194hess-15-2789-2011copy Author(s) 2011 CC Attribution 30 License

Hydrology andEarth System

Sciences

Climate change impacts on snow water availability in theEuphrates-Tigris basin

M Ozdogan

Center for Sustainability and the Global Environment (SAGE) amp Department of Forest and Wildlife Ecology University ofWisconsin-Madison 1710 University Avenue Madison WI 53726 USA

Received 27 March 2011 ndash Published in Hydrol Earth Syst Sci Discuss 15 April 2011Revised 9 August 2011 ndash Accepted 3 September 2011 ndash Published 8 September 2011

Abstract This study investigates the effects of projectedclimate change on snow water availability in the Euphrates-Tigris basin using the Variable Infiltration Capacity (VIC)macro scale hydrologic model and a set of regional climate-change outputs from 13 global circulation models (GCMs)forced with two greenhouse gas emission scenarios for twotime periods in the 21st century (2050 and 2090) The hy-drologic model produces a reasonable simulation of seasonaland spatial variation in snow cover and associated snow wa-ter equivalent (SWE) in the mountainous areas of the basinalthough its performance is poorer at marginal snow coversites While there is great variation across GCM outputsinfluencing snow water availability the majority of modelsand scenarios suggest a significant decline (between 10 and60 percent) in available snow water particularly under thehigh-impact A2 climate change scenario and later in the 21stcentury The changes in SWE are more stable when multi-model ensemble GCM outputs are used to minimize inter-model variability suggesting a consistent and significant de-crease in snow-covered areas and associated water availabil-ity in the headwaters of the Euphrates-Tigris basin Detailedanalysis of future climatic conditions point to the combinedeffects of reduced precipitation and increased temperaturesas primary drivers of reduced snowpack Results also indi-cate a more rapid decline in snow cover in the lower eleva-tion zones than the higher areas in a changing climate butthese findings also contain a larger uncertainty The simu-lated changes in snow water availability have important im-plications for the future of water resources and associatedhydropower generation and land-use management and plan-ning in a region already ripe for interstate water conflictWhile the changes in the frequency and intensity of snow-bearing circulation systems or the interannual variability re-

Correspondence toM Ozdogan(ozdoganwiscedu)

lated to climate were not considered the simulated changesin snow water availability presented here are likely to be in-dicative of climate change impacts on the water resources ofthe Euphrates-Tigris basin

1 Introduction

It is widely accepted that projected changes in our climate as-sociated with increasing concentrations of greenhouse gasesin the atmosphere will fundamentally alter the magnitudeand seasonal variations in temperature and precipitation pat-terns in many parts of the globe (IPCC 2007a) What is lessknown however is the impact this change will have on wa-ter resources and freshwater ecosystems especially in moun-tainous regions where much of the regional water supply isstored as snowpack and glaciers that melt into rivers (Ham-let and Lettenmaier 1999 Beniston et al 2003 Mote et al2005)

In mountainous ecosystems of the world seasonal snow-pack is a key component of the hydrologic cycle storingwater in winter and releasing it in spring and early summerwhen hydroelectric agricultural ecological and recreationaldemands for water are greatest In many mountainous riverbasins snow is the largest component of water storage butthe availability of snow also makes these basins most vulner-able to climatic variations and changes that influence win-ter snowpack For example decline in snow storage due toincreased warming could reduce snowmelt contributions torunoff and the timing of runoff events but not necessarilyoverall runoff volumes As a result too much snowmelt-generated runoff could occur in early spring when it is leastneeded but this will not help in summer when it is essen-tial for societal and ecological needs Less snow also meansreduced power output in downstream generators as well asnegative impacts to ecosystem functions including aquatichabitat fish migration and wetland replenishment

Published by Copernicus Publications on behalf of the European Geosciences Union

2790 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

The vulnerability from climatic variations and changesthat influence winter snowpack is perhaps best exemplifiedin the Euphrates-Tigris (E-T) river basin Aggressive wa-ter management programs by Turkey Syria and Iraq overthe course of the last three decades have resulted in dramaticchanges in the land-use patterns and the hydrology of thebasin Most of these changes take the form of capturingsnow-generated runoff in a series of dams in the upstreamcountries and releasing it for irrigation and other economicpurposes in the warm season As a result these countriesnow more heavily depend on the continuous existence of sea-sonal mountain snowpack more than anytime in the historyHowever if climate change alters snow-related processes inthe mountainous upstream countries the changes in season-ality and to a lesser extent amount of snow accumulationand melting would have important implications for down-stream uses such as storage requirements and dam operationAs a result climate change could have significant implica-tions for evaporative loss water demand and agriculturaloutput as well as long-term hydro-energy planning in thebasin (Gleick 2000) Similarly continued high populationgrowth in the riparian countries will require sustained (andperhaps increased) levels of snow-generated runoff to main-tain the cultural and economical benefits of water use in thebasin Hence climate change either on its own or when com-bined with other factors in the long term may provide condi-tions for conflict between the riparian countries within theE-T watershed

Several studies have investigated the effects of climatechange on winter snow dynamics and associated changes inrunoff (Milly et al 2005) For example Hamlet and Let-tenmaier (1999) used four global climate models (GCMs) toevaluate possible future changes to the Pacific Northwest cli-mate and the surface water response of the Columbia Riverbasin in the US Their results showed significant increases inwinter runoff volumes due to increased winter precipitationand warmer winter temperatures In particular the reducedsnowpack and earlier snow melt coupled with higher evapo-transpiration in early summer led to earlier spring peak flowsand reduced warm-season runoff volumes In the same loca-tion McCabe and Wolock (1999) used a conceptual snow ac-cumulation and melt model to estimate future snowpack byusing changes in monthly temperature and precipitation sim-ulated by two GCMs Their results indicated that althoughwinter precipitation is estimated to increase in the futureincreases in temperatures will result in large decreases inspring snowpack for the entire western US Christensen andLettenmaier (2007) investigated the implications of 21st cen-tury climate change on the hydrology and water resources ofthe Colorado River Basin using a multi-model ensemble ap-proach with 11 downscaled and bias-corrected global circu-lation model outputs and a macroscale hydrological modelTheir results indicated that modeled April first snow waterequivalent (SWE) declined for all ensemble members andtime periods In addition the maximum reduction (ensem-

ble mean) of 38 for the A2 aggressive climate change sce-nario occurred later in the century Associated with decreas-ing snow water availability and the shift in increased evap-otranspiration over the seasonal precipitation average totalbasin reservoir storage and average hydropower productiongenerally declined with a large range across the ensembleIn their more recent work McCabe and Wolock (2009) simu-lated mid-spring SWE between 1900 and 2008 in the WesternUS and concluded that periods of higher-than-average SWEare associated with higher precipitation and lower tempera-ture While this work focused on past climate conditionsit provides empirical evidence for climate-induced variationsin snow pack dynamics that could be affected under futureclimate change scenarios

A number of studies also investigated the impacts of cli-mate change in the Middle East and particularly in the E-T basin although none have concentrated specifically onthe effects to snow water availability Using a super-high-resolution GCM Kitoh et al (2008) concluded that by theend of this century the Fertile Crescent (the arc area that cov-ers the headwater region of the E-T basin) will lose its cur-rent shape and may disappear altogether due in part to sig-nificant decreases (29ndash73 ) in the annual discharge of theEuphrates River Evans (2009) examined the performanceand future predictions of 18 GCMs in the Middle East andfound an overall temperature increase of almost four degreesK and a large decrease in precipitation associated with a de-crease in storm track activity in parts of Turkey Syria andnorthern Iraq by late 21st century SimilarlyOnol and Se-mazzi (2009) assessed the role of global warming in mod-ulating the future climate over the eastern Mediterranean re-gion using a regional climate model under IntergovernmentalPanel on Climate Change (IPCC) emission scenarios Theirresults suggest a large decrease in precipitation over south-eastern Turkey where the headwaters of the E-T basin origi-nate With these results in mind the research presented hereinvestigates the effect of projected climatic changes on snowwater availability in the E-T basin using a distributed hydro-logical model capable of capturing snow-water dynamics inmountainous watersheds In particular it focuses on changesin SWE as an aggregate measure of climate change impactson snow water availability that forms the major source offlow for the twin rivers Unlike previous research this studyalso represents a comprehensive assessment that uses 52 cli-mate change scenario realizations (13 models two emissionscenarios and two time periods)

2 Methodology

21 Study area

The Euphrates and Tigris Rivers originate from the highlandsof eastern Turkey Iran and Syria and discharge south intoSyria and Iraq where they merge to feed the Mesopotamian

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2791

Fig 1 Location and physiography of the Euphrates-Tigris basinThe location of each river course is also shown The high elevationzones flanking the north and east of the basin forms the shape of theFertile Crescent

marshlands in southern Iraq (Fig 1) Nearly all of the flow(sim90 ) of the Euphrates river originates in the highlandsof eastern Turkey with modest contributions from the Syr-ian highlands but only minimal additions from Iraq (Gruen2000) The mountains in eastern Turkey contribute a smalleramount (sim40 ) to the Tigris river flow and the remaindercomes from numerous tributaries that originate in the ZagrosMountains between Iraq and Iran (Gruen 2000) A typicalcontinental climate prevails in the basin with a distinct northto south temperature and precipitation gradient Most of theprecipitation in the Turkish highlands in the north and in theZagros mountains in the east occurs between November andApril mostly in the form of snow over higher elevations Atthe height of the cold season (FebruaryndashMarch) snow coverflanks the northern and eastern edges of the basin in parallelto the shape of the Fertile Crescent (Fig 1) Precipitation re-mains in the snowpack until the spring melt season (MarchndashJune) Mountain snowpack plays a critical role in the hy-drological cycle of the basin with spring snow-melt beingthe dominant source for the flow of the two rivers Henceannual peak flows in both rivers exhibit a unimodal distri-bution in which the peak flow coincides with seasonal melt-ing of snow in May independent of in-season precipitation(Fig 2) Flow steadily decreases throughout the summer andearly fall reaching its lowest value in September-October be-fore the rainy period commences again in November Giventhe importance of water for energy production flood controlagricultural productivity with irrigation and reservoir opera-tions in the E-T basin information on water available in thesnowpack and the potential changes in the extent timing andmagnitude of snowpack under a changing climate are crucial

Fig 2 Distribution of long-term averaged Euphrates flow (black)and precipitation (gray) for a location in SE Turkey (centered at4016 E and 396 N) The flow data was obtained from the agencyresponsible for collecting and disseminating flow data in Turkey(EIEI 2008) The precipitation data was extracted from the Cli-matic Research Center (CRU) long-term precipitation climatologydataset (New et al 2002) Note the start month is October forthe water year The mismatch between precipitation and runoff inthe spring is an indication of snow melt-dominated flow in thosemonths

22 Hydrologic model and meteorological forcing data

The variable infiltration capacity model

The Variable Infiltration Capacity (VIC) model is amacroscale hydrology model that has been successfully ap-plied to many continental scale river basins in various cli-mates (Liang et al 1994 Nijssen et al 1997 Cherkauer andLettenmaier 2003) VIC solves energy and water balanceequations over each model grid cell at each time step whiledistinctly parameterizing subgrid variability in soil mois-ture precipitation topography and vegetation More specifi-cally each grid cell can have multiple soil layers and be par-tially covered by a mosaic of vegetation types and elevationbands Moisture and energy fluxes are computed separatelyfor each vegetation category and elevation band within eachgrid cell and then area-weighted and summed over the gridcell Streamflow is handled by a separate routing subsur-face and surface runoff scheme developed by Lohmann etal (1996)

In VIC cold season processes including snow accumula-tion and ablation are modeled using a two-layer energy andmass balance model where the thin surface layer is used tosolve the energy balance between the atmosphere and snow-pack while the lower layer is used as storage for the ex-cess snow mass (Storck et al 2002 Cherkauer and Letten-maier 2003) Specifically the SWE variable is modeled as

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2792 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

part of a detailed but straightforward snow module that simu-lates many of the snow accumulation and ablation processes(Storck et al 2002) The module is especially well suitedfor simulating mountain snowpack and includes the effectsof forest canopy on snow interception and the attenuationof wind and solar radiation which are important drivers ofsnow accumulation sublimation and melt processes (Storck2000 Storck et al 2002) Fundamentally the partitioningbetween the snow and rain phases of precipitation is deter-mined from a simple scheme based on estimated hourly tem-peratures that are derived from daily minimum and maxi-mum temperatures Belowminus05C precipitation is assumedto fall as 100 snow and above 05C as 100 rain A lin-ear relationship between snow and rain is assumed betweenthese two limits In addition to determining the phase of pre-cipitation temperature also influences other snow accumu-lation and melt processes including convective heat transferfrom the air to the snowpack (eg Storck 2000) attenua-tion of solar radiation and variations in long-wave radiation(determined from the difference betweenTmax and Tmin)and calculation of the dewpoint and vapor pressure deficit(from minimum temperature) that influence snow sublima-tion (Thornton and Running 1999) More specifically thetrends in temperature are indirectly related to trends in long-and shortwave radiation humidity and vapor pressure con-tributing to sublimation and snowmelt in the model Snowprocesses including SWE are modeled separately for a mo-saic of elevation bands and vegetation types in each grid cellNote that when solving for a subgrid tile of certain vegeta-tion type and elevation band VIC assumes that any snowfully covers that tile

For this study a grid layout at 18 intervals (roughly12 km on a side) were set up over the E-T basin For eachgrid cell the VIC model was run in energy balance modewhich solves the complete water balance but also minimizessurface energy balance errors The surface energy balanceis closed through an iterative process which attempts to finda surface temperature that adjusts surface energy fluxes sothat they balance the incoming solar and longwave radiationfluxes Although computationally more expensive this modesimulates the surface energy fluxes that are important to un-derstanding the snow accumulation and ablation processesAmong many hydrologic quantities that the VIC model isable to simulate only the daily SWE predictions for eachgrid cell both as individual elevation bands and as their arealaverages were used as the primary output in this study

To assess the effects of climate change on snow wateravailability the VIC model was forced with gridded fieldsof minimum and maximum temperature precipitation windspeed humidity and solar radiation at sub-daily time stepsTo establish the baseline snow dynamics in the basin min-imum and maximum temperature precipitation solar radia-tion wind speed and humidity on a 6 h time step were ob-tained from the National Centers for Environmental Predic-tion (NCEP) Global Reanalysis project data (Kalnay et al

1996) that were downscaled to 18 resolution using a bilin-ear interpolator To account for topographic effects on snowdynamics not captured at the coarse scale nature of the Re-analysis data correction factors were applied to the tempera-ture (lapsed by 61C kmminus1) and precipitation data followingthe Precipitation Regression on Independent Slopes Method(PRISM) algorithm (Daly et al 1994) Even with these cor-rections the derived forcing data did not exhibit large spa-tial variation within the mountainous portions of the basin(its effects are discussed in later sections) Nevertheless thelimitations in the forcing data may not have significant im-pacts because the snow accumulation and ablation processesin mountainous regions are controlled by the accumulatedprecipitation in snowpacks and thus a lack of spatial vari-ability in the input data may not be important (Hamlet etal 2005) Further evidence that lack of spatial variabilitymay not contribute to large model errors is shown by theVIC modelrsquos ability to capture the temporal and spatial pat-terns of snow covered areas and SWE (Fig 4 through 6)Moreover a comparison of the Reanalysis-derived climaticvariables to those obtained from the National Climatic DataCenter (NCDC) global summary of day reports for five me-teorological stations located in the study area (NCDC 2010)revealed a strong correlation for all forcing variables exceptprecipitation (Table 1) The precipitation correlation was lessthan ideal but may be attributable to an average 100 mm pos-itive bias in the NCEP Reanalysis data

In Table 1 the quantitative comparison between NCEPand station data was achieved using the correlation coeffi-cient(r) mean absolute error (MAE) root mean square er-ror (RMSE) and the index of agreement (d) following Will-mott (1981) The index of agreement is a measure of rela-tive error in model estimates It is a dimensionless numberbetween 00 and 10 where 00 describes complete disagree-ment between estimated and observed values and 10 indi-cates that the simulated and observed values are identicalThe index of agreement is included because the correlationcoefficient(r) cannot account for additive differences or dif-ferences in proportionality (Willmott 1981) Neverthelessr (measure of covariability between observed and modeledvalues) and RMSE (measure of average difference betweenobserved and modeled values) statistics are also included be-cause they describe different components of model error in-dicated byd

In its default configuration the VIC model treats eachgrid cell to have a single elevation value (ie grid cells areflat) In mountainous areas where snow processes are animportant part of the hydrological budget the single eleva-tion assumption can lead to inaccuracies in snow pack es-timates To overcome this issue each grid cell in the E-T basin was broken up into a total of six elevation bands(also called snow bands) and simulated individually Alongwith the PRISM correction described above VIC was usedto lapse the grid cell average temperature pressure and pre-cipitation to a more accurate local estimate based on each

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2793

Table 1 Statistics describing the strength of the relationship between three meteorological forcing variables measured at a single site (locatedat 3796 N and 3963 E) and extracted from the NCEP dataset for the same location The metrics used in comparison and their descriptionand formulas are also provided

var N mod obs Smod Sobs MAE RMSE RMSES RMSEU d

Tmax 13767 211 225 1302 118 296 39 119 1302 097Tmin 13762 64 87 85 895 303 38 925 85 095PREC 13672 16 13 425 409 111 321 41 425 083

Statistic Description

N number of samplesmod mean modeled quantity (mm dminus1 for precipitation andC for temperature)obs mean observed quantity (mm dminus1 for precipitation andC for temperature)Smod standard deviation of modeled quantity (mm dminus1 for precipitation andC for temperature)Sobs standard deviation of observed quantity (mm dminus1 for precipitation andC for temperature)

MAE mean absolute error1N

nsumi=1

| modi minusobsi |

RMSE root mean squared error[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSES systematic RMSE[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSEU unsystematic RMSE[Nminus1nsum

i=1( modi minus modi)

2]05

d index of agreementd = 10minus

nsum

i=1( mod iminusobsi )2

nsumi=1

(∣∣ mod iminusobs

∣∣+∣∣obsiminusobs∣∣)2

bandrsquos mean elevation value As a result VIC has the abilityto predict snow-related variables by elevation band althoughdistribution within each band in non-spatial ndash ie the loca-tion of snow accumulationablation in the elevation band isnot known Aside from the simple climatic downscaling ofthe input climatic drivers and this correction the VIC modeldoes not introduce additional variability

While this study primarily makes use of the detailed en-ergy balance snow model described above (Cherkauer andLettenmaier 2003) the VIC model has been well validatedwith streamflow observations particularly in the mountain-ous western US (Christensen et al 2004 Hamlet et al2005 Maurer et al 2002) The model has also been usedextensively for streamflow forecasting applications (Hamletand Lettenmaier 1999 Wood et al 2002) and for climatechange assessments (Lettenmaier et al 1999 Christensen etal 2004 Hamlet and Lettenmaier 1999 Payne et al 2004Snover et al 2003)

23 Climate change scenarios

The climatic conditions used to represent a changing cli-mate were constructed from the output of 13 Global Cir-culation Models (GCMs) obtained from the Coupled ModelIntercomparison Project phase 3 (CMIP3) of the World Cli-

Table 2 List of GCMs and relevant references used in this study

Model Origin Reference

BCCR-BCM20 Norway Deque et al (1994)CCMA CGCM31 (T63) Canada Scinocca et al (2008)CNRM CM3 France Salas-Melia et al (2005)CSIROMk35 Australia Gordon et al (2002)GFDL CM21 USA Delworth et al (2006)GISSER USA Russell et al (2000)INM CM30 Russia Diansky and Volodin (2002)IPSL CM4 France IPSL (2005)MIROC32(med res) Japan K-1 model developers (2004)MPI ECHAM5 Germany Jungclaus et al (2006)MRI CDCM232 Japan Yukimoto et al (2001)NCAR CCSM3 USA Collins et al (2006)UKMO HadCM3 UK Gordon et al (2000)

mate Research Program (WCRP) (Meehl et al 2007) (Ta-ble 2) Variables for the VIC model were extracted fromeach model by selecting the grid boxes most closely alignedwith the study area Note that the selected grid boxes oc-cupy slightly different areas and locations between modelsdue to different projection characteristics of each modelrsquosgrid layout Monthly data from the control and the perturbed

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2794 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

(ie climate change) integrations of each model run wereused to calculate the anomalies that were then applied tothe observed subdaily baseline The anomalies includedmonthly changes in precipitation intensity (in relative units)changes in the duration of wet and dry spells changes inmonthly temperature means (in absolute terms) and changesin monthly mean radiation (in absolute terms) Using thesevarious datasets the observed (ie baseline) data were ad-justed to reflect each modelrsquos climate change predictions atmonthly integrations assuming stationarity in sub-monthlyvariability Using monthly deviations from baseline obser-vations five years of synthetic sub-daily weather data weregenerated under a series of future climate scenarios Forthe climate change impact assessment only two time peri-ods were considered 2045ndash2055 (2050s) and 2085ndash2095(2090s) while the period between 1961ndash1990 represented thebaseline conditions For emission scenarios two storylineswere selected from the Special Report on Emission Scenarios(SRES) (IPCC 2000) (1) the low-impact (B1) and (2) high-impact (A2) development scenario Each storyline describesa different world evolving through the 21st century and leadsto different greenhouse gas emission trajectories driven bydemographic economic technological and land-use forcesThe B1 scenario describes a convergent world (where devel-oping and developed nations converge) with rapid changesin economic structures reductions in the intensity resourceuse and the introduction of clean technologies The empha-sis is on global solutions to bring about economic social andenvironmental sustainability including improving equity butwithout additional climate change policies The A2 storylinein contrast describes a differentiated world Economic de-velopment is primarily regional and technological changesare more fragmented in a world of self-reliance and continu-ously increasing population

231 Model calibration and validation

The VIC model was validated for snow covered areas in thenorthern mountainous part of the E-T basin using multi-yearsatellite observations The VIC model was run over the en-tire basin at 18 spatial resolution and the results were com-pared to MODerate Resolution Imaging Spectroradiometer(MODIS) Terra Snow Cover 8-Day composited L3 Global005Deg CMG (MOD10C2) data (Hall et al 2002) for theentire basin This comparison was made for four time peri-ods in January 2001

To validate the SWE output from the VIC model twodifferent observed datasets were used The first datasetcontained SWE observations obtained from the Directorateof Turkish Electrical Works snow observations book (EIEI2007) These observations are obtained roughly one timeper month in mountainous areas of the country and includesnow cover depth density and SWE For the purposes ofthis study data acquired at the station 21K07 (located at3893 N and 4024 E) were used The statistics describing

the strength of the relationship between observed and mod-eled values are given in Table 3

The second comparison involved the use of remotelysensed observations More specifically VIC output from anadditional single run at a site located in the headwaters of theEuphrates river in southeastern Turkey (centered at 405 Elongitude 385 N latitude) were compared to SWE datafrom the 8-day blended Special Sensor Microwave Imager(SSMI) and the MODIS snow cover remote sensing product(Brodzik et al 2007) The SWE dataset was derived fromthe SSMI Pathfinder daily brightness temperatures Thisdataset was enhanced by the MODIS snow cover area prod-uct derived from the MODISTerra Snow Cover 8-Day L3Global 005Deg CMG (MOD10C2) data regridded to theEASE-Grid While spatially coarse these remote sensing-based SWE datasets provide a first-order approximation ofseasonal and interannual changes in SWE

The final form of model evaluation was made by com-paring modeled runoff generated by VICrsquos routing modelagainst the observed runoff data obtained from the Direc-torate of Electrical Works monthly runoff book series (EIEI2008) We used data obtained from the Logmar station (lo-cated at 3852 N and 3949 E) at a monthly time step be-tween 1994 and 2004 The quality of fit between mod-eled and observed discharge was obtained by Nash-Sutcliffemodel efficiency coefficient commonly used to assess thepredictive power of models (Nash and Sutcliffe 1970)

3 Results

31 Expected changes in temperature and precipitation

The deviations from baseline climatic conditions of the 20thcentury (1961ndash1990) predicted by 13 climate models undertwo climate change scenarios (A2 B1) for two time peri-ods are provided as a temperature-precipitation cross-plot inFig 3 The temperature change was computed as the abso-lute difference (inC) between the baseline conditions andthe climate change scenarios thus positive values indicateincreases in air temperature The precipitation change wascomputed as the relative change from baseline conditionsnegative numbers indicate a reduction in precipitation fromthe baseline

In the future climate the mean cold season air tem-peratures show a consistent pattern of increase under allmodelscenarioyear combinations The amount of increaseranges from 07C under the SRES B1 scenario in 2050 togreater than 5C under the SRES A2 scenario at the end ofthe 21st century All models also suggest consistent increasesfrom 2050 to 2090 regardless of the emission scenario

With respect to precipitation most models suggest vari-ous levels of reduction ndash from 5 to 45 percent ndash under allscenarios and all time periods A few models suggest slightincreases in precipitation (up to 20 percent by one model)

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2795

Table 3 Various statistical metrics describing the strength of the relationship between observed and modeled SWE at one location (centeredat 3892 N and 4025 E) The description and the formulas used to derive the metrics are also provided

Statistic Value Description

N 138 number of samplesmod 932 mean modeled SWE (mm monthminus1)obs 1134 mean observed SWE (mm monthminus1)Smod 992 standard deviation of modeled SWE (mm monthminus1)Sobs 1048 standard deviation of observed SWE (mm monthminus1)

MAE 469 mean absolute error1N

nsumi=1

| modi minusobsi |

RMSE 737 root mean squared error[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSES 1064 systematic RMSE[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSEU 989 unsystematic RMSE[Nminus1nsum

i=1( modi minus modi)

2]05

d 086 index of agreementd = 10minus

nsum

i=1( mod iminusobsi )2

nsumi=1

(∣∣ mod iminusobs

∣∣+∣∣obsiminusobs∣∣)2

especially under the B1 scenario in the mid 21st centuryWith the exception of one model (NCAR) no precipitationincrease is suggested under the A2 scenario in either the mid-dle or at the end of the 21st century In general precipitationchange is greater under A2 as much as 25 percent lower thanaverage baseline conditions within each modelrsquos outcomeIn contrast to other models the NCAR model predicts no netchange nor cooler temperatures (by as much as 05C) un-der the B1 scenario but instead predicts a shift to warmingunder the A2 scenario These results suggest that in generalthe climate in the E-T basin will be warmer and dryer withsubsequent implications for the snow processes and availablemelt water

32 Performance of the VIC model

The first set of results illustrates the ability of the VIC modelto accurately simulate SWE in the E-T basin The compari-son between modeled and observed SWE at a single site usedto acquire ground SWE observations shows a strong correla-tion over time (Fig 4) Detailed statistics for this compari-son are given in Table 3 Please see the discussion providedabove for justification of using certain statistical measure-ment variables as suggested by Willmott (1981) When ob-served and modeled SWE data are compared in the form ofa scatter plot (not shown) there is no systematic bias at thesite level as shown in Table 3

A comparison of modeled SWE data to remotely sensedobservations also suggest that the VIC model is able to cor-rectly simulate the available water in accumulated snow Fig-ure 5 displays the comparison between SSMI-observed andVIC-modeled SWE (in mm) in the form of a time-series plot

Fig 3 Cold season (DJFMA) averaged changes in near surfaceair temperature and precipitation from baseline conditions (1961ndash1990) predicted by 13 models for four scenariotime combina-tions Different colors correspond to different models while markershapes indicates the scenariotime pair Positive values in the tem-perature field indicate increases in temperature while negative val-ues in the precipitation field indicate relative decreases in precipita-tion amounts as a result of climate change

between 2001 and 2006 averaged over 8-day intervals Ingeneral there is very good agreement between the two es-timates (r = 083) With a few exceptions the VIC modelaccurately simulates the seasonal and interannual variationof SWE in this part of the basin There is remarkable

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2796 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 4 Comparison of single day SWE observations to modeledquantities at a location in northern part of the E-T basin The siteis located at (3892 N and 4025 E) and includes once-per-monthobservations Sample dates of observed data are provided for refer-ence

Fig 5 A comparison of modeled (red) and observed (black) SWEat a location in southeastern Turkey (405 E longitude 385 N lat-itude) within the Euphrates-Tigris basin for 2001ndash2006 The ob-served data are extracted from the 8-day blended Special SensorMicrowave Imager (SSMI) data (Brodzik et al 2007)

correspondence between the VIC model and the SSMI ob-servations for the timing of accumulation maximum and ab-lation of snow The figure also points to interesting dynam-ics of SWE in the basin First snow accumulation begins inNovember of each year gradually increases to a maximumin late Marchndashearly April with a subsequent sharp declineinto the summer season Another important observation from

45

Figure 6 Comparison of observed (top) and simulated (bottom) snow covered areas in the Euphrates-Tigris basin across

different periods in winter 2001 Snow cover is indicated in white against the color-coded topography of the basin

Observations in the top row were extracted from Terra MODIS 8-day snow cover product (MOD10C2) at 005 degree

spatial resolution while the simulated results of the bottom row were produced in this research using the VIC model

More details on the MODIS snow cover product can be found in Hall et al (2002) Each date at the top indicate the

beginning date of the 8-day compositing period

January252001January172001January92001 February22001

MODIS

VIC

Fig 6 Comparison of observed (top) and simulated (bottom) snowcovered areas in the Euphrates-Tigris basin across different periodsin winter 2001 Snow cover is indicated in white against the color-coded topography of the basin Observations in the top row were ex-tracted from Terra MODIS 8-day snow cover product (MOD10C2)at 005 spatial resolution while the simulated results of the bot-tom row were produced in this research using the VIC model Moredetails on the MODIS snow cover product can be found in Hall etal (2002) Each date at the top indicate the beginning date of the8-day compositing period

Fig 4 is the interannual variability of SWE Both the mod-eled and remote sensing results had larger maximum SWEvalues (around 200 mm) in years 2001 2002 and 2006 hadthan 2004 and 2005 Thus while 2002 and 2003 are largesnow accumulation years (and therefore larger SWE) 2004and 2005 appear to be drier years with much less accumula-tion It is interesting to note that both the VIC model and thesatellite observations capture this variability quite well

Further assessment of the VIC model performance wasmade by visually comparing modeled snow covered areas tosatellite observations of snow cover from the MODIS sen-sor (Fig 6) The map comparison between satellite-observedand modeled snow area data for four dates in January 2001indicate a noticeable agreement Even though VIC predicteda larger area covered by snow on each date especially in theeastern mountainous zone of the basin the general patternof snow cover accumulating in the arc-shaped northern andeastern portions of the basin is remarkably similar betweenthe two data sets One reason for the larger area snow accu-mulation in VIC simulations is the coarse nature of the grid-ded weather data that smooth over the fine scale topographicvariation present in the MODIS data

Comparison of modeled discharge to observed quantitiesat a location in the northeastern portion of the basin furthershows a notable correlation especially in the timing of peakflows that are indicative of timing of snowmelt (Fig 7) Themodeled discharge data were obtained by routing VIC sur-face and baseflow predictions through the routing model ofLohmann et al (1996) The quality of the match between theobserved and modeled discharge determined by the Nash-Sutcliffe coefficient of 05 is moderate although the timingof peak flows is matched nicely indicating that VIC is ableto simulate the accumulation and melting of snow in accept-able form

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2797

Fig 7 Comparison of monthly observed and modeled discharge atthe Logmar station (located at 3851 N and 3949 E) While theabsolute value flow is less than ideal in some cases VIC modelcaptures the timing of peak flow quite well Nash-Sutcliffe (NS)statistic is also provided to show the strength of correlation

33 Changes in SWE for each GCM model outputs

The changes in SWE across 12 of the 13 models are givenin Fig 8 where in each panel deviations from the baselineconditions are depicted as a relative change in months defin-ing the cold season Note that even though the month ofApril is not generally considered part of the cold season inmany northern hemisphere studies it was included in thecurrent study due to its importance for snow ablation andmelting Also shown with text in the figure are the largestabsolute changes in SWE (in mm) for the month in questionThese absolute values reflect the averaged largest single daychanges in SWE in each month The direction of and mag-nitude of these absolute changes must be interpreted in thecontext of relative changes shown in the y-axis

Overall the following patterns emerge from Fig 8 Firstalmost all models predict a decline in SWE in the DecemberndashJanuary period However there are fewer models predictinga decline in SWE in the MarchndashApril period indicating thatthe reliability in predicting a decline in SWE is reduced whenmoving from the DecemberndashJanuary period to the MarchndashApril period Second there is considerable variability acrossthe models For example the GFDL model predicts a strongnegative change in SWE across all months years and emis-sion scenarios while the CNNM model predicts a weak neg-ative response in SWE under a changing climate (less than40 percent) In some cases the models even predict a positiveresponse Across all models the predicted increase althoughrare occurs in the spring during the normal thawing periodin 10 out of 13 models This represents a shift in the timingof snow accumulation moving from the DecemberndashFebruaryperiod to the MarchndashApril period Another interesting obser-

vation concerning the increase in SWE with climate changeis that it occurs more often under the B1 than the A2 sce-nario and more frequently in the middle of the 21st century(2050s) than in the later period (2090s)

There are notable exceptions to these general trends Forinstance the NCAR model does not fit this description re-sults from this model suggest a steady increase in SWE avail-ability under the aggressive A2 scenario at the end of the 21stcentury (although this increase occurs more in April than inJanuary) Several other models including the IPSL and theMIROC models also show similar trends but to a lesser ex-tent However all of these exceptions tend to occur in thespring rather than during the cold season under changing cli-matic conditions

Absolute changes in SWE as depicted by the number be-low bars representing each month also tell a similar story butin different context First the absolute magnitude of SWEchanges is highly variable ranging from less than 20 to over230 mm When placed in context of available snow waterat the peak snow accumulation period (as much as 400 mm)these absolute changes reflect 10 to 60 percent of the avail-able snow water Second the largest changes (and the mostvariation across models) occur in April followed by MarchThe timing of these largest changes in the spring are in con-trast with the large relative changes that occur in the winterperiod Although these largest changes do not necessarilyreflect the largest relative changes they indicate the impor-tance of climate change for accumulation (winter) and decay(spring) of SWE

Given the model-derived variability in predicting climatevariables that are important for snow accumulation the ques-tion arises as to which models or strategies should be used todistill information of use to stake-holders One common ap-proach is simply to average all models This approach com-monly referred to as a multi-model ensemble dilutes modelsthat provide poor outcomes with those that do a good job insimulating a region (Lambert and Boer 2001 Pierce et al2009) Using this approach a new multi-model ensembleclimate forcing dataset was generated for input to the VICmodel By averaging the outcomes of all climate models theensemble results suggest that SWE could decline as muchas 40 percent under all emission scenarios all time periodsand all months considered in this study (Fig 9) Of greaterinterest however is the gradual decline in the reduction inSWE from December (largest decrease) to April (smallestdecrease) These findings suggest that the amount of snowthat help build SWE during the cold season could graduallylessen leaving little or no snowmelt generated flow in thespring As with individual models the absolute changes in-crease from winter to spring

Rather than average the negative or positive changes as inthe ensemble approach it is possible to explore the frequencywith which the models predict positive or negative changesby month In Fig 10 the number of times an increase or de-crease was predicted by a model is plotted where the length

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2798 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 8 Changes in basin-averaged SWE relative to baseline conditions (1961ndash1990) predicted by 12 GCMs for the cold season (DJFMA)Negative values indicate a reduction and positive values indicate an increase in SWE relative to baseline conditions Also shown are themaximum averaged absolute changes in SWE (in units of mmmonth) The sign of the absolute change must be interpreted with the directionof relative change on the y-axis Note that only 12 out of 13 models used in the study are shown

of each bar is equal to 13 representing the total number ofmodels used The first finding from this plot is that all modelsindicate a clear decline in December under changing climateconditions Moreover many of the models predict a declinein SWE January through March Yet only half the modelspredict an increase while the other half predict a decrease inApril and this result is consistent across all emission scenar-ios and time periods This lack of consensus among modelsin April is mostly due to lack of snow in this period In otherwords the amount of reduction on an already small snowquantity in the spring is lost between model variability In

the winter period however all models do predict snow avail-ability but simply less of it under a changing climate

Finally Fig 11 shows the changes in modeled SWE acrosselevation zones (or bands) for four scenariotime combina-tions The elevation bands are given at 250 m incrementson the y-axis while the x-axis depicts the departure of SWEfrom the baseline conditions in relative terms The results re-veal that the largest changes (greater than 50 percent) occurin the lower elevation bands (under 500 m) and as altitudeincreases the reduction in SWE diminishes exponentially toa minimum value of about 10 percent Above 2000 m the

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2799

Fig 8 Continued

predicted decline in SWE is constant with a mean value of10 percent However slight positive bias in VIC-simulatedsnow area and SWE amounts in lower elevations suggeststhat the magnitude of change in lower elevations must beviewed with caution This is also corroborated by lack ofmodel consensus in April which affect lower elevations firstThese findings suggest that as the climate warms and pre-cipitation declines lower elevation zones could experience amore rapid reduction in snow accumulation than the higherelevation zones of the basin but these results are as good asthe quality of the model predictions in the lower areas

4 Discussion

In this research the effects of climate change on theamount of water stored in snowpack in the mountains ofthe Euphrates-Tigris river basin were assessed using the VICmacroscale hydrological model linked to climate model re-sults Inevitably the approach adopted in this study has sev-eral limitations With respect to the climate change predic-tions this study only considered the averages of the regionalclimate system but did not assess changes in its variabil-ity While the approach taken here involving the perturba-tion of observed climate allows for some random variationin day-to-day weather events the true variability includingclimatic extremes is not captured Work on the sensitivity of

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2791

Fig 1 Location and physiography of the Euphrates-Tigris basinThe location of each river course is also shown The high elevationzones flanking the north and east of the basin forms the shape of theFertile Crescent

marshlands in southern Iraq (Fig 1) Nearly all of the flow(sim90 ) of the Euphrates river originates in the highlandsof eastern Turkey with modest contributions from the Syr-ian highlands but only minimal additions from Iraq (Gruen2000) The mountains in eastern Turkey contribute a smalleramount (sim40 ) to the Tigris river flow and the remaindercomes from numerous tributaries that originate in the ZagrosMountains between Iraq and Iran (Gruen 2000) A typicalcontinental climate prevails in the basin with a distinct northto south temperature and precipitation gradient Most of theprecipitation in the Turkish highlands in the north and in theZagros mountains in the east occurs between November andApril mostly in the form of snow over higher elevations Atthe height of the cold season (FebruaryndashMarch) snow coverflanks the northern and eastern edges of the basin in parallelto the shape of the Fertile Crescent (Fig 1) Precipitation re-mains in the snowpack until the spring melt season (MarchndashJune) Mountain snowpack plays a critical role in the hy-drological cycle of the basin with spring snow-melt beingthe dominant source for the flow of the two rivers Henceannual peak flows in both rivers exhibit a unimodal distri-bution in which the peak flow coincides with seasonal melt-ing of snow in May independent of in-season precipitation(Fig 2) Flow steadily decreases throughout the summer andearly fall reaching its lowest value in September-October be-fore the rainy period commences again in November Giventhe importance of water for energy production flood controlagricultural productivity with irrigation and reservoir opera-tions in the E-T basin information on water available in thesnowpack and the potential changes in the extent timing andmagnitude of snowpack under a changing climate are crucial

Fig 2 Distribution of long-term averaged Euphrates flow (black)and precipitation (gray) for a location in SE Turkey (centered at4016 E and 396 N) The flow data was obtained from the agencyresponsible for collecting and disseminating flow data in Turkey(EIEI 2008) The precipitation data was extracted from the Cli-matic Research Center (CRU) long-term precipitation climatologydataset (New et al 2002) Note the start month is October forthe water year The mismatch between precipitation and runoff inthe spring is an indication of snow melt-dominated flow in thosemonths

22 Hydrologic model and meteorological forcing data

The variable infiltration capacity model

The Variable Infiltration Capacity (VIC) model is amacroscale hydrology model that has been successfully ap-plied to many continental scale river basins in various cli-mates (Liang et al 1994 Nijssen et al 1997 Cherkauer andLettenmaier 2003) VIC solves energy and water balanceequations over each model grid cell at each time step whiledistinctly parameterizing subgrid variability in soil mois-ture precipitation topography and vegetation More specifi-cally each grid cell can have multiple soil layers and be par-tially covered by a mosaic of vegetation types and elevationbands Moisture and energy fluxes are computed separatelyfor each vegetation category and elevation band within eachgrid cell and then area-weighted and summed over the gridcell Streamflow is handled by a separate routing subsur-face and surface runoff scheme developed by Lohmann etal (1996)

In VIC cold season processes including snow accumula-tion and ablation are modeled using a two-layer energy andmass balance model where the thin surface layer is used tosolve the energy balance between the atmosphere and snow-pack while the lower layer is used as storage for the ex-cess snow mass (Storck et al 2002 Cherkauer and Letten-maier 2003) Specifically the SWE variable is modeled as

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2792 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

part of a detailed but straightforward snow module that simu-lates many of the snow accumulation and ablation processes(Storck et al 2002) The module is especially well suitedfor simulating mountain snowpack and includes the effectsof forest canopy on snow interception and the attenuationof wind and solar radiation which are important drivers ofsnow accumulation sublimation and melt processes (Storck2000 Storck et al 2002) Fundamentally the partitioningbetween the snow and rain phases of precipitation is deter-mined from a simple scheme based on estimated hourly tem-peratures that are derived from daily minimum and maxi-mum temperatures Belowminus05C precipitation is assumedto fall as 100 snow and above 05C as 100 rain A lin-ear relationship between snow and rain is assumed betweenthese two limits In addition to determining the phase of pre-cipitation temperature also influences other snow accumu-lation and melt processes including convective heat transferfrom the air to the snowpack (eg Storck 2000) attenua-tion of solar radiation and variations in long-wave radiation(determined from the difference betweenTmax and Tmin)and calculation of the dewpoint and vapor pressure deficit(from minimum temperature) that influence snow sublima-tion (Thornton and Running 1999) More specifically thetrends in temperature are indirectly related to trends in long-and shortwave radiation humidity and vapor pressure con-tributing to sublimation and snowmelt in the model Snowprocesses including SWE are modeled separately for a mo-saic of elevation bands and vegetation types in each grid cellNote that when solving for a subgrid tile of certain vegeta-tion type and elevation band VIC assumes that any snowfully covers that tile

For this study a grid layout at 18 intervals (roughly12 km on a side) were set up over the E-T basin For eachgrid cell the VIC model was run in energy balance modewhich solves the complete water balance but also minimizessurface energy balance errors The surface energy balanceis closed through an iterative process which attempts to finda surface temperature that adjusts surface energy fluxes sothat they balance the incoming solar and longwave radiationfluxes Although computationally more expensive this modesimulates the surface energy fluxes that are important to un-derstanding the snow accumulation and ablation processesAmong many hydrologic quantities that the VIC model isable to simulate only the daily SWE predictions for eachgrid cell both as individual elevation bands and as their arealaverages were used as the primary output in this study

To assess the effects of climate change on snow wateravailability the VIC model was forced with gridded fieldsof minimum and maximum temperature precipitation windspeed humidity and solar radiation at sub-daily time stepsTo establish the baseline snow dynamics in the basin min-imum and maximum temperature precipitation solar radia-tion wind speed and humidity on a 6 h time step were ob-tained from the National Centers for Environmental Predic-tion (NCEP) Global Reanalysis project data (Kalnay et al

1996) that were downscaled to 18 resolution using a bilin-ear interpolator To account for topographic effects on snowdynamics not captured at the coarse scale nature of the Re-analysis data correction factors were applied to the tempera-ture (lapsed by 61C kmminus1) and precipitation data followingthe Precipitation Regression on Independent Slopes Method(PRISM) algorithm (Daly et al 1994) Even with these cor-rections the derived forcing data did not exhibit large spa-tial variation within the mountainous portions of the basin(its effects are discussed in later sections) Nevertheless thelimitations in the forcing data may not have significant im-pacts because the snow accumulation and ablation processesin mountainous regions are controlled by the accumulatedprecipitation in snowpacks and thus a lack of spatial vari-ability in the input data may not be important (Hamlet etal 2005) Further evidence that lack of spatial variabilitymay not contribute to large model errors is shown by theVIC modelrsquos ability to capture the temporal and spatial pat-terns of snow covered areas and SWE (Fig 4 through 6)Moreover a comparison of the Reanalysis-derived climaticvariables to those obtained from the National Climatic DataCenter (NCDC) global summary of day reports for five me-teorological stations located in the study area (NCDC 2010)revealed a strong correlation for all forcing variables exceptprecipitation (Table 1) The precipitation correlation was lessthan ideal but may be attributable to an average 100 mm pos-itive bias in the NCEP Reanalysis data

In Table 1 the quantitative comparison between NCEPand station data was achieved using the correlation coeffi-cient(r) mean absolute error (MAE) root mean square er-ror (RMSE) and the index of agreement (d) following Will-mott (1981) The index of agreement is a measure of rela-tive error in model estimates It is a dimensionless numberbetween 00 and 10 where 00 describes complete disagree-ment between estimated and observed values and 10 indi-cates that the simulated and observed values are identicalThe index of agreement is included because the correlationcoefficient(r) cannot account for additive differences or dif-ferences in proportionality (Willmott 1981) Neverthelessr (measure of covariability between observed and modeledvalues) and RMSE (measure of average difference betweenobserved and modeled values) statistics are also included be-cause they describe different components of model error in-dicated byd

In its default configuration the VIC model treats eachgrid cell to have a single elevation value (ie grid cells areflat) In mountainous areas where snow processes are animportant part of the hydrological budget the single eleva-tion assumption can lead to inaccuracies in snow pack es-timates To overcome this issue each grid cell in the E-T basin was broken up into a total of six elevation bands(also called snow bands) and simulated individually Alongwith the PRISM correction described above VIC was usedto lapse the grid cell average temperature pressure and pre-cipitation to a more accurate local estimate based on each

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2793

Table 1 Statistics describing the strength of the relationship between three meteorological forcing variables measured at a single site (locatedat 3796 N and 3963 E) and extracted from the NCEP dataset for the same location The metrics used in comparison and their descriptionand formulas are also provided

var N mod obs Smod Sobs MAE RMSE RMSES RMSEU d

Tmax 13767 211 225 1302 118 296 39 119 1302 097Tmin 13762 64 87 85 895 303 38 925 85 095PREC 13672 16 13 425 409 111 321 41 425 083

Statistic Description

N number of samplesmod mean modeled quantity (mm dminus1 for precipitation andC for temperature)obs mean observed quantity (mm dminus1 for precipitation andC for temperature)Smod standard deviation of modeled quantity (mm dminus1 for precipitation andC for temperature)Sobs standard deviation of observed quantity (mm dminus1 for precipitation andC for temperature)

MAE mean absolute error1N

nsumi=1

| modi minusobsi |

RMSE root mean squared error[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSES systematic RMSE[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSEU unsystematic RMSE[Nminus1nsum

i=1( modi minus modi)

2]05

d index of agreementd = 10minus

nsum

i=1( mod iminusobsi )2

nsumi=1

(∣∣ mod iminusobs

∣∣+∣∣obsiminusobs∣∣)2

bandrsquos mean elevation value As a result VIC has the abilityto predict snow-related variables by elevation band althoughdistribution within each band in non-spatial ndash ie the loca-tion of snow accumulationablation in the elevation band isnot known Aside from the simple climatic downscaling ofthe input climatic drivers and this correction the VIC modeldoes not introduce additional variability

While this study primarily makes use of the detailed en-ergy balance snow model described above (Cherkauer andLettenmaier 2003) the VIC model has been well validatedwith streamflow observations particularly in the mountain-ous western US (Christensen et al 2004 Hamlet et al2005 Maurer et al 2002) The model has also been usedextensively for streamflow forecasting applications (Hamletand Lettenmaier 1999 Wood et al 2002) and for climatechange assessments (Lettenmaier et al 1999 Christensen etal 2004 Hamlet and Lettenmaier 1999 Payne et al 2004Snover et al 2003)

23 Climate change scenarios

The climatic conditions used to represent a changing cli-mate were constructed from the output of 13 Global Cir-culation Models (GCMs) obtained from the Coupled ModelIntercomparison Project phase 3 (CMIP3) of the World Cli-

Table 2 List of GCMs and relevant references used in this study

Model Origin Reference

BCCR-BCM20 Norway Deque et al (1994)CCMA CGCM31 (T63) Canada Scinocca et al (2008)CNRM CM3 France Salas-Melia et al (2005)CSIROMk35 Australia Gordon et al (2002)GFDL CM21 USA Delworth et al (2006)GISSER USA Russell et al (2000)INM CM30 Russia Diansky and Volodin (2002)IPSL CM4 France IPSL (2005)MIROC32(med res) Japan K-1 model developers (2004)MPI ECHAM5 Germany Jungclaus et al (2006)MRI CDCM232 Japan Yukimoto et al (2001)NCAR CCSM3 USA Collins et al (2006)UKMO HadCM3 UK Gordon et al (2000)

mate Research Program (WCRP) (Meehl et al 2007) (Ta-ble 2) Variables for the VIC model were extracted fromeach model by selecting the grid boxes most closely alignedwith the study area Note that the selected grid boxes oc-cupy slightly different areas and locations between modelsdue to different projection characteristics of each modelrsquosgrid layout Monthly data from the control and the perturbed

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2794 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

(ie climate change) integrations of each model run wereused to calculate the anomalies that were then applied tothe observed subdaily baseline The anomalies includedmonthly changes in precipitation intensity (in relative units)changes in the duration of wet and dry spells changes inmonthly temperature means (in absolute terms) and changesin monthly mean radiation (in absolute terms) Using thesevarious datasets the observed (ie baseline) data were ad-justed to reflect each modelrsquos climate change predictions atmonthly integrations assuming stationarity in sub-monthlyvariability Using monthly deviations from baseline obser-vations five years of synthetic sub-daily weather data weregenerated under a series of future climate scenarios Forthe climate change impact assessment only two time peri-ods were considered 2045ndash2055 (2050s) and 2085ndash2095(2090s) while the period between 1961ndash1990 represented thebaseline conditions For emission scenarios two storylineswere selected from the Special Report on Emission Scenarios(SRES) (IPCC 2000) (1) the low-impact (B1) and (2) high-impact (A2) development scenario Each storyline describesa different world evolving through the 21st century and leadsto different greenhouse gas emission trajectories driven bydemographic economic technological and land-use forcesThe B1 scenario describes a convergent world (where devel-oping and developed nations converge) with rapid changesin economic structures reductions in the intensity resourceuse and the introduction of clean technologies The empha-sis is on global solutions to bring about economic social andenvironmental sustainability including improving equity butwithout additional climate change policies The A2 storylinein contrast describes a differentiated world Economic de-velopment is primarily regional and technological changesare more fragmented in a world of self-reliance and continu-ously increasing population

231 Model calibration and validation

The VIC model was validated for snow covered areas in thenorthern mountainous part of the E-T basin using multi-yearsatellite observations The VIC model was run over the en-tire basin at 18 spatial resolution and the results were com-pared to MODerate Resolution Imaging Spectroradiometer(MODIS) Terra Snow Cover 8-Day composited L3 Global005Deg CMG (MOD10C2) data (Hall et al 2002) for theentire basin This comparison was made for four time peri-ods in January 2001

To validate the SWE output from the VIC model twodifferent observed datasets were used The first datasetcontained SWE observations obtained from the Directorateof Turkish Electrical Works snow observations book (EIEI2007) These observations are obtained roughly one timeper month in mountainous areas of the country and includesnow cover depth density and SWE For the purposes ofthis study data acquired at the station 21K07 (located at3893 N and 4024 E) were used The statistics describing

the strength of the relationship between observed and mod-eled values are given in Table 3

The second comparison involved the use of remotelysensed observations More specifically VIC output from anadditional single run at a site located in the headwaters of theEuphrates river in southeastern Turkey (centered at 405 Elongitude 385 N latitude) were compared to SWE datafrom the 8-day blended Special Sensor Microwave Imager(SSMI) and the MODIS snow cover remote sensing product(Brodzik et al 2007) The SWE dataset was derived fromthe SSMI Pathfinder daily brightness temperatures Thisdataset was enhanced by the MODIS snow cover area prod-uct derived from the MODISTerra Snow Cover 8-Day L3Global 005Deg CMG (MOD10C2) data regridded to theEASE-Grid While spatially coarse these remote sensing-based SWE datasets provide a first-order approximation ofseasonal and interannual changes in SWE

The final form of model evaluation was made by com-paring modeled runoff generated by VICrsquos routing modelagainst the observed runoff data obtained from the Direc-torate of Electrical Works monthly runoff book series (EIEI2008) We used data obtained from the Logmar station (lo-cated at 3852 N and 3949 E) at a monthly time step be-tween 1994 and 2004 The quality of fit between mod-eled and observed discharge was obtained by Nash-Sutcliffemodel efficiency coefficient commonly used to assess thepredictive power of models (Nash and Sutcliffe 1970)

3 Results

31 Expected changes in temperature and precipitation

The deviations from baseline climatic conditions of the 20thcentury (1961ndash1990) predicted by 13 climate models undertwo climate change scenarios (A2 B1) for two time peri-ods are provided as a temperature-precipitation cross-plot inFig 3 The temperature change was computed as the abso-lute difference (inC) between the baseline conditions andthe climate change scenarios thus positive values indicateincreases in air temperature The precipitation change wascomputed as the relative change from baseline conditionsnegative numbers indicate a reduction in precipitation fromthe baseline

In the future climate the mean cold season air tem-peratures show a consistent pattern of increase under allmodelscenarioyear combinations The amount of increaseranges from 07C under the SRES B1 scenario in 2050 togreater than 5C under the SRES A2 scenario at the end ofthe 21st century All models also suggest consistent increasesfrom 2050 to 2090 regardless of the emission scenario

With respect to precipitation most models suggest vari-ous levels of reduction ndash from 5 to 45 percent ndash under allscenarios and all time periods A few models suggest slightincreases in precipitation (up to 20 percent by one model)

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2795

Table 3 Various statistical metrics describing the strength of the relationship between observed and modeled SWE at one location (centeredat 3892 N and 4025 E) The description and the formulas used to derive the metrics are also provided

Statistic Value Description

N 138 number of samplesmod 932 mean modeled SWE (mm monthminus1)obs 1134 mean observed SWE (mm monthminus1)Smod 992 standard deviation of modeled SWE (mm monthminus1)Sobs 1048 standard deviation of observed SWE (mm monthminus1)

MAE 469 mean absolute error1N

nsumi=1

| modi minusobsi |

RMSE 737 root mean squared error[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSES 1064 systematic RMSE[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSEU 989 unsystematic RMSE[Nminus1nsum

i=1( modi minus modi)

2]05

d 086 index of agreementd = 10minus

nsum

i=1( mod iminusobsi )2

nsumi=1

(∣∣ mod iminusobs

∣∣+∣∣obsiminusobs∣∣)2

especially under the B1 scenario in the mid 21st centuryWith the exception of one model (NCAR) no precipitationincrease is suggested under the A2 scenario in either the mid-dle or at the end of the 21st century In general precipitationchange is greater under A2 as much as 25 percent lower thanaverage baseline conditions within each modelrsquos outcomeIn contrast to other models the NCAR model predicts no netchange nor cooler temperatures (by as much as 05C) un-der the B1 scenario but instead predicts a shift to warmingunder the A2 scenario These results suggest that in generalthe climate in the E-T basin will be warmer and dryer withsubsequent implications for the snow processes and availablemelt water

32 Performance of the VIC model

The first set of results illustrates the ability of the VIC modelto accurately simulate SWE in the E-T basin The compari-son between modeled and observed SWE at a single site usedto acquire ground SWE observations shows a strong correla-tion over time (Fig 4) Detailed statistics for this compari-son are given in Table 3 Please see the discussion providedabove for justification of using certain statistical measure-ment variables as suggested by Willmott (1981) When ob-served and modeled SWE data are compared in the form ofa scatter plot (not shown) there is no systematic bias at thesite level as shown in Table 3

A comparison of modeled SWE data to remotely sensedobservations also suggest that the VIC model is able to cor-rectly simulate the available water in accumulated snow Fig-ure 5 displays the comparison between SSMI-observed andVIC-modeled SWE (in mm) in the form of a time-series plot

Fig 3 Cold season (DJFMA) averaged changes in near surfaceair temperature and precipitation from baseline conditions (1961ndash1990) predicted by 13 models for four scenariotime combina-tions Different colors correspond to different models while markershapes indicates the scenariotime pair Positive values in the tem-perature field indicate increases in temperature while negative val-ues in the precipitation field indicate relative decreases in precipita-tion amounts as a result of climate change

between 2001 and 2006 averaged over 8-day intervals Ingeneral there is very good agreement between the two es-timates (r = 083) With a few exceptions the VIC modelaccurately simulates the seasonal and interannual variationof SWE in this part of the basin There is remarkable

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2796 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 4 Comparison of single day SWE observations to modeledquantities at a location in northern part of the E-T basin The siteis located at (3892 N and 4025 E) and includes once-per-monthobservations Sample dates of observed data are provided for refer-ence

Fig 5 A comparison of modeled (red) and observed (black) SWEat a location in southeastern Turkey (405 E longitude 385 N lat-itude) within the Euphrates-Tigris basin for 2001ndash2006 The ob-served data are extracted from the 8-day blended Special SensorMicrowave Imager (SSMI) data (Brodzik et al 2007)

correspondence between the VIC model and the SSMI ob-servations for the timing of accumulation maximum and ab-lation of snow The figure also points to interesting dynam-ics of SWE in the basin First snow accumulation begins inNovember of each year gradually increases to a maximumin late Marchndashearly April with a subsequent sharp declineinto the summer season Another important observation from

45

Figure 6 Comparison of observed (top) and simulated (bottom) snow covered areas in the Euphrates-Tigris basin across

different periods in winter 2001 Snow cover is indicated in white against the color-coded topography of the basin

Observations in the top row were extracted from Terra MODIS 8-day snow cover product (MOD10C2) at 005 degree

spatial resolution while the simulated results of the bottom row were produced in this research using the VIC model

More details on the MODIS snow cover product can be found in Hall et al (2002) Each date at the top indicate the

beginning date of the 8-day compositing period

January252001January172001January92001 February22001

MODIS

VIC

Fig 6 Comparison of observed (top) and simulated (bottom) snowcovered areas in the Euphrates-Tigris basin across different periodsin winter 2001 Snow cover is indicated in white against the color-coded topography of the basin Observations in the top row were ex-tracted from Terra MODIS 8-day snow cover product (MOD10C2)at 005 spatial resolution while the simulated results of the bot-tom row were produced in this research using the VIC model Moredetails on the MODIS snow cover product can be found in Hall etal (2002) Each date at the top indicate the beginning date of the8-day compositing period

Fig 4 is the interannual variability of SWE Both the mod-eled and remote sensing results had larger maximum SWEvalues (around 200 mm) in years 2001 2002 and 2006 hadthan 2004 and 2005 Thus while 2002 and 2003 are largesnow accumulation years (and therefore larger SWE) 2004and 2005 appear to be drier years with much less accumula-tion It is interesting to note that both the VIC model and thesatellite observations capture this variability quite well

Further assessment of the VIC model performance wasmade by visually comparing modeled snow covered areas tosatellite observations of snow cover from the MODIS sen-sor (Fig 6) The map comparison between satellite-observedand modeled snow area data for four dates in January 2001indicate a noticeable agreement Even though VIC predicteda larger area covered by snow on each date especially in theeastern mountainous zone of the basin the general patternof snow cover accumulating in the arc-shaped northern andeastern portions of the basin is remarkably similar betweenthe two data sets One reason for the larger area snow accu-mulation in VIC simulations is the coarse nature of the grid-ded weather data that smooth over the fine scale topographicvariation present in the MODIS data

Comparison of modeled discharge to observed quantitiesat a location in the northeastern portion of the basin furthershows a notable correlation especially in the timing of peakflows that are indicative of timing of snowmelt (Fig 7) Themodeled discharge data were obtained by routing VIC sur-face and baseflow predictions through the routing model ofLohmann et al (1996) The quality of the match between theobserved and modeled discharge determined by the Nash-Sutcliffe coefficient of 05 is moderate although the timingof peak flows is matched nicely indicating that VIC is ableto simulate the accumulation and melting of snow in accept-able form

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2797

Fig 7 Comparison of monthly observed and modeled discharge atthe Logmar station (located at 3851 N and 3949 E) While theabsolute value flow is less than ideal in some cases VIC modelcaptures the timing of peak flow quite well Nash-Sutcliffe (NS)statistic is also provided to show the strength of correlation

33 Changes in SWE for each GCM model outputs

The changes in SWE across 12 of the 13 models are givenin Fig 8 where in each panel deviations from the baselineconditions are depicted as a relative change in months defin-ing the cold season Note that even though the month ofApril is not generally considered part of the cold season inmany northern hemisphere studies it was included in thecurrent study due to its importance for snow ablation andmelting Also shown with text in the figure are the largestabsolute changes in SWE (in mm) for the month in questionThese absolute values reflect the averaged largest single daychanges in SWE in each month The direction of and mag-nitude of these absolute changes must be interpreted in thecontext of relative changes shown in the y-axis

Overall the following patterns emerge from Fig 8 Firstalmost all models predict a decline in SWE in the DecemberndashJanuary period However there are fewer models predictinga decline in SWE in the MarchndashApril period indicating thatthe reliability in predicting a decline in SWE is reduced whenmoving from the DecemberndashJanuary period to the MarchndashApril period Second there is considerable variability acrossthe models For example the GFDL model predicts a strongnegative change in SWE across all months years and emis-sion scenarios while the CNNM model predicts a weak neg-ative response in SWE under a changing climate (less than40 percent) In some cases the models even predict a positiveresponse Across all models the predicted increase althoughrare occurs in the spring during the normal thawing periodin 10 out of 13 models This represents a shift in the timingof snow accumulation moving from the DecemberndashFebruaryperiod to the MarchndashApril period Another interesting obser-

vation concerning the increase in SWE with climate changeis that it occurs more often under the B1 than the A2 sce-nario and more frequently in the middle of the 21st century(2050s) than in the later period (2090s)

There are notable exceptions to these general trends Forinstance the NCAR model does not fit this description re-sults from this model suggest a steady increase in SWE avail-ability under the aggressive A2 scenario at the end of the 21stcentury (although this increase occurs more in April than inJanuary) Several other models including the IPSL and theMIROC models also show similar trends but to a lesser ex-tent However all of these exceptions tend to occur in thespring rather than during the cold season under changing cli-matic conditions

Absolute changes in SWE as depicted by the number be-low bars representing each month also tell a similar story butin different context First the absolute magnitude of SWEchanges is highly variable ranging from less than 20 to over230 mm When placed in context of available snow waterat the peak snow accumulation period (as much as 400 mm)these absolute changes reflect 10 to 60 percent of the avail-able snow water Second the largest changes (and the mostvariation across models) occur in April followed by MarchThe timing of these largest changes in the spring are in con-trast with the large relative changes that occur in the winterperiod Although these largest changes do not necessarilyreflect the largest relative changes they indicate the impor-tance of climate change for accumulation (winter) and decay(spring) of SWE

Given the model-derived variability in predicting climatevariables that are important for snow accumulation the ques-tion arises as to which models or strategies should be used todistill information of use to stake-holders One common ap-proach is simply to average all models This approach com-monly referred to as a multi-model ensemble dilutes modelsthat provide poor outcomes with those that do a good job insimulating a region (Lambert and Boer 2001 Pierce et al2009) Using this approach a new multi-model ensembleclimate forcing dataset was generated for input to the VICmodel By averaging the outcomes of all climate models theensemble results suggest that SWE could decline as muchas 40 percent under all emission scenarios all time periodsand all months considered in this study (Fig 9) Of greaterinterest however is the gradual decline in the reduction inSWE from December (largest decrease) to April (smallestdecrease) These findings suggest that the amount of snowthat help build SWE during the cold season could graduallylessen leaving little or no snowmelt generated flow in thespring As with individual models the absolute changes in-crease from winter to spring

Rather than average the negative or positive changes as inthe ensemble approach it is possible to explore the frequencywith which the models predict positive or negative changesby month In Fig 10 the number of times an increase or de-crease was predicted by a model is plotted where the length

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2798 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 8 Changes in basin-averaged SWE relative to baseline conditions (1961ndash1990) predicted by 12 GCMs for the cold season (DJFMA)Negative values indicate a reduction and positive values indicate an increase in SWE relative to baseline conditions Also shown are themaximum averaged absolute changes in SWE (in units of mmmonth) The sign of the absolute change must be interpreted with the directionof relative change on the y-axis Note that only 12 out of 13 models used in the study are shown

of each bar is equal to 13 representing the total number ofmodels used The first finding from this plot is that all modelsindicate a clear decline in December under changing climateconditions Moreover many of the models predict a declinein SWE January through March Yet only half the modelspredict an increase while the other half predict a decrease inApril and this result is consistent across all emission scenar-ios and time periods This lack of consensus among modelsin April is mostly due to lack of snow in this period In otherwords the amount of reduction on an already small snowquantity in the spring is lost between model variability In

the winter period however all models do predict snow avail-ability but simply less of it under a changing climate

Finally Fig 11 shows the changes in modeled SWE acrosselevation zones (or bands) for four scenariotime combina-tions The elevation bands are given at 250 m incrementson the y-axis while the x-axis depicts the departure of SWEfrom the baseline conditions in relative terms The results re-veal that the largest changes (greater than 50 percent) occurin the lower elevation bands (under 500 m) and as altitudeincreases the reduction in SWE diminishes exponentially toa minimum value of about 10 percent Above 2000 m the

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2799

Fig 8 Continued

predicted decline in SWE is constant with a mean value of10 percent However slight positive bias in VIC-simulatedsnow area and SWE amounts in lower elevations suggeststhat the magnitude of change in lower elevations must beviewed with caution This is also corroborated by lack ofmodel consensus in April which affect lower elevations firstThese findings suggest that as the climate warms and pre-cipitation declines lower elevation zones could experience amore rapid reduction in snow accumulation than the higherelevation zones of the basin but these results are as good asthe quality of the model predictions in the lower areas

4 Discussion

In this research the effects of climate change on theamount of water stored in snowpack in the mountains ofthe Euphrates-Tigris river basin were assessed using the VICmacroscale hydrological model linked to climate model re-sults Inevitably the approach adopted in this study has sev-eral limitations With respect to the climate change predic-tions this study only considered the averages of the regionalclimate system but did not assess changes in its variabil-ity While the approach taken here involving the perturba-tion of observed climate allows for some random variationin day-to-day weather events the true variability includingclimatic extremes is not captured Work on the sensitivity of

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

2792 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

part of a detailed but straightforward snow module that simu-lates many of the snow accumulation and ablation processes(Storck et al 2002) The module is especially well suitedfor simulating mountain snowpack and includes the effectsof forest canopy on snow interception and the attenuationof wind and solar radiation which are important drivers ofsnow accumulation sublimation and melt processes (Storck2000 Storck et al 2002) Fundamentally the partitioningbetween the snow and rain phases of precipitation is deter-mined from a simple scheme based on estimated hourly tem-peratures that are derived from daily minimum and maxi-mum temperatures Belowminus05C precipitation is assumedto fall as 100 snow and above 05C as 100 rain A lin-ear relationship between snow and rain is assumed betweenthese two limits In addition to determining the phase of pre-cipitation temperature also influences other snow accumu-lation and melt processes including convective heat transferfrom the air to the snowpack (eg Storck 2000) attenua-tion of solar radiation and variations in long-wave radiation(determined from the difference betweenTmax and Tmin)and calculation of the dewpoint and vapor pressure deficit(from minimum temperature) that influence snow sublima-tion (Thornton and Running 1999) More specifically thetrends in temperature are indirectly related to trends in long-and shortwave radiation humidity and vapor pressure con-tributing to sublimation and snowmelt in the model Snowprocesses including SWE are modeled separately for a mo-saic of elevation bands and vegetation types in each grid cellNote that when solving for a subgrid tile of certain vegeta-tion type and elevation band VIC assumes that any snowfully covers that tile

For this study a grid layout at 18 intervals (roughly12 km on a side) were set up over the E-T basin For eachgrid cell the VIC model was run in energy balance modewhich solves the complete water balance but also minimizessurface energy balance errors The surface energy balanceis closed through an iterative process which attempts to finda surface temperature that adjusts surface energy fluxes sothat they balance the incoming solar and longwave radiationfluxes Although computationally more expensive this modesimulates the surface energy fluxes that are important to un-derstanding the snow accumulation and ablation processesAmong many hydrologic quantities that the VIC model isable to simulate only the daily SWE predictions for eachgrid cell both as individual elevation bands and as their arealaverages were used as the primary output in this study

To assess the effects of climate change on snow wateravailability the VIC model was forced with gridded fieldsof minimum and maximum temperature precipitation windspeed humidity and solar radiation at sub-daily time stepsTo establish the baseline snow dynamics in the basin min-imum and maximum temperature precipitation solar radia-tion wind speed and humidity on a 6 h time step were ob-tained from the National Centers for Environmental Predic-tion (NCEP) Global Reanalysis project data (Kalnay et al

1996) that were downscaled to 18 resolution using a bilin-ear interpolator To account for topographic effects on snowdynamics not captured at the coarse scale nature of the Re-analysis data correction factors were applied to the tempera-ture (lapsed by 61C kmminus1) and precipitation data followingthe Precipitation Regression on Independent Slopes Method(PRISM) algorithm (Daly et al 1994) Even with these cor-rections the derived forcing data did not exhibit large spa-tial variation within the mountainous portions of the basin(its effects are discussed in later sections) Nevertheless thelimitations in the forcing data may not have significant im-pacts because the snow accumulation and ablation processesin mountainous regions are controlled by the accumulatedprecipitation in snowpacks and thus a lack of spatial vari-ability in the input data may not be important (Hamlet etal 2005) Further evidence that lack of spatial variabilitymay not contribute to large model errors is shown by theVIC modelrsquos ability to capture the temporal and spatial pat-terns of snow covered areas and SWE (Fig 4 through 6)Moreover a comparison of the Reanalysis-derived climaticvariables to those obtained from the National Climatic DataCenter (NCDC) global summary of day reports for five me-teorological stations located in the study area (NCDC 2010)revealed a strong correlation for all forcing variables exceptprecipitation (Table 1) The precipitation correlation was lessthan ideal but may be attributable to an average 100 mm pos-itive bias in the NCEP Reanalysis data

In Table 1 the quantitative comparison between NCEPand station data was achieved using the correlation coeffi-cient(r) mean absolute error (MAE) root mean square er-ror (RMSE) and the index of agreement (d) following Will-mott (1981) The index of agreement is a measure of rela-tive error in model estimates It is a dimensionless numberbetween 00 and 10 where 00 describes complete disagree-ment between estimated and observed values and 10 indi-cates that the simulated and observed values are identicalThe index of agreement is included because the correlationcoefficient(r) cannot account for additive differences or dif-ferences in proportionality (Willmott 1981) Neverthelessr (measure of covariability between observed and modeledvalues) and RMSE (measure of average difference betweenobserved and modeled values) statistics are also included be-cause they describe different components of model error in-dicated byd

In its default configuration the VIC model treats eachgrid cell to have a single elevation value (ie grid cells areflat) In mountainous areas where snow processes are animportant part of the hydrological budget the single eleva-tion assumption can lead to inaccuracies in snow pack es-timates To overcome this issue each grid cell in the E-T basin was broken up into a total of six elevation bands(also called snow bands) and simulated individually Alongwith the PRISM correction described above VIC was usedto lapse the grid cell average temperature pressure and pre-cipitation to a more accurate local estimate based on each

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2793

Table 1 Statistics describing the strength of the relationship between three meteorological forcing variables measured at a single site (locatedat 3796 N and 3963 E) and extracted from the NCEP dataset for the same location The metrics used in comparison and their descriptionand formulas are also provided

var N mod obs Smod Sobs MAE RMSE RMSES RMSEU d

Tmax 13767 211 225 1302 118 296 39 119 1302 097Tmin 13762 64 87 85 895 303 38 925 85 095PREC 13672 16 13 425 409 111 321 41 425 083

Statistic Description

N number of samplesmod mean modeled quantity (mm dminus1 for precipitation andC for temperature)obs mean observed quantity (mm dminus1 for precipitation andC for temperature)Smod standard deviation of modeled quantity (mm dminus1 for precipitation andC for temperature)Sobs standard deviation of observed quantity (mm dminus1 for precipitation andC for temperature)

MAE mean absolute error1N

nsumi=1

| modi minusobsi |

RMSE root mean squared error[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSES systematic RMSE[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSEU unsystematic RMSE[Nminus1nsum

i=1( modi minus modi)

2]05

d index of agreementd = 10minus

nsum

i=1( mod iminusobsi )2

nsumi=1

(∣∣ mod iminusobs

∣∣+∣∣obsiminusobs∣∣)2

bandrsquos mean elevation value As a result VIC has the abilityto predict snow-related variables by elevation band althoughdistribution within each band in non-spatial ndash ie the loca-tion of snow accumulationablation in the elevation band isnot known Aside from the simple climatic downscaling ofthe input climatic drivers and this correction the VIC modeldoes not introduce additional variability

While this study primarily makes use of the detailed en-ergy balance snow model described above (Cherkauer andLettenmaier 2003) the VIC model has been well validatedwith streamflow observations particularly in the mountain-ous western US (Christensen et al 2004 Hamlet et al2005 Maurer et al 2002) The model has also been usedextensively for streamflow forecasting applications (Hamletand Lettenmaier 1999 Wood et al 2002) and for climatechange assessments (Lettenmaier et al 1999 Christensen etal 2004 Hamlet and Lettenmaier 1999 Payne et al 2004Snover et al 2003)

23 Climate change scenarios

The climatic conditions used to represent a changing cli-mate were constructed from the output of 13 Global Cir-culation Models (GCMs) obtained from the Coupled ModelIntercomparison Project phase 3 (CMIP3) of the World Cli-

Table 2 List of GCMs and relevant references used in this study

Model Origin Reference

BCCR-BCM20 Norway Deque et al (1994)CCMA CGCM31 (T63) Canada Scinocca et al (2008)CNRM CM3 France Salas-Melia et al (2005)CSIROMk35 Australia Gordon et al (2002)GFDL CM21 USA Delworth et al (2006)GISSER USA Russell et al (2000)INM CM30 Russia Diansky and Volodin (2002)IPSL CM4 France IPSL (2005)MIROC32(med res) Japan K-1 model developers (2004)MPI ECHAM5 Germany Jungclaus et al (2006)MRI CDCM232 Japan Yukimoto et al (2001)NCAR CCSM3 USA Collins et al (2006)UKMO HadCM3 UK Gordon et al (2000)

mate Research Program (WCRP) (Meehl et al 2007) (Ta-ble 2) Variables for the VIC model were extracted fromeach model by selecting the grid boxes most closely alignedwith the study area Note that the selected grid boxes oc-cupy slightly different areas and locations between modelsdue to different projection characteristics of each modelrsquosgrid layout Monthly data from the control and the perturbed

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2794 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

(ie climate change) integrations of each model run wereused to calculate the anomalies that were then applied tothe observed subdaily baseline The anomalies includedmonthly changes in precipitation intensity (in relative units)changes in the duration of wet and dry spells changes inmonthly temperature means (in absolute terms) and changesin monthly mean radiation (in absolute terms) Using thesevarious datasets the observed (ie baseline) data were ad-justed to reflect each modelrsquos climate change predictions atmonthly integrations assuming stationarity in sub-monthlyvariability Using monthly deviations from baseline obser-vations five years of synthetic sub-daily weather data weregenerated under a series of future climate scenarios Forthe climate change impact assessment only two time peri-ods were considered 2045ndash2055 (2050s) and 2085ndash2095(2090s) while the period between 1961ndash1990 represented thebaseline conditions For emission scenarios two storylineswere selected from the Special Report on Emission Scenarios(SRES) (IPCC 2000) (1) the low-impact (B1) and (2) high-impact (A2) development scenario Each storyline describesa different world evolving through the 21st century and leadsto different greenhouse gas emission trajectories driven bydemographic economic technological and land-use forcesThe B1 scenario describes a convergent world (where devel-oping and developed nations converge) with rapid changesin economic structures reductions in the intensity resourceuse and the introduction of clean technologies The empha-sis is on global solutions to bring about economic social andenvironmental sustainability including improving equity butwithout additional climate change policies The A2 storylinein contrast describes a differentiated world Economic de-velopment is primarily regional and technological changesare more fragmented in a world of self-reliance and continu-ously increasing population

231 Model calibration and validation

The VIC model was validated for snow covered areas in thenorthern mountainous part of the E-T basin using multi-yearsatellite observations The VIC model was run over the en-tire basin at 18 spatial resolution and the results were com-pared to MODerate Resolution Imaging Spectroradiometer(MODIS) Terra Snow Cover 8-Day composited L3 Global005Deg CMG (MOD10C2) data (Hall et al 2002) for theentire basin This comparison was made for four time peri-ods in January 2001

To validate the SWE output from the VIC model twodifferent observed datasets were used The first datasetcontained SWE observations obtained from the Directorateof Turkish Electrical Works snow observations book (EIEI2007) These observations are obtained roughly one timeper month in mountainous areas of the country and includesnow cover depth density and SWE For the purposes ofthis study data acquired at the station 21K07 (located at3893 N and 4024 E) were used The statistics describing

the strength of the relationship between observed and mod-eled values are given in Table 3

The second comparison involved the use of remotelysensed observations More specifically VIC output from anadditional single run at a site located in the headwaters of theEuphrates river in southeastern Turkey (centered at 405 Elongitude 385 N latitude) were compared to SWE datafrom the 8-day blended Special Sensor Microwave Imager(SSMI) and the MODIS snow cover remote sensing product(Brodzik et al 2007) The SWE dataset was derived fromthe SSMI Pathfinder daily brightness temperatures Thisdataset was enhanced by the MODIS snow cover area prod-uct derived from the MODISTerra Snow Cover 8-Day L3Global 005Deg CMG (MOD10C2) data regridded to theEASE-Grid While spatially coarse these remote sensing-based SWE datasets provide a first-order approximation ofseasonal and interannual changes in SWE

The final form of model evaluation was made by com-paring modeled runoff generated by VICrsquos routing modelagainst the observed runoff data obtained from the Direc-torate of Electrical Works monthly runoff book series (EIEI2008) We used data obtained from the Logmar station (lo-cated at 3852 N and 3949 E) at a monthly time step be-tween 1994 and 2004 The quality of fit between mod-eled and observed discharge was obtained by Nash-Sutcliffemodel efficiency coefficient commonly used to assess thepredictive power of models (Nash and Sutcliffe 1970)

3 Results

31 Expected changes in temperature and precipitation

The deviations from baseline climatic conditions of the 20thcentury (1961ndash1990) predicted by 13 climate models undertwo climate change scenarios (A2 B1) for two time peri-ods are provided as a temperature-precipitation cross-plot inFig 3 The temperature change was computed as the abso-lute difference (inC) between the baseline conditions andthe climate change scenarios thus positive values indicateincreases in air temperature The precipitation change wascomputed as the relative change from baseline conditionsnegative numbers indicate a reduction in precipitation fromthe baseline

In the future climate the mean cold season air tem-peratures show a consistent pattern of increase under allmodelscenarioyear combinations The amount of increaseranges from 07C under the SRES B1 scenario in 2050 togreater than 5C under the SRES A2 scenario at the end ofthe 21st century All models also suggest consistent increasesfrom 2050 to 2090 regardless of the emission scenario

With respect to precipitation most models suggest vari-ous levels of reduction ndash from 5 to 45 percent ndash under allscenarios and all time periods A few models suggest slightincreases in precipitation (up to 20 percent by one model)

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2795

Table 3 Various statistical metrics describing the strength of the relationship between observed and modeled SWE at one location (centeredat 3892 N and 4025 E) The description and the formulas used to derive the metrics are also provided

Statistic Value Description

N 138 number of samplesmod 932 mean modeled SWE (mm monthminus1)obs 1134 mean observed SWE (mm monthminus1)Smod 992 standard deviation of modeled SWE (mm monthminus1)Sobs 1048 standard deviation of observed SWE (mm monthminus1)

MAE 469 mean absolute error1N

nsumi=1

| modi minusobsi |

RMSE 737 root mean squared error[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSES 1064 systematic RMSE[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSEU 989 unsystematic RMSE[Nminus1nsum

i=1( modi minus modi)

2]05

d 086 index of agreementd = 10minus

nsum

i=1( mod iminusobsi )2

nsumi=1

(∣∣ mod iminusobs

∣∣+∣∣obsiminusobs∣∣)2

especially under the B1 scenario in the mid 21st centuryWith the exception of one model (NCAR) no precipitationincrease is suggested under the A2 scenario in either the mid-dle or at the end of the 21st century In general precipitationchange is greater under A2 as much as 25 percent lower thanaverage baseline conditions within each modelrsquos outcomeIn contrast to other models the NCAR model predicts no netchange nor cooler temperatures (by as much as 05C) un-der the B1 scenario but instead predicts a shift to warmingunder the A2 scenario These results suggest that in generalthe climate in the E-T basin will be warmer and dryer withsubsequent implications for the snow processes and availablemelt water

32 Performance of the VIC model

The first set of results illustrates the ability of the VIC modelto accurately simulate SWE in the E-T basin The compari-son between modeled and observed SWE at a single site usedto acquire ground SWE observations shows a strong correla-tion over time (Fig 4) Detailed statistics for this compari-son are given in Table 3 Please see the discussion providedabove for justification of using certain statistical measure-ment variables as suggested by Willmott (1981) When ob-served and modeled SWE data are compared in the form ofa scatter plot (not shown) there is no systematic bias at thesite level as shown in Table 3

A comparison of modeled SWE data to remotely sensedobservations also suggest that the VIC model is able to cor-rectly simulate the available water in accumulated snow Fig-ure 5 displays the comparison between SSMI-observed andVIC-modeled SWE (in mm) in the form of a time-series plot

Fig 3 Cold season (DJFMA) averaged changes in near surfaceair temperature and precipitation from baseline conditions (1961ndash1990) predicted by 13 models for four scenariotime combina-tions Different colors correspond to different models while markershapes indicates the scenariotime pair Positive values in the tem-perature field indicate increases in temperature while negative val-ues in the precipitation field indicate relative decreases in precipita-tion amounts as a result of climate change

between 2001 and 2006 averaged over 8-day intervals Ingeneral there is very good agreement between the two es-timates (r = 083) With a few exceptions the VIC modelaccurately simulates the seasonal and interannual variationof SWE in this part of the basin There is remarkable

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2796 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 4 Comparison of single day SWE observations to modeledquantities at a location in northern part of the E-T basin The siteis located at (3892 N and 4025 E) and includes once-per-monthobservations Sample dates of observed data are provided for refer-ence

Fig 5 A comparison of modeled (red) and observed (black) SWEat a location in southeastern Turkey (405 E longitude 385 N lat-itude) within the Euphrates-Tigris basin for 2001ndash2006 The ob-served data are extracted from the 8-day blended Special SensorMicrowave Imager (SSMI) data (Brodzik et al 2007)

correspondence between the VIC model and the SSMI ob-servations for the timing of accumulation maximum and ab-lation of snow The figure also points to interesting dynam-ics of SWE in the basin First snow accumulation begins inNovember of each year gradually increases to a maximumin late Marchndashearly April with a subsequent sharp declineinto the summer season Another important observation from

45

Figure 6 Comparison of observed (top) and simulated (bottom) snow covered areas in the Euphrates-Tigris basin across

different periods in winter 2001 Snow cover is indicated in white against the color-coded topography of the basin

Observations in the top row were extracted from Terra MODIS 8-day snow cover product (MOD10C2) at 005 degree

spatial resolution while the simulated results of the bottom row were produced in this research using the VIC model

More details on the MODIS snow cover product can be found in Hall et al (2002) Each date at the top indicate the

beginning date of the 8-day compositing period

January252001January172001January92001 February22001

MODIS

VIC

Fig 6 Comparison of observed (top) and simulated (bottom) snowcovered areas in the Euphrates-Tigris basin across different periodsin winter 2001 Snow cover is indicated in white against the color-coded topography of the basin Observations in the top row were ex-tracted from Terra MODIS 8-day snow cover product (MOD10C2)at 005 spatial resolution while the simulated results of the bot-tom row were produced in this research using the VIC model Moredetails on the MODIS snow cover product can be found in Hall etal (2002) Each date at the top indicate the beginning date of the8-day compositing period

Fig 4 is the interannual variability of SWE Both the mod-eled and remote sensing results had larger maximum SWEvalues (around 200 mm) in years 2001 2002 and 2006 hadthan 2004 and 2005 Thus while 2002 and 2003 are largesnow accumulation years (and therefore larger SWE) 2004and 2005 appear to be drier years with much less accumula-tion It is interesting to note that both the VIC model and thesatellite observations capture this variability quite well

Further assessment of the VIC model performance wasmade by visually comparing modeled snow covered areas tosatellite observations of snow cover from the MODIS sen-sor (Fig 6) The map comparison between satellite-observedand modeled snow area data for four dates in January 2001indicate a noticeable agreement Even though VIC predicteda larger area covered by snow on each date especially in theeastern mountainous zone of the basin the general patternof snow cover accumulating in the arc-shaped northern andeastern portions of the basin is remarkably similar betweenthe two data sets One reason for the larger area snow accu-mulation in VIC simulations is the coarse nature of the grid-ded weather data that smooth over the fine scale topographicvariation present in the MODIS data

Comparison of modeled discharge to observed quantitiesat a location in the northeastern portion of the basin furthershows a notable correlation especially in the timing of peakflows that are indicative of timing of snowmelt (Fig 7) Themodeled discharge data were obtained by routing VIC sur-face and baseflow predictions through the routing model ofLohmann et al (1996) The quality of the match between theobserved and modeled discharge determined by the Nash-Sutcliffe coefficient of 05 is moderate although the timingof peak flows is matched nicely indicating that VIC is ableto simulate the accumulation and melting of snow in accept-able form

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2797

Fig 7 Comparison of monthly observed and modeled discharge atthe Logmar station (located at 3851 N and 3949 E) While theabsolute value flow is less than ideal in some cases VIC modelcaptures the timing of peak flow quite well Nash-Sutcliffe (NS)statistic is also provided to show the strength of correlation

33 Changes in SWE for each GCM model outputs

The changes in SWE across 12 of the 13 models are givenin Fig 8 where in each panel deviations from the baselineconditions are depicted as a relative change in months defin-ing the cold season Note that even though the month ofApril is not generally considered part of the cold season inmany northern hemisphere studies it was included in thecurrent study due to its importance for snow ablation andmelting Also shown with text in the figure are the largestabsolute changes in SWE (in mm) for the month in questionThese absolute values reflect the averaged largest single daychanges in SWE in each month The direction of and mag-nitude of these absolute changes must be interpreted in thecontext of relative changes shown in the y-axis

Overall the following patterns emerge from Fig 8 Firstalmost all models predict a decline in SWE in the DecemberndashJanuary period However there are fewer models predictinga decline in SWE in the MarchndashApril period indicating thatthe reliability in predicting a decline in SWE is reduced whenmoving from the DecemberndashJanuary period to the MarchndashApril period Second there is considerable variability acrossthe models For example the GFDL model predicts a strongnegative change in SWE across all months years and emis-sion scenarios while the CNNM model predicts a weak neg-ative response in SWE under a changing climate (less than40 percent) In some cases the models even predict a positiveresponse Across all models the predicted increase althoughrare occurs in the spring during the normal thawing periodin 10 out of 13 models This represents a shift in the timingof snow accumulation moving from the DecemberndashFebruaryperiod to the MarchndashApril period Another interesting obser-

vation concerning the increase in SWE with climate changeis that it occurs more often under the B1 than the A2 sce-nario and more frequently in the middle of the 21st century(2050s) than in the later period (2090s)

There are notable exceptions to these general trends Forinstance the NCAR model does not fit this description re-sults from this model suggest a steady increase in SWE avail-ability under the aggressive A2 scenario at the end of the 21stcentury (although this increase occurs more in April than inJanuary) Several other models including the IPSL and theMIROC models also show similar trends but to a lesser ex-tent However all of these exceptions tend to occur in thespring rather than during the cold season under changing cli-matic conditions

Absolute changes in SWE as depicted by the number be-low bars representing each month also tell a similar story butin different context First the absolute magnitude of SWEchanges is highly variable ranging from less than 20 to over230 mm When placed in context of available snow waterat the peak snow accumulation period (as much as 400 mm)these absolute changes reflect 10 to 60 percent of the avail-able snow water Second the largest changes (and the mostvariation across models) occur in April followed by MarchThe timing of these largest changes in the spring are in con-trast with the large relative changes that occur in the winterperiod Although these largest changes do not necessarilyreflect the largest relative changes they indicate the impor-tance of climate change for accumulation (winter) and decay(spring) of SWE

Given the model-derived variability in predicting climatevariables that are important for snow accumulation the ques-tion arises as to which models or strategies should be used todistill information of use to stake-holders One common ap-proach is simply to average all models This approach com-monly referred to as a multi-model ensemble dilutes modelsthat provide poor outcomes with those that do a good job insimulating a region (Lambert and Boer 2001 Pierce et al2009) Using this approach a new multi-model ensembleclimate forcing dataset was generated for input to the VICmodel By averaging the outcomes of all climate models theensemble results suggest that SWE could decline as muchas 40 percent under all emission scenarios all time periodsand all months considered in this study (Fig 9) Of greaterinterest however is the gradual decline in the reduction inSWE from December (largest decrease) to April (smallestdecrease) These findings suggest that the amount of snowthat help build SWE during the cold season could graduallylessen leaving little or no snowmelt generated flow in thespring As with individual models the absolute changes in-crease from winter to spring

Rather than average the negative or positive changes as inthe ensemble approach it is possible to explore the frequencywith which the models predict positive or negative changesby month In Fig 10 the number of times an increase or de-crease was predicted by a model is plotted where the length

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2798 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 8 Changes in basin-averaged SWE relative to baseline conditions (1961ndash1990) predicted by 12 GCMs for the cold season (DJFMA)Negative values indicate a reduction and positive values indicate an increase in SWE relative to baseline conditions Also shown are themaximum averaged absolute changes in SWE (in units of mmmonth) The sign of the absolute change must be interpreted with the directionof relative change on the y-axis Note that only 12 out of 13 models used in the study are shown

of each bar is equal to 13 representing the total number ofmodels used The first finding from this plot is that all modelsindicate a clear decline in December under changing climateconditions Moreover many of the models predict a declinein SWE January through March Yet only half the modelspredict an increase while the other half predict a decrease inApril and this result is consistent across all emission scenar-ios and time periods This lack of consensus among modelsin April is mostly due to lack of snow in this period In otherwords the amount of reduction on an already small snowquantity in the spring is lost between model variability In

the winter period however all models do predict snow avail-ability but simply less of it under a changing climate

Finally Fig 11 shows the changes in modeled SWE acrosselevation zones (or bands) for four scenariotime combina-tions The elevation bands are given at 250 m incrementson the y-axis while the x-axis depicts the departure of SWEfrom the baseline conditions in relative terms The results re-veal that the largest changes (greater than 50 percent) occurin the lower elevation bands (under 500 m) and as altitudeincreases the reduction in SWE diminishes exponentially toa minimum value of about 10 percent Above 2000 m the

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2799

Fig 8 Continued

predicted decline in SWE is constant with a mean value of10 percent However slight positive bias in VIC-simulatedsnow area and SWE amounts in lower elevations suggeststhat the magnitude of change in lower elevations must beviewed with caution This is also corroborated by lack ofmodel consensus in April which affect lower elevations firstThese findings suggest that as the climate warms and pre-cipitation declines lower elevation zones could experience amore rapid reduction in snow accumulation than the higherelevation zones of the basin but these results are as good asthe quality of the model predictions in the lower areas

4 Discussion

In this research the effects of climate change on theamount of water stored in snowpack in the mountains ofthe Euphrates-Tigris river basin were assessed using the VICmacroscale hydrological model linked to climate model re-sults Inevitably the approach adopted in this study has sev-eral limitations With respect to the climate change predic-tions this study only considered the averages of the regionalclimate system but did not assess changes in its variabil-ity While the approach taken here involving the perturba-tion of observed climate allows for some random variationin day-to-day weather events the true variability includingclimatic extremes is not captured Work on the sensitivity of

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2793

Table 1 Statistics describing the strength of the relationship between three meteorological forcing variables measured at a single site (locatedat 3796 N and 3963 E) and extracted from the NCEP dataset for the same location The metrics used in comparison and their descriptionand formulas are also provided

var N mod obs Smod Sobs MAE RMSE RMSES RMSEU d

Tmax 13767 211 225 1302 118 296 39 119 1302 097Tmin 13762 64 87 85 895 303 38 925 85 095PREC 13672 16 13 425 409 111 321 41 425 083

Statistic Description

N number of samplesmod mean modeled quantity (mm dminus1 for precipitation andC for temperature)obs mean observed quantity (mm dminus1 for precipitation andC for temperature)Smod standard deviation of modeled quantity (mm dminus1 for precipitation andC for temperature)Sobs standard deviation of observed quantity (mm dminus1 for precipitation andC for temperature)

MAE mean absolute error1N

nsumi=1

| modi minusobsi |

RMSE root mean squared error[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSES systematic RMSE[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSEU unsystematic RMSE[Nminus1nsum

i=1( modi minus modi)

2]05

d index of agreementd = 10minus

nsum

i=1( mod iminusobsi )2

nsumi=1

(∣∣ mod iminusobs

∣∣+∣∣obsiminusobs∣∣)2

bandrsquos mean elevation value As a result VIC has the abilityto predict snow-related variables by elevation band althoughdistribution within each band in non-spatial ndash ie the loca-tion of snow accumulationablation in the elevation band isnot known Aside from the simple climatic downscaling ofthe input climatic drivers and this correction the VIC modeldoes not introduce additional variability

While this study primarily makes use of the detailed en-ergy balance snow model described above (Cherkauer andLettenmaier 2003) the VIC model has been well validatedwith streamflow observations particularly in the mountain-ous western US (Christensen et al 2004 Hamlet et al2005 Maurer et al 2002) The model has also been usedextensively for streamflow forecasting applications (Hamletand Lettenmaier 1999 Wood et al 2002) and for climatechange assessments (Lettenmaier et al 1999 Christensen etal 2004 Hamlet and Lettenmaier 1999 Payne et al 2004Snover et al 2003)

23 Climate change scenarios

The climatic conditions used to represent a changing cli-mate were constructed from the output of 13 Global Cir-culation Models (GCMs) obtained from the Coupled ModelIntercomparison Project phase 3 (CMIP3) of the World Cli-

Table 2 List of GCMs and relevant references used in this study

Model Origin Reference

BCCR-BCM20 Norway Deque et al (1994)CCMA CGCM31 (T63) Canada Scinocca et al (2008)CNRM CM3 France Salas-Melia et al (2005)CSIROMk35 Australia Gordon et al (2002)GFDL CM21 USA Delworth et al (2006)GISSER USA Russell et al (2000)INM CM30 Russia Diansky and Volodin (2002)IPSL CM4 France IPSL (2005)MIROC32(med res) Japan K-1 model developers (2004)MPI ECHAM5 Germany Jungclaus et al (2006)MRI CDCM232 Japan Yukimoto et al (2001)NCAR CCSM3 USA Collins et al (2006)UKMO HadCM3 UK Gordon et al (2000)

mate Research Program (WCRP) (Meehl et al 2007) (Ta-ble 2) Variables for the VIC model were extracted fromeach model by selecting the grid boxes most closely alignedwith the study area Note that the selected grid boxes oc-cupy slightly different areas and locations between modelsdue to different projection characteristics of each modelrsquosgrid layout Monthly data from the control and the perturbed

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2794 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

(ie climate change) integrations of each model run wereused to calculate the anomalies that were then applied tothe observed subdaily baseline The anomalies includedmonthly changes in precipitation intensity (in relative units)changes in the duration of wet and dry spells changes inmonthly temperature means (in absolute terms) and changesin monthly mean radiation (in absolute terms) Using thesevarious datasets the observed (ie baseline) data were ad-justed to reflect each modelrsquos climate change predictions atmonthly integrations assuming stationarity in sub-monthlyvariability Using monthly deviations from baseline obser-vations five years of synthetic sub-daily weather data weregenerated under a series of future climate scenarios Forthe climate change impact assessment only two time peri-ods were considered 2045ndash2055 (2050s) and 2085ndash2095(2090s) while the period between 1961ndash1990 represented thebaseline conditions For emission scenarios two storylineswere selected from the Special Report on Emission Scenarios(SRES) (IPCC 2000) (1) the low-impact (B1) and (2) high-impact (A2) development scenario Each storyline describesa different world evolving through the 21st century and leadsto different greenhouse gas emission trajectories driven bydemographic economic technological and land-use forcesThe B1 scenario describes a convergent world (where devel-oping and developed nations converge) with rapid changesin economic structures reductions in the intensity resourceuse and the introduction of clean technologies The empha-sis is on global solutions to bring about economic social andenvironmental sustainability including improving equity butwithout additional climate change policies The A2 storylinein contrast describes a differentiated world Economic de-velopment is primarily regional and technological changesare more fragmented in a world of self-reliance and continu-ously increasing population

231 Model calibration and validation

The VIC model was validated for snow covered areas in thenorthern mountainous part of the E-T basin using multi-yearsatellite observations The VIC model was run over the en-tire basin at 18 spatial resolution and the results were com-pared to MODerate Resolution Imaging Spectroradiometer(MODIS) Terra Snow Cover 8-Day composited L3 Global005Deg CMG (MOD10C2) data (Hall et al 2002) for theentire basin This comparison was made for four time peri-ods in January 2001

To validate the SWE output from the VIC model twodifferent observed datasets were used The first datasetcontained SWE observations obtained from the Directorateof Turkish Electrical Works snow observations book (EIEI2007) These observations are obtained roughly one timeper month in mountainous areas of the country and includesnow cover depth density and SWE For the purposes ofthis study data acquired at the station 21K07 (located at3893 N and 4024 E) were used The statistics describing

the strength of the relationship between observed and mod-eled values are given in Table 3

The second comparison involved the use of remotelysensed observations More specifically VIC output from anadditional single run at a site located in the headwaters of theEuphrates river in southeastern Turkey (centered at 405 Elongitude 385 N latitude) were compared to SWE datafrom the 8-day blended Special Sensor Microwave Imager(SSMI) and the MODIS snow cover remote sensing product(Brodzik et al 2007) The SWE dataset was derived fromthe SSMI Pathfinder daily brightness temperatures Thisdataset was enhanced by the MODIS snow cover area prod-uct derived from the MODISTerra Snow Cover 8-Day L3Global 005Deg CMG (MOD10C2) data regridded to theEASE-Grid While spatially coarse these remote sensing-based SWE datasets provide a first-order approximation ofseasonal and interannual changes in SWE

The final form of model evaluation was made by com-paring modeled runoff generated by VICrsquos routing modelagainst the observed runoff data obtained from the Direc-torate of Electrical Works monthly runoff book series (EIEI2008) We used data obtained from the Logmar station (lo-cated at 3852 N and 3949 E) at a monthly time step be-tween 1994 and 2004 The quality of fit between mod-eled and observed discharge was obtained by Nash-Sutcliffemodel efficiency coefficient commonly used to assess thepredictive power of models (Nash and Sutcliffe 1970)

3 Results

31 Expected changes in temperature and precipitation

The deviations from baseline climatic conditions of the 20thcentury (1961ndash1990) predicted by 13 climate models undertwo climate change scenarios (A2 B1) for two time peri-ods are provided as a temperature-precipitation cross-plot inFig 3 The temperature change was computed as the abso-lute difference (inC) between the baseline conditions andthe climate change scenarios thus positive values indicateincreases in air temperature The precipitation change wascomputed as the relative change from baseline conditionsnegative numbers indicate a reduction in precipitation fromthe baseline

In the future climate the mean cold season air tem-peratures show a consistent pattern of increase under allmodelscenarioyear combinations The amount of increaseranges from 07C under the SRES B1 scenario in 2050 togreater than 5C under the SRES A2 scenario at the end ofthe 21st century All models also suggest consistent increasesfrom 2050 to 2090 regardless of the emission scenario

With respect to precipitation most models suggest vari-ous levels of reduction ndash from 5 to 45 percent ndash under allscenarios and all time periods A few models suggest slightincreases in precipitation (up to 20 percent by one model)

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2795

Table 3 Various statistical metrics describing the strength of the relationship between observed and modeled SWE at one location (centeredat 3892 N and 4025 E) The description and the formulas used to derive the metrics are also provided

Statistic Value Description

N 138 number of samplesmod 932 mean modeled SWE (mm monthminus1)obs 1134 mean observed SWE (mm monthminus1)Smod 992 standard deviation of modeled SWE (mm monthminus1)Sobs 1048 standard deviation of observed SWE (mm monthminus1)

MAE 469 mean absolute error1N

nsumi=1

| modi minusobsi |

RMSE 737 root mean squared error[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSES 1064 systematic RMSE[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSEU 989 unsystematic RMSE[Nminus1nsum

i=1( modi minus modi)

2]05

d 086 index of agreementd = 10minus

nsum

i=1( mod iminusobsi )2

nsumi=1

(∣∣ mod iminusobs

∣∣+∣∣obsiminusobs∣∣)2

especially under the B1 scenario in the mid 21st centuryWith the exception of one model (NCAR) no precipitationincrease is suggested under the A2 scenario in either the mid-dle or at the end of the 21st century In general precipitationchange is greater under A2 as much as 25 percent lower thanaverage baseline conditions within each modelrsquos outcomeIn contrast to other models the NCAR model predicts no netchange nor cooler temperatures (by as much as 05C) un-der the B1 scenario but instead predicts a shift to warmingunder the A2 scenario These results suggest that in generalthe climate in the E-T basin will be warmer and dryer withsubsequent implications for the snow processes and availablemelt water

32 Performance of the VIC model

The first set of results illustrates the ability of the VIC modelto accurately simulate SWE in the E-T basin The compari-son between modeled and observed SWE at a single site usedto acquire ground SWE observations shows a strong correla-tion over time (Fig 4) Detailed statistics for this compari-son are given in Table 3 Please see the discussion providedabove for justification of using certain statistical measure-ment variables as suggested by Willmott (1981) When ob-served and modeled SWE data are compared in the form ofa scatter plot (not shown) there is no systematic bias at thesite level as shown in Table 3

A comparison of modeled SWE data to remotely sensedobservations also suggest that the VIC model is able to cor-rectly simulate the available water in accumulated snow Fig-ure 5 displays the comparison between SSMI-observed andVIC-modeled SWE (in mm) in the form of a time-series plot

Fig 3 Cold season (DJFMA) averaged changes in near surfaceair temperature and precipitation from baseline conditions (1961ndash1990) predicted by 13 models for four scenariotime combina-tions Different colors correspond to different models while markershapes indicates the scenariotime pair Positive values in the tem-perature field indicate increases in temperature while negative val-ues in the precipitation field indicate relative decreases in precipita-tion amounts as a result of climate change

between 2001 and 2006 averaged over 8-day intervals Ingeneral there is very good agreement between the two es-timates (r = 083) With a few exceptions the VIC modelaccurately simulates the seasonal and interannual variationof SWE in this part of the basin There is remarkable

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2796 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 4 Comparison of single day SWE observations to modeledquantities at a location in northern part of the E-T basin The siteis located at (3892 N and 4025 E) and includes once-per-monthobservations Sample dates of observed data are provided for refer-ence

Fig 5 A comparison of modeled (red) and observed (black) SWEat a location in southeastern Turkey (405 E longitude 385 N lat-itude) within the Euphrates-Tigris basin for 2001ndash2006 The ob-served data are extracted from the 8-day blended Special SensorMicrowave Imager (SSMI) data (Brodzik et al 2007)

correspondence between the VIC model and the SSMI ob-servations for the timing of accumulation maximum and ab-lation of snow The figure also points to interesting dynam-ics of SWE in the basin First snow accumulation begins inNovember of each year gradually increases to a maximumin late Marchndashearly April with a subsequent sharp declineinto the summer season Another important observation from

45

Figure 6 Comparison of observed (top) and simulated (bottom) snow covered areas in the Euphrates-Tigris basin across

different periods in winter 2001 Snow cover is indicated in white against the color-coded topography of the basin

Observations in the top row were extracted from Terra MODIS 8-day snow cover product (MOD10C2) at 005 degree

spatial resolution while the simulated results of the bottom row were produced in this research using the VIC model

More details on the MODIS snow cover product can be found in Hall et al (2002) Each date at the top indicate the

beginning date of the 8-day compositing period

January252001January172001January92001 February22001

MODIS

VIC

Fig 6 Comparison of observed (top) and simulated (bottom) snowcovered areas in the Euphrates-Tigris basin across different periodsin winter 2001 Snow cover is indicated in white against the color-coded topography of the basin Observations in the top row were ex-tracted from Terra MODIS 8-day snow cover product (MOD10C2)at 005 spatial resolution while the simulated results of the bot-tom row were produced in this research using the VIC model Moredetails on the MODIS snow cover product can be found in Hall etal (2002) Each date at the top indicate the beginning date of the8-day compositing period

Fig 4 is the interannual variability of SWE Both the mod-eled and remote sensing results had larger maximum SWEvalues (around 200 mm) in years 2001 2002 and 2006 hadthan 2004 and 2005 Thus while 2002 and 2003 are largesnow accumulation years (and therefore larger SWE) 2004and 2005 appear to be drier years with much less accumula-tion It is interesting to note that both the VIC model and thesatellite observations capture this variability quite well

Further assessment of the VIC model performance wasmade by visually comparing modeled snow covered areas tosatellite observations of snow cover from the MODIS sen-sor (Fig 6) The map comparison between satellite-observedand modeled snow area data for four dates in January 2001indicate a noticeable agreement Even though VIC predicteda larger area covered by snow on each date especially in theeastern mountainous zone of the basin the general patternof snow cover accumulating in the arc-shaped northern andeastern portions of the basin is remarkably similar betweenthe two data sets One reason for the larger area snow accu-mulation in VIC simulations is the coarse nature of the grid-ded weather data that smooth over the fine scale topographicvariation present in the MODIS data

Comparison of modeled discharge to observed quantitiesat a location in the northeastern portion of the basin furthershows a notable correlation especially in the timing of peakflows that are indicative of timing of snowmelt (Fig 7) Themodeled discharge data were obtained by routing VIC sur-face and baseflow predictions through the routing model ofLohmann et al (1996) The quality of the match between theobserved and modeled discharge determined by the Nash-Sutcliffe coefficient of 05 is moderate although the timingof peak flows is matched nicely indicating that VIC is ableto simulate the accumulation and melting of snow in accept-able form

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2797

Fig 7 Comparison of monthly observed and modeled discharge atthe Logmar station (located at 3851 N and 3949 E) While theabsolute value flow is less than ideal in some cases VIC modelcaptures the timing of peak flow quite well Nash-Sutcliffe (NS)statistic is also provided to show the strength of correlation

33 Changes in SWE for each GCM model outputs

The changes in SWE across 12 of the 13 models are givenin Fig 8 where in each panel deviations from the baselineconditions are depicted as a relative change in months defin-ing the cold season Note that even though the month ofApril is not generally considered part of the cold season inmany northern hemisphere studies it was included in thecurrent study due to its importance for snow ablation andmelting Also shown with text in the figure are the largestabsolute changes in SWE (in mm) for the month in questionThese absolute values reflect the averaged largest single daychanges in SWE in each month The direction of and mag-nitude of these absolute changes must be interpreted in thecontext of relative changes shown in the y-axis

Overall the following patterns emerge from Fig 8 Firstalmost all models predict a decline in SWE in the DecemberndashJanuary period However there are fewer models predictinga decline in SWE in the MarchndashApril period indicating thatthe reliability in predicting a decline in SWE is reduced whenmoving from the DecemberndashJanuary period to the MarchndashApril period Second there is considerable variability acrossthe models For example the GFDL model predicts a strongnegative change in SWE across all months years and emis-sion scenarios while the CNNM model predicts a weak neg-ative response in SWE under a changing climate (less than40 percent) In some cases the models even predict a positiveresponse Across all models the predicted increase althoughrare occurs in the spring during the normal thawing periodin 10 out of 13 models This represents a shift in the timingof snow accumulation moving from the DecemberndashFebruaryperiod to the MarchndashApril period Another interesting obser-

vation concerning the increase in SWE with climate changeis that it occurs more often under the B1 than the A2 sce-nario and more frequently in the middle of the 21st century(2050s) than in the later period (2090s)

There are notable exceptions to these general trends Forinstance the NCAR model does not fit this description re-sults from this model suggest a steady increase in SWE avail-ability under the aggressive A2 scenario at the end of the 21stcentury (although this increase occurs more in April than inJanuary) Several other models including the IPSL and theMIROC models also show similar trends but to a lesser ex-tent However all of these exceptions tend to occur in thespring rather than during the cold season under changing cli-matic conditions

Absolute changes in SWE as depicted by the number be-low bars representing each month also tell a similar story butin different context First the absolute magnitude of SWEchanges is highly variable ranging from less than 20 to over230 mm When placed in context of available snow waterat the peak snow accumulation period (as much as 400 mm)these absolute changes reflect 10 to 60 percent of the avail-able snow water Second the largest changes (and the mostvariation across models) occur in April followed by MarchThe timing of these largest changes in the spring are in con-trast with the large relative changes that occur in the winterperiod Although these largest changes do not necessarilyreflect the largest relative changes they indicate the impor-tance of climate change for accumulation (winter) and decay(spring) of SWE

Given the model-derived variability in predicting climatevariables that are important for snow accumulation the ques-tion arises as to which models or strategies should be used todistill information of use to stake-holders One common ap-proach is simply to average all models This approach com-monly referred to as a multi-model ensemble dilutes modelsthat provide poor outcomes with those that do a good job insimulating a region (Lambert and Boer 2001 Pierce et al2009) Using this approach a new multi-model ensembleclimate forcing dataset was generated for input to the VICmodel By averaging the outcomes of all climate models theensemble results suggest that SWE could decline as muchas 40 percent under all emission scenarios all time periodsand all months considered in this study (Fig 9) Of greaterinterest however is the gradual decline in the reduction inSWE from December (largest decrease) to April (smallestdecrease) These findings suggest that the amount of snowthat help build SWE during the cold season could graduallylessen leaving little or no snowmelt generated flow in thespring As with individual models the absolute changes in-crease from winter to spring

Rather than average the negative or positive changes as inthe ensemble approach it is possible to explore the frequencywith which the models predict positive or negative changesby month In Fig 10 the number of times an increase or de-crease was predicted by a model is plotted where the length

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2798 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 8 Changes in basin-averaged SWE relative to baseline conditions (1961ndash1990) predicted by 12 GCMs for the cold season (DJFMA)Negative values indicate a reduction and positive values indicate an increase in SWE relative to baseline conditions Also shown are themaximum averaged absolute changes in SWE (in units of mmmonth) The sign of the absolute change must be interpreted with the directionof relative change on the y-axis Note that only 12 out of 13 models used in the study are shown

of each bar is equal to 13 representing the total number ofmodels used The first finding from this plot is that all modelsindicate a clear decline in December under changing climateconditions Moreover many of the models predict a declinein SWE January through March Yet only half the modelspredict an increase while the other half predict a decrease inApril and this result is consistent across all emission scenar-ios and time periods This lack of consensus among modelsin April is mostly due to lack of snow in this period In otherwords the amount of reduction on an already small snowquantity in the spring is lost between model variability In

the winter period however all models do predict snow avail-ability but simply less of it under a changing climate

Finally Fig 11 shows the changes in modeled SWE acrosselevation zones (or bands) for four scenariotime combina-tions The elevation bands are given at 250 m incrementson the y-axis while the x-axis depicts the departure of SWEfrom the baseline conditions in relative terms The results re-veal that the largest changes (greater than 50 percent) occurin the lower elevation bands (under 500 m) and as altitudeincreases the reduction in SWE diminishes exponentially toa minimum value of about 10 percent Above 2000 m the

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2799

Fig 8 Continued

predicted decline in SWE is constant with a mean value of10 percent However slight positive bias in VIC-simulatedsnow area and SWE amounts in lower elevations suggeststhat the magnitude of change in lower elevations must beviewed with caution This is also corroborated by lack ofmodel consensus in April which affect lower elevations firstThese findings suggest that as the climate warms and pre-cipitation declines lower elevation zones could experience amore rapid reduction in snow accumulation than the higherelevation zones of the basin but these results are as good asthe quality of the model predictions in the lower areas

4 Discussion

In this research the effects of climate change on theamount of water stored in snowpack in the mountains ofthe Euphrates-Tigris river basin were assessed using the VICmacroscale hydrological model linked to climate model re-sults Inevitably the approach adopted in this study has sev-eral limitations With respect to the climate change predic-tions this study only considered the averages of the regionalclimate system but did not assess changes in its variabil-ity While the approach taken here involving the perturba-tion of observed climate allows for some random variationin day-to-day weather events the true variability includingclimatic extremes is not captured Work on the sensitivity of

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

2794 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

(ie climate change) integrations of each model run wereused to calculate the anomalies that were then applied tothe observed subdaily baseline The anomalies includedmonthly changes in precipitation intensity (in relative units)changes in the duration of wet and dry spells changes inmonthly temperature means (in absolute terms) and changesin monthly mean radiation (in absolute terms) Using thesevarious datasets the observed (ie baseline) data were ad-justed to reflect each modelrsquos climate change predictions atmonthly integrations assuming stationarity in sub-monthlyvariability Using monthly deviations from baseline obser-vations five years of synthetic sub-daily weather data weregenerated under a series of future climate scenarios Forthe climate change impact assessment only two time peri-ods were considered 2045ndash2055 (2050s) and 2085ndash2095(2090s) while the period between 1961ndash1990 represented thebaseline conditions For emission scenarios two storylineswere selected from the Special Report on Emission Scenarios(SRES) (IPCC 2000) (1) the low-impact (B1) and (2) high-impact (A2) development scenario Each storyline describesa different world evolving through the 21st century and leadsto different greenhouse gas emission trajectories driven bydemographic economic technological and land-use forcesThe B1 scenario describes a convergent world (where devel-oping and developed nations converge) with rapid changesin economic structures reductions in the intensity resourceuse and the introduction of clean technologies The empha-sis is on global solutions to bring about economic social andenvironmental sustainability including improving equity butwithout additional climate change policies The A2 storylinein contrast describes a differentiated world Economic de-velopment is primarily regional and technological changesare more fragmented in a world of self-reliance and continu-ously increasing population

231 Model calibration and validation

The VIC model was validated for snow covered areas in thenorthern mountainous part of the E-T basin using multi-yearsatellite observations The VIC model was run over the en-tire basin at 18 spatial resolution and the results were com-pared to MODerate Resolution Imaging Spectroradiometer(MODIS) Terra Snow Cover 8-Day composited L3 Global005Deg CMG (MOD10C2) data (Hall et al 2002) for theentire basin This comparison was made for four time peri-ods in January 2001

To validate the SWE output from the VIC model twodifferent observed datasets were used The first datasetcontained SWE observations obtained from the Directorateof Turkish Electrical Works snow observations book (EIEI2007) These observations are obtained roughly one timeper month in mountainous areas of the country and includesnow cover depth density and SWE For the purposes ofthis study data acquired at the station 21K07 (located at3893 N and 4024 E) were used The statistics describing

the strength of the relationship between observed and mod-eled values are given in Table 3

The second comparison involved the use of remotelysensed observations More specifically VIC output from anadditional single run at a site located in the headwaters of theEuphrates river in southeastern Turkey (centered at 405 Elongitude 385 N latitude) were compared to SWE datafrom the 8-day blended Special Sensor Microwave Imager(SSMI) and the MODIS snow cover remote sensing product(Brodzik et al 2007) The SWE dataset was derived fromthe SSMI Pathfinder daily brightness temperatures Thisdataset was enhanced by the MODIS snow cover area prod-uct derived from the MODISTerra Snow Cover 8-Day L3Global 005Deg CMG (MOD10C2) data regridded to theEASE-Grid While spatially coarse these remote sensing-based SWE datasets provide a first-order approximation ofseasonal and interannual changes in SWE

The final form of model evaluation was made by com-paring modeled runoff generated by VICrsquos routing modelagainst the observed runoff data obtained from the Direc-torate of Electrical Works monthly runoff book series (EIEI2008) We used data obtained from the Logmar station (lo-cated at 3852 N and 3949 E) at a monthly time step be-tween 1994 and 2004 The quality of fit between mod-eled and observed discharge was obtained by Nash-Sutcliffemodel efficiency coefficient commonly used to assess thepredictive power of models (Nash and Sutcliffe 1970)

3 Results

31 Expected changes in temperature and precipitation

The deviations from baseline climatic conditions of the 20thcentury (1961ndash1990) predicted by 13 climate models undertwo climate change scenarios (A2 B1) for two time peri-ods are provided as a temperature-precipitation cross-plot inFig 3 The temperature change was computed as the abso-lute difference (inC) between the baseline conditions andthe climate change scenarios thus positive values indicateincreases in air temperature The precipitation change wascomputed as the relative change from baseline conditionsnegative numbers indicate a reduction in precipitation fromthe baseline

In the future climate the mean cold season air tem-peratures show a consistent pattern of increase under allmodelscenarioyear combinations The amount of increaseranges from 07C under the SRES B1 scenario in 2050 togreater than 5C under the SRES A2 scenario at the end ofthe 21st century All models also suggest consistent increasesfrom 2050 to 2090 regardless of the emission scenario

With respect to precipitation most models suggest vari-ous levels of reduction ndash from 5 to 45 percent ndash under allscenarios and all time periods A few models suggest slightincreases in precipitation (up to 20 percent by one model)

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2795

Table 3 Various statistical metrics describing the strength of the relationship between observed and modeled SWE at one location (centeredat 3892 N and 4025 E) The description and the formulas used to derive the metrics are also provided

Statistic Value Description

N 138 number of samplesmod 932 mean modeled SWE (mm monthminus1)obs 1134 mean observed SWE (mm monthminus1)Smod 992 standard deviation of modeled SWE (mm monthminus1)Sobs 1048 standard deviation of observed SWE (mm monthminus1)

MAE 469 mean absolute error1N

nsumi=1

| modi minusobsi |

RMSE 737 root mean squared error[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSES 1064 systematic RMSE[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSEU 989 unsystematic RMSE[Nminus1nsum

i=1( modi minus modi)

2]05

d 086 index of agreementd = 10minus

nsum

i=1( mod iminusobsi )2

nsumi=1

(∣∣ mod iminusobs

∣∣+∣∣obsiminusobs∣∣)2

especially under the B1 scenario in the mid 21st centuryWith the exception of one model (NCAR) no precipitationincrease is suggested under the A2 scenario in either the mid-dle or at the end of the 21st century In general precipitationchange is greater under A2 as much as 25 percent lower thanaverage baseline conditions within each modelrsquos outcomeIn contrast to other models the NCAR model predicts no netchange nor cooler temperatures (by as much as 05C) un-der the B1 scenario but instead predicts a shift to warmingunder the A2 scenario These results suggest that in generalthe climate in the E-T basin will be warmer and dryer withsubsequent implications for the snow processes and availablemelt water

32 Performance of the VIC model

The first set of results illustrates the ability of the VIC modelto accurately simulate SWE in the E-T basin The compari-son between modeled and observed SWE at a single site usedto acquire ground SWE observations shows a strong correla-tion over time (Fig 4) Detailed statistics for this compari-son are given in Table 3 Please see the discussion providedabove for justification of using certain statistical measure-ment variables as suggested by Willmott (1981) When ob-served and modeled SWE data are compared in the form ofa scatter plot (not shown) there is no systematic bias at thesite level as shown in Table 3

A comparison of modeled SWE data to remotely sensedobservations also suggest that the VIC model is able to cor-rectly simulate the available water in accumulated snow Fig-ure 5 displays the comparison between SSMI-observed andVIC-modeled SWE (in mm) in the form of a time-series plot

Fig 3 Cold season (DJFMA) averaged changes in near surfaceair temperature and precipitation from baseline conditions (1961ndash1990) predicted by 13 models for four scenariotime combina-tions Different colors correspond to different models while markershapes indicates the scenariotime pair Positive values in the tem-perature field indicate increases in temperature while negative val-ues in the precipitation field indicate relative decreases in precipita-tion amounts as a result of climate change

between 2001 and 2006 averaged over 8-day intervals Ingeneral there is very good agreement between the two es-timates (r = 083) With a few exceptions the VIC modelaccurately simulates the seasonal and interannual variationof SWE in this part of the basin There is remarkable

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2796 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 4 Comparison of single day SWE observations to modeledquantities at a location in northern part of the E-T basin The siteis located at (3892 N and 4025 E) and includes once-per-monthobservations Sample dates of observed data are provided for refer-ence

Fig 5 A comparison of modeled (red) and observed (black) SWEat a location in southeastern Turkey (405 E longitude 385 N lat-itude) within the Euphrates-Tigris basin for 2001ndash2006 The ob-served data are extracted from the 8-day blended Special SensorMicrowave Imager (SSMI) data (Brodzik et al 2007)

correspondence between the VIC model and the SSMI ob-servations for the timing of accumulation maximum and ab-lation of snow The figure also points to interesting dynam-ics of SWE in the basin First snow accumulation begins inNovember of each year gradually increases to a maximumin late Marchndashearly April with a subsequent sharp declineinto the summer season Another important observation from

45

Figure 6 Comparison of observed (top) and simulated (bottom) snow covered areas in the Euphrates-Tigris basin across

different periods in winter 2001 Snow cover is indicated in white against the color-coded topography of the basin

Observations in the top row were extracted from Terra MODIS 8-day snow cover product (MOD10C2) at 005 degree

spatial resolution while the simulated results of the bottom row were produced in this research using the VIC model

More details on the MODIS snow cover product can be found in Hall et al (2002) Each date at the top indicate the

beginning date of the 8-day compositing period

January252001January172001January92001 February22001

MODIS

VIC

Fig 6 Comparison of observed (top) and simulated (bottom) snowcovered areas in the Euphrates-Tigris basin across different periodsin winter 2001 Snow cover is indicated in white against the color-coded topography of the basin Observations in the top row were ex-tracted from Terra MODIS 8-day snow cover product (MOD10C2)at 005 spatial resolution while the simulated results of the bot-tom row were produced in this research using the VIC model Moredetails on the MODIS snow cover product can be found in Hall etal (2002) Each date at the top indicate the beginning date of the8-day compositing period

Fig 4 is the interannual variability of SWE Both the mod-eled and remote sensing results had larger maximum SWEvalues (around 200 mm) in years 2001 2002 and 2006 hadthan 2004 and 2005 Thus while 2002 and 2003 are largesnow accumulation years (and therefore larger SWE) 2004and 2005 appear to be drier years with much less accumula-tion It is interesting to note that both the VIC model and thesatellite observations capture this variability quite well

Further assessment of the VIC model performance wasmade by visually comparing modeled snow covered areas tosatellite observations of snow cover from the MODIS sen-sor (Fig 6) The map comparison between satellite-observedand modeled snow area data for four dates in January 2001indicate a noticeable agreement Even though VIC predicteda larger area covered by snow on each date especially in theeastern mountainous zone of the basin the general patternof snow cover accumulating in the arc-shaped northern andeastern portions of the basin is remarkably similar betweenthe two data sets One reason for the larger area snow accu-mulation in VIC simulations is the coarse nature of the grid-ded weather data that smooth over the fine scale topographicvariation present in the MODIS data

Comparison of modeled discharge to observed quantitiesat a location in the northeastern portion of the basin furthershows a notable correlation especially in the timing of peakflows that are indicative of timing of snowmelt (Fig 7) Themodeled discharge data were obtained by routing VIC sur-face and baseflow predictions through the routing model ofLohmann et al (1996) The quality of the match between theobserved and modeled discharge determined by the Nash-Sutcliffe coefficient of 05 is moderate although the timingof peak flows is matched nicely indicating that VIC is ableto simulate the accumulation and melting of snow in accept-able form

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2797

Fig 7 Comparison of monthly observed and modeled discharge atthe Logmar station (located at 3851 N and 3949 E) While theabsolute value flow is less than ideal in some cases VIC modelcaptures the timing of peak flow quite well Nash-Sutcliffe (NS)statistic is also provided to show the strength of correlation

33 Changes in SWE for each GCM model outputs

The changes in SWE across 12 of the 13 models are givenin Fig 8 where in each panel deviations from the baselineconditions are depicted as a relative change in months defin-ing the cold season Note that even though the month ofApril is not generally considered part of the cold season inmany northern hemisphere studies it was included in thecurrent study due to its importance for snow ablation andmelting Also shown with text in the figure are the largestabsolute changes in SWE (in mm) for the month in questionThese absolute values reflect the averaged largest single daychanges in SWE in each month The direction of and mag-nitude of these absolute changes must be interpreted in thecontext of relative changes shown in the y-axis

Overall the following patterns emerge from Fig 8 Firstalmost all models predict a decline in SWE in the DecemberndashJanuary period However there are fewer models predictinga decline in SWE in the MarchndashApril period indicating thatthe reliability in predicting a decline in SWE is reduced whenmoving from the DecemberndashJanuary period to the MarchndashApril period Second there is considerable variability acrossthe models For example the GFDL model predicts a strongnegative change in SWE across all months years and emis-sion scenarios while the CNNM model predicts a weak neg-ative response in SWE under a changing climate (less than40 percent) In some cases the models even predict a positiveresponse Across all models the predicted increase althoughrare occurs in the spring during the normal thawing periodin 10 out of 13 models This represents a shift in the timingof snow accumulation moving from the DecemberndashFebruaryperiod to the MarchndashApril period Another interesting obser-

vation concerning the increase in SWE with climate changeis that it occurs more often under the B1 than the A2 sce-nario and more frequently in the middle of the 21st century(2050s) than in the later period (2090s)

There are notable exceptions to these general trends Forinstance the NCAR model does not fit this description re-sults from this model suggest a steady increase in SWE avail-ability under the aggressive A2 scenario at the end of the 21stcentury (although this increase occurs more in April than inJanuary) Several other models including the IPSL and theMIROC models also show similar trends but to a lesser ex-tent However all of these exceptions tend to occur in thespring rather than during the cold season under changing cli-matic conditions

Absolute changes in SWE as depicted by the number be-low bars representing each month also tell a similar story butin different context First the absolute magnitude of SWEchanges is highly variable ranging from less than 20 to over230 mm When placed in context of available snow waterat the peak snow accumulation period (as much as 400 mm)these absolute changes reflect 10 to 60 percent of the avail-able snow water Second the largest changes (and the mostvariation across models) occur in April followed by MarchThe timing of these largest changes in the spring are in con-trast with the large relative changes that occur in the winterperiod Although these largest changes do not necessarilyreflect the largest relative changes they indicate the impor-tance of climate change for accumulation (winter) and decay(spring) of SWE

Given the model-derived variability in predicting climatevariables that are important for snow accumulation the ques-tion arises as to which models or strategies should be used todistill information of use to stake-holders One common ap-proach is simply to average all models This approach com-monly referred to as a multi-model ensemble dilutes modelsthat provide poor outcomes with those that do a good job insimulating a region (Lambert and Boer 2001 Pierce et al2009) Using this approach a new multi-model ensembleclimate forcing dataset was generated for input to the VICmodel By averaging the outcomes of all climate models theensemble results suggest that SWE could decline as muchas 40 percent under all emission scenarios all time periodsand all months considered in this study (Fig 9) Of greaterinterest however is the gradual decline in the reduction inSWE from December (largest decrease) to April (smallestdecrease) These findings suggest that the amount of snowthat help build SWE during the cold season could graduallylessen leaving little or no snowmelt generated flow in thespring As with individual models the absolute changes in-crease from winter to spring

Rather than average the negative or positive changes as inthe ensemble approach it is possible to explore the frequencywith which the models predict positive or negative changesby month In Fig 10 the number of times an increase or de-crease was predicted by a model is plotted where the length

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2798 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 8 Changes in basin-averaged SWE relative to baseline conditions (1961ndash1990) predicted by 12 GCMs for the cold season (DJFMA)Negative values indicate a reduction and positive values indicate an increase in SWE relative to baseline conditions Also shown are themaximum averaged absolute changes in SWE (in units of mmmonth) The sign of the absolute change must be interpreted with the directionof relative change on the y-axis Note that only 12 out of 13 models used in the study are shown

of each bar is equal to 13 representing the total number ofmodels used The first finding from this plot is that all modelsindicate a clear decline in December under changing climateconditions Moreover many of the models predict a declinein SWE January through March Yet only half the modelspredict an increase while the other half predict a decrease inApril and this result is consistent across all emission scenar-ios and time periods This lack of consensus among modelsin April is mostly due to lack of snow in this period In otherwords the amount of reduction on an already small snowquantity in the spring is lost between model variability In

the winter period however all models do predict snow avail-ability but simply less of it under a changing climate

Finally Fig 11 shows the changes in modeled SWE acrosselevation zones (or bands) for four scenariotime combina-tions The elevation bands are given at 250 m incrementson the y-axis while the x-axis depicts the departure of SWEfrom the baseline conditions in relative terms The results re-veal that the largest changes (greater than 50 percent) occurin the lower elevation bands (under 500 m) and as altitudeincreases the reduction in SWE diminishes exponentially toa minimum value of about 10 percent Above 2000 m the

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2799

Fig 8 Continued

predicted decline in SWE is constant with a mean value of10 percent However slight positive bias in VIC-simulatedsnow area and SWE amounts in lower elevations suggeststhat the magnitude of change in lower elevations must beviewed with caution This is also corroborated by lack ofmodel consensus in April which affect lower elevations firstThese findings suggest that as the climate warms and pre-cipitation declines lower elevation zones could experience amore rapid reduction in snow accumulation than the higherelevation zones of the basin but these results are as good asthe quality of the model predictions in the lower areas

4 Discussion

In this research the effects of climate change on theamount of water stored in snowpack in the mountains ofthe Euphrates-Tigris river basin were assessed using the VICmacroscale hydrological model linked to climate model re-sults Inevitably the approach adopted in this study has sev-eral limitations With respect to the climate change predic-tions this study only considered the averages of the regionalclimate system but did not assess changes in its variabil-ity While the approach taken here involving the perturba-tion of observed climate allows for some random variationin day-to-day weather events the true variability includingclimatic extremes is not captured Work on the sensitivity of

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2795

Table 3 Various statistical metrics describing the strength of the relationship between observed and modeled SWE at one location (centeredat 3892 N and 4025 E) The description and the formulas used to derive the metrics are also provided

Statistic Value Description

N 138 number of samplesmod 932 mean modeled SWE (mm monthminus1)obs 1134 mean observed SWE (mm monthminus1)Smod 992 standard deviation of modeled SWE (mm monthminus1)Sobs 1048 standard deviation of observed SWE (mm monthminus1)

MAE 469 mean absolute error1N

nsumi=1

| modi minusobsi |

RMSE 737 root mean squared error[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSES 1064 systematic RMSE[Nminus1nsum

i=1( modi minusobsi)

2]05

RMSEU 989 unsystematic RMSE[Nminus1nsum

i=1( modi minus modi)

2]05

d 086 index of agreementd = 10minus

nsum

i=1( mod iminusobsi )2

nsumi=1

(∣∣ mod iminusobs

∣∣+∣∣obsiminusobs∣∣)2

especially under the B1 scenario in the mid 21st centuryWith the exception of one model (NCAR) no precipitationincrease is suggested under the A2 scenario in either the mid-dle or at the end of the 21st century In general precipitationchange is greater under A2 as much as 25 percent lower thanaverage baseline conditions within each modelrsquos outcomeIn contrast to other models the NCAR model predicts no netchange nor cooler temperatures (by as much as 05C) un-der the B1 scenario but instead predicts a shift to warmingunder the A2 scenario These results suggest that in generalthe climate in the E-T basin will be warmer and dryer withsubsequent implications for the snow processes and availablemelt water

32 Performance of the VIC model

The first set of results illustrates the ability of the VIC modelto accurately simulate SWE in the E-T basin The compari-son between modeled and observed SWE at a single site usedto acquire ground SWE observations shows a strong correla-tion over time (Fig 4) Detailed statistics for this compari-son are given in Table 3 Please see the discussion providedabove for justification of using certain statistical measure-ment variables as suggested by Willmott (1981) When ob-served and modeled SWE data are compared in the form ofa scatter plot (not shown) there is no systematic bias at thesite level as shown in Table 3

A comparison of modeled SWE data to remotely sensedobservations also suggest that the VIC model is able to cor-rectly simulate the available water in accumulated snow Fig-ure 5 displays the comparison between SSMI-observed andVIC-modeled SWE (in mm) in the form of a time-series plot

Fig 3 Cold season (DJFMA) averaged changes in near surfaceair temperature and precipitation from baseline conditions (1961ndash1990) predicted by 13 models for four scenariotime combina-tions Different colors correspond to different models while markershapes indicates the scenariotime pair Positive values in the tem-perature field indicate increases in temperature while negative val-ues in the precipitation field indicate relative decreases in precipita-tion amounts as a result of climate change

between 2001 and 2006 averaged over 8-day intervals Ingeneral there is very good agreement between the two es-timates (r = 083) With a few exceptions the VIC modelaccurately simulates the seasonal and interannual variationof SWE in this part of the basin There is remarkable

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2796 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 4 Comparison of single day SWE observations to modeledquantities at a location in northern part of the E-T basin The siteis located at (3892 N and 4025 E) and includes once-per-monthobservations Sample dates of observed data are provided for refer-ence

Fig 5 A comparison of modeled (red) and observed (black) SWEat a location in southeastern Turkey (405 E longitude 385 N lat-itude) within the Euphrates-Tigris basin for 2001ndash2006 The ob-served data are extracted from the 8-day blended Special SensorMicrowave Imager (SSMI) data (Brodzik et al 2007)

correspondence between the VIC model and the SSMI ob-servations for the timing of accumulation maximum and ab-lation of snow The figure also points to interesting dynam-ics of SWE in the basin First snow accumulation begins inNovember of each year gradually increases to a maximumin late Marchndashearly April with a subsequent sharp declineinto the summer season Another important observation from

45

Figure 6 Comparison of observed (top) and simulated (bottom) snow covered areas in the Euphrates-Tigris basin across

different periods in winter 2001 Snow cover is indicated in white against the color-coded topography of the basin

Observations in the top row were extracted from Terra MODIS 8-day snow cover product (MOD10C2) at 005 degree

spatial resolution while the simulated results of the bottom row were produced in this research using the VIC model

More details on the MODIS snow cover product can be found in Hall et al (2002) Each date at the top indicate the

beginning date of the 8-day compositing period

January252001January172001January92001 February22001

MODIS

VIC

Fig 6 Comparison of observed (top) and simulated (bottom) snowcovered areas in the Euphrates-Tigris basin across different periodsin winter 2001 Snow cover is indicated in white against the color-coded topography of the basin Observations in the top row were ex-tracted from Terra MODIS 8-day snow cover product (MOD10C2)at 005 spatial resolution while the simulated results of the bot-tom row were produced in this research using the VIC model Moredetails on the MODIS snow cover product can be found in Hall etal (2002) Each date at the top indicate the beginning date of the8-day compositing period

Fig 4 is the interannual variability of SWE Both the mod-eled and remote sensing results had larger maximum SWEvalues (around 200 mm) in years 2001 2002 and 2006 hadthan 2004 and 2005 Thus while 2002 and 2003 are largesnow accumulation years (and therefore larger SWE) 2004and 2005 appear to be drier years with much less accumula-tion It is interesting to note that both the VIC model and thesatellite observations capture this variability quite well

Further assessment of the VIC model performance wasmade by visually comparing modeled snow covered areas tosatellite observations of snow cover from the MODIS sen-sor (Fig 6) The map comparison between satellite-observedand modeled snow area data for four dates in January 2001indicate a noticeable agreement Even though VIC predicteda larger area covered by snow on each date especially in theeastern mountainous zone of the basin the general patternof snow cover accumulating in the arc-shaped northern andeastern portions of the basin is remarkably similar betweenthe two data sets One reason for the larger area snow accu-mulation in VIC simulations is the coarse nature of the grid-ded weather data that smooth over the fine scale topographicvariation present in the MODIS data

Comparison of modeled discharge to observed quantitiesat a location in the northeastern portion of the basin furthershows a notable correlation especially in the timing of peakflows that are indicative of timing of snowmelt (Fig 7) Themodeled discharge data were obtained by routing VIC sur-face and baseflow predictions through the routing model ofLohmann et al (1996) The quality of the match between theobserved and modeled discharge determined by the Nash-Sutcliffe coefficient of 05 is moderate although the timingof peak flows is matched nicely indicating that VIC is ableto simulate the accumulation and melting of snow in accept-able form

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2797

Fig 7 Comparison of monthly observed and modeled discharge atthe Logmar station (located at 3851 N and 3949 E) While theabsolute value flow is less than ideal in some cases VIC modelcaptures the timing of peak flow quite well Nash-Sutcliffe (NS)statistic is also provided to show the strength of correlation

33 Changes in SWE for each GCM model outputs

The changes in SWE across 12 of the 13 models are givenin Fig 8 where in each panel deviations from the baselineconditions are depicted as a relative change in months defin-ing the cold season Note that even though the month ofApril is not generally considered part of the cold season inmany northern hemisphere studies it was included in thecurrent study due to its importance for snow ablation andmelting Also shown with text in the figure are the largestabsolute changes in SWE (in mm) for the month in questionThese absolute values reflect the averaged largest single daychanges in SWE in each month The direction of and mag-nitude of these absolute changes must be interpreted in thecontext of relative changes shown in the y-axis

Overall the following patterns emerge from Fig 8 Firstalmost all models predict a decline in SWE in the DecemberndashJanuary period However there are fewer models predictinga decline in SWE in the MarchndashApril period indicating thatthe reliability in predicting a decline in SWE is reduced whenmoving from the DecemberndashJanuary period to the MarchndashApril period Second there is considerable variability acrossthe models For example the GFDL model predicts a strongnegative change in SWE across all months years and emis-sion scenarios while the CNNM model predicts a weak neg-ative response in SWE under a changing climate (less than40 percent) In some cases the models even predict a positiveresponse Across all models the predicted increase althoughrare occurs in the spring during the normal thawing periodin 10 out of 13 models This represents a shift in the timingof snow accumulation moving from the DecemberndashFebruaryperiod to the MarchndashApril period Another interesting obser-

vation concerning the increase in SWE with climate changeis that it occurs more often under the B1 than the A2 sce-nario and more frequently in the middle of the 21st century(2050s) than in the later period (2090s)

There are notable exceptions to these general trends Forinstance the NCAR model does not fit this description re-sults from this model suggest a steady increase in SWE avail-ability under the aggressive A2 scenario at the end of the 21stcentury (although this increase occurs more in April than inJanuary) Several other models including the IPSL and theMIROC models also show similar trends but to a lesser ex-tent However all of these exceptions tend to occur in thespring rather than during the cold season under changing cli-matic conditions

Absolute changes in SWE as depicted by the number be-low bars representing each month also tell a similar story butin different context First the absolute magnitude of SWEchanges is highly variable ranging from less than 20 to over230 mm When placed in context of available snow waterat the peak snow accumulation period (as much as 400 mm)these absolute changes reflect 10 to 60 percent of the avail-able snow water Second the largest changes (and the mostvariation across models) occur in April followed by MarchThe timing of these largest changes in the spring are in con-trast with the large relative changes that occur in the winterperiod Although these largest changes do not necessarilyreflect the largest relative changes they indicate the impor-tance of climate change for accumulation (winter) and decay(spring) of SWE

Given the model-derived variability in predicting climatevariables that are important for snow accumulation the ques-tion arises as to which models or strategies should be used todistill information of use to stake-holders One common ap-proach is simply to average all models This approach com-monly referred to as a multi-model ensemble dilutes modelsthat provide poor outcomes with those that do a good job insimulating a region (Lambert and Boer 2001 Pierce et al2009) Using this approach a new multi-model ensembleclimate forcing dataset was generated for input to the VICmodel By averaging the outcomes of all climate models theensemble results suggest that SWE could decline as muchas 40 percent under all emission scenarios all time periodsand all months considered in this study (Fig 9) Of greaterinterest however is the gradual decline in the reduction inSWE from December (largest decrease) to April (smallestdecrease) These findings suggest that the amount of snowthat help build SWE during the cold season could graduallylessen leaving little or no snowmelt generated flow in thespring As with individual models the absolute changes in-crease from winter to spring

Rather than average the negative or positive changes as inthe ensemble approach it is possible to explore the frequencywith which the models predict positive or negative changesby month In Fig 10 the number of times an increase or de-crease was predicted by a model is plotted where the length

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2798 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 8 Changes in basin-averaged SWE relative to baseline conditions (1961ndash1990) predicted by 12 GCMs for the cold season (DJFMA)Negative values indicate a reduction and positive values indicate an increase in SWE relative to baseline conditions Also shown are themaximum averaged absolute changes in SWE (in units of mmmonth) The sign of the absolute change must be interpreted with the directionof relative change on the y-axis Note that only 12 out of 13 models used in the study are shown

of each bar is equal to 13 representing the total number ofmodels used The first finding from this plot is that all modelsindicate a clear decline in December under changing climateconditions Moreover many of the models predict a declinein SWE January through March Yet only half the modelspredict an increase while the other half predict a decrease inApril and this result is consistent across all emission scenar-ios and time periods This lack of consensus among modelsin April is mostly due to lack of snow in this period In otherwords the amount of reduction on an already small snowquantity in the spring is lost between model variability In

the winter period however all models do predict snow avail-ability but simply less of it under a changing climate

Finally Fig 11 shows the changes in modeled SWE acrosselevation zones (or bands) for four scenariotime combina-tions The elevation bands are given at 250 m incrementson the y-axis while the x-axis depicts the departure of SWEfrom the baseline conditions in relative terms The results re-veal that the largest changes (greater than 50 percent) occurin the lower elevation bands (under 500 m) and as altitudeincreases the reduction in SWE diminishes exponentially toa minimum value of about 10 percent Above 2000 m the

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2799

Fig 8 Continued

predicted decline in SWE is constant with a mean value of10 percent However slight positive bias in VIC-simulatedsnow area and SWE amounts in lower elevations suggeststhat the magnitude of change in lower elevations must beviewed with caution This is also corroborated by lack ofmodel consensus in April which affect lower elevations firstThese findings suggest that as the climate warms and pre-cipitation declines lower elevation zones could experience amore rapid reduction in snow accumulation than the higherelevation zones of the basin but these results are as good asthe quality of the model predictions in the lower areas

4 Discussion

In this research the effects of climate change on theamount of water stored in snowpack in the mountains ofthe Euphrates-Tigris river basin were assessed using the VICmacroscale hydrological model linked to climate model re-sults Inevitably the approach adopted in this study has sev-eral limitations With respect to the climate change predic-tions this study only considered the averages of the regionalclimate system but did not assess changes in its variabil-ity While the approach taken here involving the perturba-tion of observed climate allows for some random variationin day-to-day weather events the true variability includingclimatic extremes is not captured Work on the sensitivity of

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

2796 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 4 Comparison of single day SWE observations to modeledquantities at a location in northern part of the E-T basin The siteis located at (3892 N and 4025 E) and includes once-per-monthobservations Sample dates of observed data are provided for refer-ence

Fig 5 A comparison of modeled (red) and observed (black) SWEat a location in southeastern Turkey (405 E longitude 385 N lat-itude) within the Euphrates-Tigris basin for 2001ndash2006 The ob-served data are extracted from the 8-day blended Special SensorMicrowave Imager (SSMI) data (Brodzik et al 2007)

correspondence between the VIC model and the SSMI ob-servations for the timing of accumulation maximum and ab-lation of snow The figure also points to interesting dynam-ics of SWE in the basin First snow accumulation begins inNovember of each year gradually increases to a maximumin late Marchndashearly April with a subsequent sharp declineinto the summer season Another important observation from

45

Figure 6 Comparison of observed (top) and simulated (bottom) snow covered areas in the Euphrates-Tigris basin across

different periods in winter 2001 Snow cover is indicated in white against the color-coded topography of the basin

Observations in the top row were extracted from Terra MODIS 8-day snow cover product (MOD10C2) at 005 degree

spatial resolution while the simulated results of the bottom row were produced in this research using the VIC model

More details on the MODIS snow cover product can be found in Hall et al (2002) Each date at the top indicate the

beginning date of the 8-day compositing period

January252001January172001January92001 February22001

MODIS

VIC

Fig 6 Comparison of observed (top) and simulated (bottom) snowcovered areas in the Euphrates-Tigris basin across different periodsin winter 2001 Snow cover is indicated in white against the color-coded topography of the basin Observations in the top row were ex-tracted from Terra MODIS 8-day snow cover product (MOD10C2)at 005 spatial resolution while the simulated results of the bot-tom row were produced in this research using the VIC model Moredetails on the MODIS snow cover product can be found in Hall etal (2002) Each date at the top indicate the beginning date of the8-day compositing period

Fig 4 is the interannual variability of SWE Both the mod-eled and remote sensing results had larger maximum SWEvalues (around 200 mm) in years 2001 2002 and 2006 hadthan 2004 and 2005 Thus while 2002 and 2003 are largesnow accumulation years (and therefore larger SWE) 2004and 2005 appear to be drier years with much less accumula-tion It is interesting to note that both the VIC model and thesatellite observations capture this variability quite well

Further assessment of the VIC model performance wasmade by visually comparing modeled snow covered areas tosatellite observations of snow cover from the MODIS sen-sor (Fig 6) The map comparison between satellite-observedand modeled snow area data for four dates in January 2001indicate a noticeable agreement Even though VIC predicteda larger area covered by snow on each date especially in theeastern mountainous zone of the basin the general patternof snow cover accumulating in the arc-shaped northern andeastern portions of the basin is remarkably similar betweenthe two data sets One reason for the larger area snow accu-mulation in VIC simulations is the coarse nature of the grid-ded weather data that smooth over the fine scale topographicvariation present in the MODIS data

Comparison of modeled discharge to observed quantitiesat a location in the northeastern portion of the basin furthershows a notable correlation especially in the timing of peakflows that are indicative of timing of snowmelt (Fig 7) Themodeled discharge data were obtained by routing VIC sur-face and baseflow predictions through the routing model ofLohmann et al (1996) The quality of the match between theobserved and modeled discharge determined by the Nash-Sutcliffe coefficient of 05 is moderate although the timingof peak flows is matched nicely indicating that VIC is ableto simulate the accumulation and melting of snow in accept-able form

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2797

Fig 7 Comparison of monthly observed and modeled discharge atthe Logmar station (located at 3851 N and 3949 E) While theabsolute value flow is less than ideal in some cases VIC modelcaptures the timing of peak flow quite well Nash-Sutcliffe (NS)statistic is also provided to show the strength of correlation

33 Changes in SWE for each GCM model outputs

The changes in SWE across 12 of the 13 models are givenin Fig 8 where in each panel deviations from the baselineconditions are depicted as a relative change in months defin-ing the cold season Note that even though the month ofApril is not generally considered part of the cold season inmany northern hemisphere studies it was included in thecurrent study due to its importance for snow ablation andmelting Also shown with text in the figure are the largestabsolute changes in SWE (in mm) for the month in questionThese absolute values reflect the averaged largest single daychanges in SWE in each month The direction of and mag-nitude of these absolute changes must be interpreted in thecontext of relative changes shown in the y-axis

Overall the following patterns emerge from Fig 8 Firstalmost all models predict a decline in SWE in the DecemberndashJanuary period However there are fewer models predictinga decline in SWE in the MarchndashApril period indicating thatthe reliability in predicting a decline in SWE is reduced whenmoving from the DecemberndashJanuary period to the MarchndashApril period Second there is considerable variability acrossthe models For example the GFDL model predicts a strongnegative change in SWE across all months years and emis-sion scenarios while the CNNM model predicts a weak neg-ative response in SWE under a changing climate (less than40 percent) In some cases the models even predict a positiveresponse Across all models the predicted increase althoughrare occurs in the spring during the normal thawing periodin 10 out of 13 models This represents a shift in the timingof snow accumulation moving from the DecemberndashFebruaryperiod to the MarchndashApril period Another interesting obser-

vation concerning the increase in SWE with climate changeis that it occurs more often under the B1 than the A2 sce-nario and more frequently in the middle of the 21st century(2050s) than in the later period (2090s)

There are notable exceptions to these general trends Forinstance the NCAR model does not fit this description re-sults from this model suggest a steady increase in SWE avail-ability under the aggressive A2 scenario at the end of the 21stcentury (although this increase occurs more in April than inJanuary) Several other models including the IPSL and theMIROC models also show similar trends but to a lesser ex-tent However all of these exceptions tend to occur in thespring rather than during the cold season under changing cli-matic conditions

Absolute changes in SWE as depicted by the number be-low bars representing each month also tell a similar story butin different context First the absolute magnitude of SWEchanges is highly variable ranging from less than 20 to over230 mm When placed in context of available snow waterat the peak snow accumulation period (as much as 400 mm)these absolute changes reflect 10 to 60 percent of the avail-able snow water Second the largest changes (and the mostvariation across models) occur in April followed by MarchThe timing of these largest changes in the spring are in con-trast with the large relative changes that occur in the winterperiod Although these largest changes do not necessarilyreflect the largest relative changes they indicate the impor-tance of climate change for accumulation (winter) and decay(spring) of SWE

Given the model-derived variability in predicting climatevariables that are important for snow accumulation the ques-tion arises as to which models or strategies should be used todistill information of use to stake-holders One common ap-proach is simply to average all models This approach com-monly referred to as a multi-model ensemble dilutes modelsthat provide poor outcomes with those that do a good job insimulating a region (Lambert and Boer 2001 Pierce et al2009) Using this approach a new multi-model ensembleclimate forcing dataset was generated for input to the VICmodel By averaging the outcomes of all climate models theensemble results suggest that SWE could decline as muchas 40 percent under all emission scenarios all time periodsand all months considered in this study (Fig 9) Of greaterinterest however is the gradual decline in the reduction inSWE from December (largest decrease) to April (smallestdecrease) These findings suggest that the amount of snowthat help build SWE during the cold season could graduallylessen leaving little or no snowmelt generated flow in thespring As with individual models the absolute changes in-crease from winter to spring

Rather than average the negative or positive changes as inthe ensemble approach it is possible to explore the frequencywith which the models predict positive or negative changesby month In Fig 10 the number of times an increase or de-crease was predicted by a model is plotted where the length

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2798 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 8 Changes in basin-averaged SWE relative to baseline conditions (1961ndash1990) predicted by 12 GCMs for the cold season (DJFMA)Negative values indicate a reduction and positive values indicate an increase in SWE relative to baseline conditions Also shown are themaximum averaged absolute changes in SWE (in units of mmmonth) The sign of the absolute change must be interpreted with the directionof relative change on the y-axis Note that only 12 out of 13 models used in the study are shown

of each bar is equal to 13 representing the total number ofmodels used The first finding from this plot is that all modelsindicate a clear decline in December under changing climateconditions Moreover many of the models predict a declinein SWE January through March Yet only half the modelspredict an increase while the other half predict a decrease inApril and this result is consistent across all emission scenar-ios and time periods This lack of consensus among modelsin April is mostly due to lack of snow in this period In otherwords the amount of reduction on an already small snowquantity in the spring is lost between model variability In

the winter period however all models do predict snow avail-ability but simply less of it under a changing climate

Finally Fig 11 shows the changes in modeled SWE acrosselevation zones (or bands) for four scenariotime combina-tions The elevation bands are given at 250 m incrementson the y-axis while the x-axis depicts the departure of SWEfrom the baseline conditions in relative terms The results re-veal that the largest changes (greater than 50 percent) occurin the lower elevation bands (under 500 m) and as altitudeincreases the reduction in SWE diminishes exponentially toa minimum value of about 10 percent Above 2000 m the

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2799

Fig 8 Continued

predicted decline in SWE is constant with a mean value of10 percent However slight positive bias in VIC-simulatedsnow area and SWE amounts in lower elevations suggeststhat the magnitude of change in lower elevations must beviewed with caution This is also corroborated by lack ofmodel consensus in April which affect lower elevations firstThese findings suggest that as the climate warms and pre-cipitation declines lower elevation zones could experience amore rapid reduction in snow accumulation than the higherelevation zones of the basin but these results are as good asthe quality of the model predictions in the lower areas

4 Discussion

In this research the effects of climate change on theamount of water stored in snowpack in the mountains ofthe Euphrates-Tigris river basin were assessed using the VICmacroscale hydrological model linked to climate model re-sults Inevitably the approach adopted in this study has sev-eral limitations With respect to the climate change predic-tions this study only considered the averages of the regionalclimate system but did not assess changes in its variabil-ity While the approach taken here involving the perturba-tion of observed climate allows for some random variationin day-to-day weather events the true variability includingclimatic extremes is not captured Work on the sensitivity of

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2797

Fig 7 Comparison of monthly observed and modeled discharge atthe Logmar station (located at 3851 N and 3949 E) While theabsolute value flow is less than ideal in some cases VIC modelcaptures the timing of peak flow quite well Nash-Sutcliffe (NS)statistic is also provided to show the strength of correlation

33 Changes in SWE for each GCM model outputs

The changes in SWE across 12 of the 13 models are givenin Fig 8 where in each panel deviations from the baselineconditions are depicted as a relative change in months defin-ing the cold season Note that even though the month ofApril is not generally considered part of the cold season inmany northern hemisphere studies it was included in thecurrent study due to its importance for snow ablation andmelting Also shown with text in the figure are the largestabsolute changes in SWE (in mm) for the month in questionThese absolute values reflect the averaged largest single daychanges in SWE in each month The direction of and mag-nitude of these absolute changes must be interpreted in thecontext of relative changes shown in the y-axis

Overall the following patterns emerge from Fig 8 Firstalmost all models predict a decline in SWE in the DecemberndashJanuary period However there are fewer models predictinga decline in SWE in the MarchndashApril period indicating thatthe reliability in predicting a decline in SWE is reduced whenmoving from the DecemberndashJanuary period to the MarchndashApril period Second there is considerable variability acrossthe models For example the GFDL model predicts a strongnegative change in SWE across all months years and emis-sion scenarios while the CNNM model predicts a weak neg-ative response in SWE under a changing climate (less than40 percent) In some cases the models even predict a positiveresponse Across all models the predicted increase althoughrare occurs in the spring during the normal thawing periodin 10 out of 13 models This represents a shift in the timingof snow accumulation moving from the DecemberndashFebruaryperiod to the MarchndashApril period Another interesting obser-

vation concerning the increase in SWE with climate changeis that it occurs more often under the B1 than the A2 sce-nario and more frequently in the middle of the 21st century(2050s) than in the later period (2090s)

There are notable exceptions to these general trends Forinstance the NCAR model does not fit this description re-sults from this model suggest a steady increase in SWE avail-ability under the aggressive A2 scenario at the end of the 21stcentury (although this increase occurs more in April than inJanuary) Several other models including the IPSL and theMIROC models also show similar trends but to a lesser ex-tent However all of these exceptions tend to occur in thespring rather than during the cold season under changing cli-matic conditions

Absolute changes in SWE as depicted by the number be-low bars representing each month also tell a similar story butin different context First the absolute magnitude of SWEchanges is highly variable ranging from less than 20 to over230 mm When placed in context of available snow waterat the peak snow accumulation period (as much as 400 mm)these absolute changes reflect 10 to 60 percent of the avail-able snow water Second the largest changes (and the mostvariation across models) occur in April followed by MarchThe timing of these largest changes in the spring are in con-trast with the large relative changes that occur in the winterperiod Although these largest changes do not necessarilyreflect the largest relative changes they indicate the impor-tance of climate change for accumulation (winter) and decay(spring) of SWE

Given the model-derived variability in predicting climatevariables that are important for snow accumulation the ques-tion arises as to which models or strategies should be used todistill information of use to stake-holders One common ap-proach is simply to average all models This approach com-monly referred to as a multi-model ensemble dilutes modelsthat provide poor outcomes with those that do a good job insimulating a region (Lambert and Boer 2001 Pierce et al2009) Using this approach a new multi-model ensembleclimate forcing dataset was generated for input to the VICmodel By averaging the outcomes of all climate models theensemble results suggest that SWE could decline as muchas 40 percent under all emission scenarios all time periodsand all months considered in this study (Fig 9) Of greaterinterest however is the gradual decline in the reduction inSWE from December (largest decrease) to April (smallestdecrease) These findings suggest that the amount of snowthat help build SWE during the cold season could graduallylessen leaving little or no snowmelt generated flow in thespring As with individual models the absolute changes in-crease from winter to spring

Rather than average the negative or positive changes as inthe ensemble approach it is possible to explore the frequencywith which the models predict positive or negative changesby month In Fig 10 the number of times an increase or de-crease was predicted by a model is plotted where the length

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2798 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 8 Changes in basin-averaged SWE relative to baseline conditions (1961ndash1990) predicted by 12 GCMs for the cold season (DJFMA)Negative values indicate a reduction and positive values indicate an increase in SWE relative to baseline conditions Also shown are themaximum averaged absolute changes in SWE (in units of mmmonth) The sign of the absolute change must be interpreted with the directionof relative change on the y-axis Note that only 12 out of 13 models used in the study are shown

of each bar is equal to 13 representing the total number ofmodels used The first finding from this plot is that all modelsindicate a clear decline in December under changing climateconditions Moreover many of the models predict a declinein SWE January through March Yet only half the modelspredict an increase while the other half predict a decrease inApril and this result is consistent across all emission scenar-ios and time periods This lack of consensus among modelsin April is mostly due to lack of snow in this period In otherwords the amount of reduction on an already small snowquantity in the spring is lost between model variability In

the winter period however all models do predict snow avail-ability but simply less of it under a changing climate

Finally Fig 11 shows the changes in modeled SWE acrosselevation zones (or bands) for four scenariotime combina-tions The elevation bands are given at 250 m incrementson the y-axis while the x-axis depicts the departure of SWEfrom the baseline conditions in relative terms The results re-veal that the largest changes (greater than 50 percent) occurin the lower elevation bands (under 500 m) and as altitudeincreases the reduction in SWE diminishes exponentially toa minimum value of about 10 percent Above 2000 m the

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2799

Fig 8 Continued

predicted decline in SWE is constant with a mean value of10 percent However slight positive bias in VIC-simulatedsnow area and SWE amounts in lower elevations suggeststhat the magnitude of change in lower elevations must beviewed with caution This is also corroborated by lack ofmodel consensus in April which affect lower elevations firstThese findings suggest that as the climate warms and pre-cipitation declines lower elevation zones could experience amore rapid reduction in snow accumulation than the higherelevation zones of the basin but these results are as good asthe quality of the model predictions in the lower areas

4 Discussion

In this research the effects of climate change on theamount of water stored in snowpack in the mountains ofthe Euphrates-Tigris river basin were assessed using the VICmacroscale hydrological model linked to climate model re-sults Inevitably the approach adopted in this study has sev-eral limitations With respect to the climate change predic-tions this study only considered the averages of the regionalclimate system but did not assess changes in its variabil-ity While the approach taken here involving the perturba-tion of observed climate allows for some random variationin day-to-day weather events the true variability includingclimatic extremes is not captured Work on the sensitivity of

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

2798 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 8 Changes in basin-averaged SWE relative to baseline conditions (1961ndash1990) predicted by 12 GCMs for the cold season (DJFMA)Negative values indicate a reduction and positive values indicate an increase in SWE relative to baseline conditions Also shown are themaximum averaged absolute changes in SWE (in units of mmmonth) The sign of the absolute change must be interpreted with the directionof relative change on the y-axis Note that only 12 out of 13 models used in the study are shown

of each bar is equal to 13 representing the total number ofmodels used The first finding from this plot is that all modelsindicate a clear decline in December under changing climateconditions Moreover many of the models predict a declinein SWE January through March Yet only half the modelspredict an increase while the other half predict a decrease inApril and this result is consistent across all emission scenar-ios and time periods This lack of consensus among modelsin April is mostly due to lack of snow in this period In otherwords the amount of reduction on an already small snowquantity in the spring is lost between model variability In

the winter period however all models do predict snow avail-ability but simply less of it under a changing climate

Finally Fig 11 shows the changes in modeled SWE acrosselevation zones (or bands) for four scenariotime combina-tions The elevation bands are given at 250 m incrementson the y-axis while the x-axis depicts the departure of SWEfrom the baseline conditions in relative terms The results re-veal that the largest changes (greater than 50 percent) occurin the lower elevation bands (under 500 m) and as altitudeincreases the reduction in SWE diminishes exponentially toa minimum value of about 10 percent Above 2000 m the

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2799

Fig 8 Continued

predicted decline in SWE is constant with a mean value of10 percent However slight positive bias in VIC-simulatedsnow area and SWE amounts in lower elevations suggeststhat the magnitude of change in lower elevations must beviewed with caution This is also corroborated by lack ofmodel consensus in April which affect lower elevations firstThese findings suggest that as the climate warms and pre-cipitation declines lower elevation zones could experience amore rapid reduction in snow accumulation than the higherelevation zones of the basin but these results are as good asthe quality of the model predictions in the lower areas

4 Discussion

In this research the effects of climate change on theamount of water stored in snowpack in the mountains ofthe Euphrates-Tigris river basin were assessed using the VICmacroscale hydrological model linked to climate model re-sults Inevitably the approach adopted in this study has sev-eral limitations With respect to the climate change predic-tions this study only considered the averages of the regionalclimate system but did not assess changes in its variabil-ity While the approach taken here involving the perturba-tion of observed climate allows for some random variationin day-to-day weather events the true variability includingclimatic extremes is not captured Work on the sensitivity of

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2799

Fig 8 Continued

predicted decline in SWE is constant with a mean value of10 percent However slight positive bias in VIC-simulatedsnow area and SWE amounts in lower elevations suggeststhat the magnitude of change in lower elevations must beviewed with caution This is also corroborated by lack ofmodel consensus in April which affect lower elevations firstThese findings suggest that as the climate warms and pre-cipitation declines lower elevation zones could experience amore rapid reduction in snow accumulation than the higherelevation zones of the basin but these results are as good asthe quality of the model predictions in the lower areas

4 Discussion

In this research the effects of climate change on theamount of water stored in snowpack in the mountains ofthe Euphrates-Tigris river basin were assessed using the VICmacroscale hydrological model linked to climate model re-sults Inevitably the approach adopted in this study has sev-eral limitations With respect to the climate change predic-tions this study only considered the averages of the regionalclimate system but did not assess changes in its variabil-ity While the approach taken here involving the perturba-tion of observed climate allows for some random variationin day-to-day weather events the true variability includingclimatic extremes is not captured Work on the sensitivity of

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

2800 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

Fig 9 Changes in basin-averaged SWE relative to baseline condi-tions (1961ndash1990) predicted by multi-model ensemble outputs forthe cold season (DJFMA) Negative values indicate a reduction andpositive values indicates an increase in SWE relative to baselineconditions

Fig 10 Number of models that predicted an increase (the upperhalf) and a decrease (the lower half) in SWE relative to baselineconditions for each monthscenarioperiod combination In gen-eral more models predict a decrease in SWE especially in theDecember-March period when snow accumulation is greatest

snowpack to changes in the frequency and magnitude of ex-treme weather events suggests long term droughts and coldspells could have important consequences for winter waterstorage (see eg Mote 2006) Thus the ldquoaveragerdquo pre-dicted changes in future snowpack presented in this studymay under- or over-estimate future water yields from snowmelt in the basin In addition studies such as this could fur-ther benefit from probabilistic modeling of uncertainty asso-ciated with climate

Fig 11 Changes in modeled SWE across elevation zones for fourscenariotime combinations Each marker plot represents the aver-age VIC SWE outcome from the 13 GCM climate forcing data

Another climate model-related limitation of this study isthe application of GCM-predicted anomalies to observedtemperature and precipitation While it is possible to down-scale large GCM grids statistically or dynamically so thatthe spatial variability across a study area can be examinedthis was not done in this study There is evidence to sug-gest that it is preferable to apply the native GCM-predictedclimate anomalies to observed data in order to construct cli-mate scenarios rather than derive them directly from the re-gional climate model simulations or by statistical downscal-ing (eg Arnell et al 2003 Salathe et al 2007)

The comparison of future climatic conditions from 13GCMs provides several important observations First thereis significant variation in future conditions across the dif-ferent models even though all models were forced withthe same emissions scenarios Second the models responddifferently to emission scenarios in different time periodsThird the model results suggest that variability in precipi-tation estimates is much greater than the variability of esti-mates for temperature For example the NCAR model pre-dicts a 20 percent increase in 2050 while the CSIRO modelsuggests a greater than 40 percent decrease in rainfall in 2050under the same emissions scenario Temperature on theother hand has an expected pattern of change across modelsgenerally increasing into the 21st century across successiveperiods but with variability in the amount of increase Forexample the variation in temperature increase from less than1C with the MIROC3 model to around 3C with the IPSLmodel This variation across models is one reason for usingthe multi-model ensemble approach adopted here This typeof multi-model ensemble approach is already in use in globalstudies that examine the mean state of a given climate and hasalso been found to be superior to any individual model output

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2801

in studies at a regional scale (Pierce et al 2009) Researchsuggests that this occurs as a result of errors in individualmodels cancelling out one another In the fourth assessmentreport IPCC points out discrepancies among models andsuggests the use of multiple models for any impact assess-ment (IPCC 2007b) In the present study both the individualmodel and the multi-model ensemble average approach wereutilized unlike previous studies that investigated the effectsof climate change on water yield in the E-T basin

Another limitation of this study is concerned with lackof subgrid variability in snow processes in the VIC modelThere is evidence to suggest that from a hydrological per-spective the subgrid snow-depth distribution is an impor-tant quantity to include within macro scale models (eg Le-ung and Qian 2003 Liston 2004) This is because sub-grid snow distribution is the critical first-order influence onsnow-cover depletion during snowmelt since the spatially-variable subgrid SWE distribution is largely responsible forthe patchy mosaic of snow covered areas In the current re-search the within-grid variability of snow covered area andassociated SWE is represented as elevation bands only andmay be under- or over-estimated In addition to elevationthe true subgrid snow distribution is also strongly influencedby spatial variations in snowndashcanopy interactions snow re-distribution by wind and orographic precipitation (Liston2004)

The results presented in this research have important im-plications for the future of water availability in the regionthey may perhaps pave the way for adaptation options Asignificant decline in water stored in snowpack is likely todepress overall production volumes and the timing of peakflow thus affecting reservoir storage for power generationand warm season irrigation Currently there are more than20 dams and reservoirs with modest to large storage capac-ities in the basin The expected snowpack reduction alongwith the shift of runoff peak from spring to winter due to cli-mate warming is likely to have important effects on powergeneration and revenues in Turkey and Iraq Since all of theexisting capacity was designed under contemporary climateconditions to take advantage of snowpack as a natural reser-voir the adaptability of the storage structures is in questionwhen climate warming is considered Expansion of storagecapacity and to a lesser extent generation capacity may in-crease water availability although such expansions might behugely cost prohibitive As a result countermeasures for wa-ter shortages could become much more difficult

In addition to its impacts on water management climatechange introduces another element of uncertainty about fu-ture water resource management in the E-T basin Water re-sources of the region are heavily managed and all aspects ofwater usage are politically charged The allocation of wa-ter between the three countries of the E-T basin has led to atleast fifteen conflicts in the past four decades (Gleick 2009)While another conflict may not be likely in the short-term(Beaumont 1998) the future reduction of snow and its as-

sociated runoff may be enough for countries in the basin torisk conflict (Beaumont 1998) Therefore implementationof adaptation measures such as water conservation use ofmarkets to allocate water and the application of appropriatemanagement practices will play an important role in deter-mining the impacts of climate change on water resources

5 Conclusions

Water stored in seasonal snowpack in the crescent-shapedmountains of the Euphrates-Tigris basin is an extremely im-portant resource for the countries of the region Using acomprehensive assessment involving 13 general circulationmodels two greenhouse gas emission scenarios and twotime periods the results of this research suggest that cli-mate change has the potential to severely reduce (between10 to 60 percent) the amount of this water source and placeincreased stress on the water-dependant societies of the re-gion These findings are verified not only by individual GCMoutcomes but also by the outputs of a multi-model ensem-ble developed to reduce inter-model variation The resultsalso reveal that the largest changes (greater than 50 percent)in SWE occur in the lower elevation bands (under 500 m)and as altitude increases the reduction in SWE will dimin-ish exponentially suggesting a more rapid reduction in snowaccumulation in the lower zones of the basin although theseresults are less certain than basin averaged changes in snowwater availability While there are uncertainties associatedwith both the climate model outputs and outputs provided bythe VIC model the results of this investigation could haveimportant implications in the development of strategies to ad-dress the climate change problem in a region already fraughtwith water-related conflicts

AcknowledgementsThis research is partially funded by a NASAApplication Science Program grant number NNX08AM69Gawarded to the author Early edits and suggestions of George Allezsignificantly improved the readability of this document The authoralso thanks to Annemarie Schneider for her comments on an earlyversion of this document that made it more clear and succinct Thecomments of three anonymous reviewers substantially improvedthe content and the readability of this manuscript Finally theauthor acknowledges the modeling groups the Program forClimate Model Diagnosis and Intercomparison (PCMDI) and theWCRPrsquos Working Group on Coupled Modelling (WGCM) for theirroles in making the WCRP CMIP3 multi-model dataset availableSupport for that dataset was provided by the Office of Science USDepartment of Energy

Edited by J Liu

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

2802 MOzdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin

References

Beaumont P Restructuring of Water Usage in the Tigris-EuphratesBasin The Impact of Modern Water Management Projects inTransformation of Middle Eastern Natural Environments Lega-cies and Lessons edited by Coppock J and Miller J A YaleUniversity New Haven 1998

Beniston M Keller F and Goyette S Snowpack in the SwissAlps under changing climatic conditions an empirical approachfor climate impacts studies Theor Appl Climatol 74 19ndash312003

Brodzik M J Armstrong R L and Savoie M Global EASE-Grid 8-day Blended SSMI and MODIS Snow Cover (2001ndash2006) Boulder Colorado USA National Snow and Ice DataCenter Digital media 2007

Christensen N S Wood A W Voisin N Lettenmaier D P andPalmer R N Effects of climate change on the hydrology andwater resources of the Colorado River Basin Clim Change 62337ndash363 2004

Christensen N S and Lettenmaier D P A multimodel ensem-ble approach to assessment of climate change impacts on thehydrology and water resources of the Colorado River Basin Hy-drol Earth Syst Sci 11 1417ndash1434doi105194hess-11-1417-2007 2007

Cherkauer K A and Lettenmaier D P Simulation of spatialvariability in snow and frozen soil J Geophys Res 108(D22)8858doi1010292003JD003575 2003

Collins W D Blackmon M L Bonan G B Hack J J Hen-derson T B Kiehl J T Large W G McKenna D S BitzC M Bretherton C S Carton J A Chang P Doney S CSanter B D and Smith R D The Community Climate SystemModel Version 3 (CCSM3) J Climate 19 2122ndash2143 2006

Delworth T L Broccoli A J Rosati A Stouffer R J BalajiV Beesley J A Cooke W F Dixon K W Dunne J DunneK A Durachta J W Findell K L Ginoux P GnanadesikanA Gordon C T Griffies S M Gudgel R Harrison M JHeld I M Hemler R S Horowitz L W Klein S A Knut-son T R Kushner P J Langenhorst A R Lee H C Lin SJ Lu J Malyshev S L Milly P C D Ramaswamy V Rus-sell J Schwarzkopf M D Shevliakova E Sirutis J J Spel-man M J Stern W F Winton M Wittenberg A T WymanB Zeng F and Zhang R GFDLrsquos CM2 global coupled cli-mate models Part I Formulation and simulation characteristicsJ Climate 19 643ndash674 2006

Deque M Dreveton C Braun A and Cariolle D TheARPEGEIFS atmosphere model A contribution to the Frenchcommunity climate modeling Clim Dynam 10 249ndash2661994

Diansky N A and Volodin E M Simulation of present-dayclimate with a coupled Atmosphere-ocean general circulationmodel Izv Atmos Ocean Phys (Engl Transl) 38 732ndash7472002

EIEI Snow Observations (1966ndash2005) Hydrologic ResearchUnit General Directorate of Electrical Works Ankara Turkey491 pp January 2007

EIEI Monthly Flow Averages of Turkish Rivers (1935ndash2005) Hy-drologic Research Unit General Directorate of Electrical WorksAnkara Turkey 740 p September 2008

Evans J P 21st century climate change in the Middle East ClimChange 92 417ndash432doi101007s10584-008-9438-5 2009

Gleick P H Water The potential consequences of climate vari-ability and change for the water resources of the United StatesReport of the Water Sector Assessment Team of the National As-sessment of the Potential Consequences of Climate Variabilityand Change Pacific Institute for Studies in Development Envi-ronment and Security Oakland CA 151 pp 2000

Gleick P H Water and conflict Chronology in The Worldrsquos Water2008ndash2009 Island Press Washington D C 151ndash194 2009

Gordon C Cooper C Senior C A Banks H T Gregory J MJohns T C Mitchell J F B and Wood R A The simulationof SST sea ice extents and ocean heat transports in a versionof the Hadley Centre coupled model without flux adjustmentsClim Dynam 16 147ndash168 2000

Gordon H B Rotstayn L D McGregor J L Dix M R Kowal-czyk EA OrsquoFarrell S P Waterman L J Hirst A C Wil-son G Collier M A Watterson I G and Elliott T I TheCSIRO Mk3 Climate System Model CSIRO Atmospheric Re-search Technical Paper No 60 CSIRO Division of AtmosphericResearch Victoria Australia 130 pp 2002

Gruen G E Turkish waters Source of regional conflict or catalystfor peace Water Air Soil Poll 123 565ndash579 2000

Hall D K Riggs G A Salomonson V V DiGirolamo N Eand Bayr K J MODIS snow-cover products Remote Sens En-viron 83 181ndash194 2002

Hamlet A F and Lettenmaier D P Effects of Climate Change onHydrology and Water Resources in the Columbia River Basin JAm Water Resour As 35 1597ndash1623 1999

Hamlet A F Mote P W Clark M P and Lettenmaier D PEffects of temperature and precipitation variability on snowpacktrends in the western United States J Climate 18 4545ndash45612005

IPCC Special Report on Emission Scenarios edited by Nakicen-ovic N and Swart R Cambridge University Press UK 570 pp2000

IPCC Climate Change 2007 The Physical Science Basis Con-tribution of Working Group I to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change edited bySolomon S Qin D Manning M Chen Z Marquis M Av-eryt K B Tignor M and Miller H L Cambridge UniversityPress Cambridge United Kingdom and New York NY USA996 pp 2007a

IPCC Climate Change 2007 Impacts Adaptation and Vulnera-bility Contribution of Working Group II to the Fourth Assess-ment Report of the Intergovernmental Panel on Climate Changeedited by Parry M L Canziani O F Palutikof J P van derLinden P J and Hanson C E Cambridge University PressCambridge United Kingdom 1000 pp 2007b

IPSL The new IPSL climate system model IPSL-CM4rsquo InstitutPierre Simon Laplace des Sciences de lrsquoEnvironnement GlobalParis France 73 pp 2005

Jungclaus J H Botzet M Haak H Keenlyside N Luo J-J Latif M Marotzke J Mikolajewicz U and RoecknerE Ocean circulation and tropical variability in the AOGCMECHAM5MPI-OM J Climate 19 3952ndash3972 2006

Kalnay E Kanamitsu M Kistler R Collins W Deaven DGandin L Iredell M Saha S White G Woollen J Zhu YLeetmaa A Reynolds R Chelliah M Ebisuzaki W Hig-gins W Janowiak J Mo K C Ropelewski C Wang JenneR and Joseph D The NCEPNCAR 40-Year Reanalysis

Hydrol Earth Syst Sci 15 2789ndash2803 2011 wwwhydrol-earth-syst-scinet1527892011

M Ozdogan Climate change impacts on snow water availability in the Euphrates-Tigris basin 2803

Project Bull Am Meteorol Soc 77 437ndash471 1996K-1 model developers K-1 coupled model (MIROC) description

K-1 technical report 1 edited by Hasumi H and EmoriS Center for Climate System Research University of Tokyo34 pp 2004

Kitoh A Yatagai A and Alpert P First super-high-resolution model projection that the ancient ldquoFertile Crescentrdquowill disappear in this century Hydrol Res Lett 2 1ndash4doi103178HRL21 2008

Lambert S J and Boer G J CMIP1 evaluation and inter-comparison of coupled climate Models Clim Dynam 17 83ndash106 2001

Lettenmaier D P Wood A W Palmer R N Wood E F andStakhiv E Z Water resources implications of global warmingA US regional perspective Clim Change 43 537ndash579 1999

Leung L R and Qian Y The sensitivity of precipitation andsnowpack simulations to model resolution via nesting in regionsof complex terrain J Hydrometeorology 4 1025ndash1043 2003

Liang X Lettenmaier D P Wood E F and Burges S J Asimple hydrologically based model of land surface water and en-ergy fluxes for general circulation models J Geophys Res 9914415ndash14428 1994

Liston G E Representing subgrid snow cover heterogeneities inregional and global models J Climate 17 1381ndash1397 2004

Lohmann D Nolte-Holube R and Raschke E A large-scalehorizontal routing model to be coupled to land surface parame-terization schemes Tellus A 48 708ndash21 1996

McCabe G J and Wolock D M General-circulation-model sim-ulations of future snowpack in the western United States J AmWater Resour As 35 1473ndash1484 1999

McCabe G J and Wolock D M Recent declines in western USsnowpack in the context of twentieth-century climate variabilityEarth Interact 13 1ndash15 2009

Meehl G A Covey C Delworth T Latif M McAvaney BMitchell J F B Stouffer R J and Taylor K E The WCRPCMIP3 multi-model dataset A new era in climate change re-search Bull Am Meteorol Soc 88 1383ndash1394 2007

Milly P C D Dunne K A and Vecchia A V Global patterns oftrends in streamflow and water availability in a changing climateNature 438 347ndash350 2005

Mote P W Hamlet A F Clark M P and Lettenmaier D PDeclining mountain snowpack in western North America BullAm Meteorol Soc 86 39ndash49 2005

Mote P W Climate-driven variability and trends in mountainsnowpack in western North America J Climate 19 6209ndash62202006

Nash J E and Sutcliffe J V River flow forecasting through con-ceptual models part I ndash A discussion of principles J Hydrol10(3) 282ndash290 1970

NCDC Global Summary of day data product available athttpwwwncdcnoaagovcgi-binres40plpage=gsodhtml(last ac-cess March 2010) 2010

New M Lister D Hulme M and Makin I A high-resolutiondata set of surface climate over global land areas Clim Res 211ndash25 2002

Nijssen B Lettenmaier D P Liang X Wetzel S W and WoodE F Streamflow simulation for continental-scale river basinsWater Resour Res 33 711ndash24 1997

Onol B and Semazzi F H M Regionalization of Climate ChangeSimulations over the Eastern Mediterranean J Climate 221944ndash1961doi1011752008JCLI18071 2009

Pierce D W Barnett T P Santer B D and Gleckler P J Se-lecting global climate models for regional climate change stud-ies P Natl Acad Sci USA 106 8441ndash8446 2009

Russell G L Miller J R Rind D Ruedy R A Schmidt GA and Sheth S Comparison of model and observed regionaltemperature changes during the past 40 years J Geophys Res105 14891ndash14898 2000

Salas-Melia D Chauvin F Deque M Douville H Gueremy JF Marquet P Planton S Royer J F and Tyteca S Descrip-tion and validation of the CNRM-CM3 global coupled modelCNRM working note 103 available athttpwwwcnrmmeteofrscenario2004papercm3pdf 2005

Salathe E P Mote P W and Wiley M W Review of scenarioselection and downscaling methods for the assessment of climatechange impacts on hydrology in the United States pacific north-west Int J Climatol 27 1611ndash1621doi101002joc15402007

Scinocca J F McFarlane N A Lazare M Li J and PlummerD Technical Note The CCCma third generation AGCM and itsextension into the middle atmosphere Atmos Chem Phys 87055ndash7074doi105194acp-8-7055-2008 2008

Storck P Trees snow and flooding An investigation of forestcanopy effects on snow accumulation and melt at the plot andwatershed scales in the Pacific Northwest Water Resources Se-ries Tech Rep 161 Department of Civil and Environmental En-gineering University of Washington 176 pp 2000

Storck P Lettenmaier D P and Bolton S Measurement of snowinterception and canopy effects on snow accumulation and meltin mountainous maritime climate Oregon USA Water ResourRes 38 1223ndash1238 2002

Willmott C J On the validation of models Phys Geogr 2 184ndash194 1981

Yukimoto S Noda A Kitoh A Sugi M Kitamura Y HosakaM Shibata K Maeda S and Uchiyama T The New Me-teorological Research Institute Coupled GCM (MRI-CGCM2)Model climate and variability Pap Meteorol Geophys 51 47ndash88 2001

wwwhydrol-earth-syst-scinet1527892011 Hydrol Earth Syst Sci 15 2789ndash2803 2011