climate change impact on available water resources...

Earth Syst. Dynam., 4, 129–144, 2013www.earth-syst-dynam.net/4/129/2013/doi:10.5194/esd-4-129-2013© Author(s) 2013. CC Attribution 3.0 License.

EGU Journal Logos (RGB)

Advances in Geosciences

Open A

ccess

Natural Hazards and Earth System

Sciences

Open A

ccess

Annales Geophysicae

Open A

ccess

Nonlinear Processes in Geophysics

Open A

ccess

Atmospheric Chemistry

and Physics

Open A

ccess

Atmospheric Chemistry

and Physics

Open A

ccess

Discussions

Atmospheric Measurement

Techniques

Open A

ccess

Atmospheric Measurement

TechniquesO

pen Access

Discussions

Biogeosciences

Open A

ccess

Open A

ccess

BiogeosciencesDiscussions

Climate of the Past

Open A

ccess

Open A

ccess

Climate of the Past

Discussions

Earth System Dynamics

Open A

ccess

Open A

ccess

Earth System Dynamics

Discussions

GeoscientificInstrumentation

Methods andData Systems

Open A

ccess

GeoscientificInstrumentation

Methods andData Systems

Open A

ccess

Discussions

GeoscientificModel Development

Open A

ccess

Open A

ccess

GeoscientificModel Development

Discussions

Hydrology and Earth System

Sciences

Open A

ccess

Hydrology and Earth System

Sciences

Open A

ccess

Discussions

Ocean Science

Open A

ccess

Open A

ccess

Ocean ScienceDiscussions

Solid Earth

Open A

ccess

Open A

ccess

Solid EarthDiscussions

The Cryosphere

Open A

ccess

Open A

ccess

The CryosphereDiscussions

Natural Hazards and Earth System

Sciences

Open A

ccess

Discussions

Climate change impact on available water resources obtainedusing multiple global climate and hydrology models

S. Hagemann1, C. Chen1, D. B. Clark2, S. Folwell2, S. N. Gosling3, I. Haddeland4, N. Hanasaki5, J. Heinke6,F. Ludwig7, F. Voss8, and A. J. Wiltshire9

1Max Planck Institute for Meteorology, Bundesstr. 53, 20146 Hamburg, Germany2Centre for Ecology and Hydrology, Wallingford, UK3School of Geography, University of Nottingham, Nottingham, UK4Norwegian Water Resources and Energy Directorate, Oslo, Norway5National Institute for Environmental Studies, Tsukuba, Japan6Potsdam Institute for Climate Impact Research, Potsdam, Germany7Wageningen University and Research Centre, Wageningen, the Netherlands8Center for Environmental Systems Research, University of Kassel, Kassel, Germany9Met Office Hadley Centre, Exeter, UK

Correspondence to:S. Hagemann ([email protected])

Received: 12 November 2012 – Published in Earth Syst. Dynam. Discuss.: 4 December 2012Revised: 28 March 2013 – Accepted: 5 April 2013 – Published: 7 May 2013

Abstract. Climate change is expected to alter the hydrologi-cal cycle resulting in large-scale impacts on water availabil-ity. However, future climate change impact assessments arehighly uncertain. For the first time, multiple global climate(three) and hydrological models (eight) were used to system-atically assess the hydrological response to climate changeand project the future state of global water resources. Thismulti-model ensemble allows us to investigate how the hy-drology models contribute to the uncertainty in projectedhydrological changes compared to the climate models. Dueto their systematic biases, GCM outputs cannot be used di-rectly in hydrological impact studies, so a statistical bias cor-rection has been applied. The results show a large spreadin projected changes in water resources within the climate–hydrology modelling chain for some regions. They clearlydemonstrate that climate models are not the only source ofuncertainty for hydrological change, and that the spread re-sulting from the choice of the hydrology model is larger thanthe spread originating from the climate models over manyareas. But there are also areas showing a robust change sig-nal, such as at high latitudes and in some midlatitude regions,where the models agree on the sign of projected hydrologi-cal changes, indicative of higher confidence in this ensem-ble mean signal. In many catchments an increase of avail-

able water resources is expected but there are some severedecreases in Central and Southern Europe, the Middle East,the Mississippi River basin, southern Africa, southern Chinaand south-eastern Australia.

1 Introduction

Global warming due to increased greenhouse gas emissionsleads to changes in the distribution of water resources overmany regions, and the global and regional hydrological cy-cles have been greatly influenced by climate change in thepast century (Brutsaert and Palange, 1998; Scanlon et al.,2007; Solomon et al., 2007). Following the greenhouse gasemission scenarios for the 21st century (Nakicenovic et al.,2000), climate change will cause increased temperaturesand changes in precipitation. Estimates of future changes inprecipitation, however, are highly uncertain and depend onwhich climate model is used. Hydrological models have beenwidely used for assessments of water resources, especiallyfor studying the impacts of climate change. Many studieshave tried to assess the impact of climate change on the pastand future global water cycle. Multiple climate models areoften used so as to consider part of the uncertainty in future

Published by Copernicus Publications on behalf of the European Geosciences Union.

130 S. Hagemann et al.: Climate change impact on available water resources

climate change, but in most cases only one or two hydrolog-ical impact models are applied (Gosling and Arnell, 2011;Oki et al., 2003; Nijssen et al., 2001; Doll et al., 2003; Hage-mann et al., 2011). Recent studies (Haddeland et al., 2011;Gosling et al., 2011), however, showed that differences be-tween hydrological models are also a major source of un-certainty, and it was suggested that multiple impact modelsshould be used for climate change impact studies (Haddelandet al., 2011). The present study summarizes some of the ma-jor outcomes of the European Union project WATCH (WATerand global CHange;http://www.eu-watch.org). Here, climateprojections from three state-of-the-art coupled atmosphere–ocean general circulation models (GCMs), eight global hy-drology models (GHMs) and two emission scenarios are usedto assess the response of the terrestrial hydrological cycleto climate change and subsequent changes in available wa-ter resources. In this respect, this is not only the first studyto investigate future water resources using multiple GCMsand GHMs and emission scenarios, but it is also rigorous be-cause eight GHMs were applied, which is by far the mostapplied in any global climate change impact study thus far.As GCM simulations are significantly affected by systematicerrors, and results from a directly forced hydrological simu-lation will be unrealistic and of little use (Sharma et al., 2007;Hansen et al., 2006), bias-corrected GCM output was used toforce the GHMs.

Section 2 describes the GCM–GHM modelling chain andthe measures used to analyse the results. Mean changes inlarge-scale water fluxes and related uncertainties are pre-sented in Sect. 3, where a comparison to water fluxes ob-tained directly from the GCMs is also included. The impactof climate change on the available water resources is esti-mated based on the multi-model ensemble results in Sect. 4.Finally, the results are summarized and discussed in Sect. 5,thereby also highlighting aspects of uncertainty introducedby GHMs.

Note that this study focuses on the impact of climatechange alone on water fluxes and resources where direct hu-man influences are not considered. However, land use andwater use practices also play a role in the assessment ofwhether and how strongly human societies are affected inregions with changing water resources. For an estimation ofcombined anthropogenic and climate change effects, wateruse and further direct anthropogenic impacts on hydrologyhave to be taken into account, which will be investigated byHaddeland et al. (2013).

2 Models and methods

2.1 Models

Three GCMs are used in this study to provide quantita-tive estimates of future climate projections following theIPCC emission scenarios A2 and B1 (Nakicenovic et al.,

2000): ECHAM5/MPIOM (denoted as ECHAM5 hence-forth) of the Max Planck Institute for Meteorology, LMDZ-4of Institute Pierre Simon Laplace (denoted as IPSL hence-forth) and CNRM-CM3 of Centre National de RecherchesMeteorologiques, Meteo-France (denoted as CNRM hence-forth). Note that the GCMs chosen belong to different modelfamilies and cover some of the range in projected precipita-tion change from the CMIP3 (Meehl et al., 2007; see alsoSect. 5) ensemble (Mason and Knutti, 2011). The selectionof GCMs for this study was imposed by the availabilityof climate model data necessary to force the GHMs. A re-lated analysis of the original GCM results over Europe wasprovided by Hagemann et al. (2008).

GCMs exhibit a number of significant systematic biases intheir ability to simulate key features of the observed climatesystem (Randall et al., 2007). Despite the biases, the IPCCconcludes that there is still considerable confidence that cli-mate models provide credible quantitative estimates of futureclimate changes (Randall et al., 2007). However, until GCMsperfectly reproduce the current climate, GCM outputs can-not be used directly in hydrological impact studies withoutsome form of bias correction. When uncorrected GCM out-put is used as input to hydrological simulations, the result-ing amount and seasonal distribution of runoff may be farfrom observations, for example see Haddeland et al. (2012),Wood et al. (2004) and Sharma et al. (2007). Consequently, astatistical bias-correction method (Piani et al., 2010a,b) wasapplied to the GCM daily land precipitation and mean, andminimum and maximum daily land temperatures. The bias-correction method is based on a fitted histogram equaliza-tion function. This function is defined daily, as opposed toearlier published versions in which they were derived yearlyor seasonally at best, while conserving properties of robust-ness and eliminating unrealistic jumps at seasonal or monthlytransitions. Bias-correction factors are derived from 1960 to1999 from observed (Weedon et al., 2011) and original GCMdata, and then applied to 1960-2100 simulations (Piani et al.,2010b). For details about the GCM simulations and the bias-corrected data, see Hagemann et al. (2011). Although biascorrection of climate forcing fields has become a necessarystep in climate impact simulations, many recent studies haveidentified limitations and pitfalls associated with this process(e.g. Haerter et al., 2011; Ehret et al., 2012). However, asstated by Piani and Haerter (2012) the bias correction usedhere has been applied successfully to regional climate modeloutput over Europe to examine the effects on both simulatedclimate and extreme hydrological events (Dosio and Paruolo,2011; Rojas et al., 2011).

Eight GHMs (MPI-HM, LPJmL, WaterGAP, VIC, Mac-PDM.09, H08, GWAVA and JULES) were used to calculatehistoric and future water fluxes and simulate the land surfacehydrology at a horizontal resolution of 0.5◦ (about 50 km gridspacing). The major model characteristics are listed in Ta-ble 1. The GHMs differ in their evapotranspiration and runoffschemes, and the differences in model parameterizations are

Earth Syst. Dynam., 4, 129–144, 2013 www.earth-syst-dynam.net/4/129/2013/

http://www.eu-watch.org

S. Hagemann et al.: Climate change impact on available water resources 131

Table 1.Participating models, including their main characteristics (adapted from Haddeland et al., 2011).

Model Time Meteorological Energy ETpot Runoff Snowstep forcing balance scheme2 scheme4 scheme

variables1

GWAVA Daily P , T , W , No Penman– Saturation DegreeQ, LWn, Monteith excess/ daySW, SP beta

function

H08 Daily R, S, T , Yes Bulk Saturation EnergyW , Q, formula3 excess/ balanceLW, SW, betaSP function

JULES 1 h R, S, T , Yes Penman– Infiltration EnergyW , Q, Monteith excess/ balanceLW, SW, DarcySP

LPJmL Daily P , T , No Priestley– Saturation DegreeLWn, SW Taylor excess day

Mac-PDM.09 Daily P , T , W , No Penman– Saturation DegreeQ, LWn, Monteith excess/ daySW beta

function

MPI-HM Daily P , T No Thornth- Saturation Degreewaite excess/ day

betafunction

VIC Daily/ P , Tmax, Snow Penman– Saturation Energy3 h Tmin, W , season Monteith excess/ balance

Q, LW, betaSW, SP function

WaterGAP Daily P , T , No Priestley– beta DegreeLWn, SW Taylor function day

1 R: rainfall rate,S: snowfall rate,P : precipitation (rain or snow distinguished in the model),T : air temperature,Tmax:maximum daily air temperature,Tmin: minimum daily air temperature,W : wind speed,Q: specific humidity, LW:longwave radiation flux (downward), LWn: longwave radiation flux (net), SW: shortwave radiation flux (downward), SP:surface pressure.2 ETpot: potential evapotranspiration.3 Bulk formula: bulk transfer coefficients are used when

calculating the turbulent heat fluxes.4 Beta function: runoff is a nonlinear function of soil moisture.

to some extent reflected in the forcing variables that are usedby each (Table 1). For associated model references and vali-dation of GHM model results using quasi-observational forc-ing data, see Haddeland et al. (2011). The variability amongthe GHM results forced with bias-corrected GCM output andassociated runoff biases for the control period 1971–2000are in accordance with the validation shown in Haddelandet al. (2011).

Note that our study focused on the impact of climatechange on hydrology, and anthropogenic influences such aswater withdrawals and reservoirs were not taken into accountin the hydrological simulations.

2.2 Experimental setup and measures

Figure 1 presents an overview on the global modellingchain developed and employed within the WATCH project(cf. Sect. 2.1). To evaluate the projected hydrological cy-cle obtained from the multi-model ensemble, the ensemblemeans and the spread around these means due to differentsources were calculated. For both emission scenarios, tran-sient simulations from 1960–2100 were conducted by theGHMs. For the high emission A2 scenario, simulations byall 8 GHMs forced by output from the 3 GCMs resultedin 24 different time series for each hydrological variable.For the low emission B1 scenario, 18 simulations were ob-tained from 6 GHMs (excluding JULES and H08) forced by

www.earth-syst-dynam.net/4/129/2013/ Earth Syst. Dynam., 4, 129–144, 2013


28

Figures 1

2

Figure 1: Global modelling chain in the WATCH project. 3

4

5

6

Climate model input from 3 GCMs: Precipitation P, 2m Temperature T, other

variables (cf. Table 1)

Interpolation to0.5 degree

Statistical bias correction of P and T fields

8 GHMs

A2 scenario B1 scenario

6 GHMs

3x8=24 simulations 3x6=18 simulations

Fig. 1.Global modelling chain in the WATCH project.

3 GCMs. The ensemble mean of each hydrological variable(evapotranspiration and runoff) was calculated for the con-trol (1971–2000) and future (2071–2100) periods, and thechanges are expressed for the future relative to the controlperiod.

In this study, uncertainty is reflected by the spread of themodel results due to the choice of the GCM (3), GHM (8)or emission scenario (2). For the first two, the spread is cal-culated from the normalized standard deviation (or coeffi-cient of variation, CV) that is commonly used to express rel-ative differences between models. Here, the spread due tothe choice of the GCM is determined by taking the ensemblemean of the 8 GHM results for each GCM, and subsequentlycalculating the standard deviation among the 3 GCMs. TheGHM spread is calculated correspondingly from the stan-dard deviation of the GCM ensemble (3) means for each ofthe 8 GHMs. For the emissions, the scenario spread is rep-resented by the differences between the ensemble mean re-sults of the high emission scenario A2 and the lower emis-sion scenario B1, obtained from those GHM simulationsthat were conducted for both scenarios (6x× 3 = 18). As thisstudy focuses on changes in available water resources, as-sociated changes in the main components of the terrestrialwater balance are considered, i.e. precipitation (simulated bythe GCMs), total runoff and evapotranspiration (simulated bythe GHMs forced with bias-corrected GCM data).

3 Mean changes in large-scale water fluxes and relateduncertainties

3.1 Results from the GCM-GHM ensemble

In the following, projected changes are associated with theA2 scenario if not mentioned otherwise. According to theresults of the bias-corrected GCM A2 simulations, precipi-tation is projected to increase by the end of the 21st century

across the higher latitude regions and in parts of the mid-dle latitudes (Fig. 2a). Parts of the Middle East, the Mediter-ranean region, the southern parts of North America, Africaand Southern Australia will receive less precipitation. Thesefuture changes in precipitation show similar patterns to theensemble of 21 GCM results summarized in the 4th IPCCAssessment Report (Solomon et al., 2007). Noticeable un-certainties in the simulated precipitation change occur overnorthern Africa, the Indian monsoon region and Himalaya,some northern and western parts of South America, a smallarea in western Australia, the southern part of North Americaand over Greenland (Fig. 2b)

Water resources depend strongly on the available runoff,which in the long term is constrained by incoming precipi-tation and outgoing evapotranspiration (ET). Runoff is pro-jected to decrease over the eastern part of Australia, southernparts of Africa and the US, the north-eastern part of SouthAmerica, the southern part of Europe, and a large part ofthe Middle East (Fig. 3a). Largely, the change pattern ofrunoff follows the ensemble mean change of precipitation(Fig. 2a). This behaviour is similar for the projected meanchanges in ET (Fig. 3b), with the noticeable exception thatET also increases in the northern high- and midlatitudes andextends further southwards into the transitional wet regions.This can be observed over south-eastern US, central Europeand eastern Asia.

Although for many large regions around the globe thereis generally a large spread of absolute changes predicted bythe different model simulations, many models agree on thesign of projected changes (Fig. 3c and d). For runoff, regionswith relatively high mean changes are generally those regionswhere the majority of the 24 GCM–GHM model combina-tions agree on the sign of change. This indicates that the pro-jected runoff changes expressed by the multi-model mean arerobust within the ensemble. The same applies for ET, exceptfor relative changes in the central Amazon and the very dryregions of the Sahara and southern Mexico where there isless agreement between the models.

For both ET and runoff, we estimated whether the largestspread in the projected changes originates from the choiceof GCM (Fig. 4a and b), GHM (Fig. 4c and d) or scenario.For ET (Fig. 5a), the uncertainty in the projected changesis largely dominated by the spread due to the choice of theGHM. Especially over high latitude regions, GHMs causenoticeable uncertainty (Fig. 4c) where the spread originatingfrom the GCMs is rather low (Fig. 4a). For runoff, the CVvalues representing the GCM spread (Fig. 4b) are often com-parable to those for the GHM spread (Fig. 4d), even thoughthe GCM spread is larger over many regions of the globe(Fig. 5b). The spread patterns associated with the runoffchanges suggest that they are partially affected by the corre-sponding spreads in the ET changes. But the associated CVvalues are strongly reduced compared to the CV of the ETchanges, especially over the mid- and low latitudes includ-ing Central and Southern Europe. This means that on one



29

1

2

3

Figure 2: Annual ensemble mean changes in bias corrected precipitation [mm/a] projected by 4

the three GCMs following the A2 scenario for 2071-2100 compared to 1971-2000 (a) and the 5

CV of these changes from the three GCMs (b). 6

7

8

9

10

a)

b)

Fig. 2. Annual ensemble mean changes in bias corrected precipita-tion (mm a−1) projected by the three GCMs following the A2 sce-nario for 2071–2100 compared to 1971–2000(a) and the CV ofthese changes from the three GCMs(b).

hand the main spread in runoff changes originates from thechoice of the GCM, particularly through the projected pre-cipitation changes (Fig. 2). On the other hand, there are sev-eral areas where the runoff spread is dominated by the spreadin ET changes that is largely induced by the GHMs, notice-ably over the high northern latitudes. Note that in a recentstudy, which follows a similar model setup as in our study,the GHM runoff spread is dominant over even larger areas(Fig. 1 in Schewe et al., 2013). This can be explained bythe fact that we are considering a common time period whileSchewe et al. (2013) are comparing model results for a uni-fied global warming (2◦C above present day) for which theGCM patterns are more similar due to the same amount ofwarming in the global mean temperature.

We now consider whether the projected changes and as-sociated spreads behave differently in dry (humid) areaswhere ET tends to be limited primarily by the availabil-ity of moisture (energy). These areas are represented bylow (high) values of the ensemble mean runoff coefficient(runoff R divided by precipitationP ) for the present-dayclimate (Fig. 6a). Figure 6b and c show that humid areasexpect increases in ET and runoff, while decreases in bothvariables occur only over some medium wet (e.g. Danube)to dry (e.g. Murray) areas (R/P < 0.6). Considering thespreads, it can be noted that for both ET and runoff the

GCM spread tends to be larger for dry areas than for hu-mid areas (Fig. 7a and b). For the GHM spread, there is aless clear tendency, even though some larger CVs (CV> 1.7for ET, CV> 1.2 for runoff) only occur over medium wet todry areas (R/P < 0.5).

In the A2 scenario some high- and midlatitude regionsshow more precipitation and runoff than the B1 scenario(Fig. S1 in the Supplement). In most other areas the projectedchanges are rather comparable. With regard to ET, most areasshow larger values in the future period for the A2 scenariothan for the B1 scenario, especially in the Amazon area. ETchanges for A2 that are smaller than for B1 are projectedover the western part of North America, the southern part ofSouth America and the Middle East. This is an indication ofa stronger drying of these regions with increased greenhousegas concentrations. Noticeable spread due to the choice ofscenario largely occurs over areas where the projected en-semble mean A2 change is relatively small. In addition,larger runoff spreads occur over the high northern latitudes,northern USA, some parts of South America and Africa thatare comparable and partially larger (see Fig. 5) than thoseoriginating from the GCMs or GHMs. Over Africa, south-ern South America and northern USA, these uncertaintiesare induced by scenario differences in precipitation that aremuch larger than those originating from the choice of theGCM. Over Siberia, the scenario spread for runoff seems tobe related to the combined effect of scenario spreads in pre-cipitation and ET (Fig. S1 in the Supplement). In most otherareas the scenario spread is smaller than those induced by theGCM or GHM.

Figure 8 sets the projected mean A2 changes in relationto the associated spreads over selected large-scale catch-ments that include the largest rivers on earth as well as somesmaller catchments in Europe (Baltic Sea, Danube) and Aus-tralia (Murray). Here the spreads are expressed by the abso-lute standard deviation about (±) the respective mean changeso as to allow direct comparisons between them. FollowingHagemann et al. (2009), a projected change is consideredrobust if the change is larger than the largest spread. Fig-ure 8 shows that for many catchments the mean A2 changeis robust compared to the different spreads, especially forrunoff. For several catchments the direction of change is notrobust, but is relatively well constrained (larger than half ofthe largest spread), i.e. Baltic Sea, Mississippi and Nile forrunoff, and Amazon, Congo, Ganges/Brahmaputra and Nilefor ET. Noticeably, the large GCM spread prohibits a con-strained runoff change signal over the Ganges/Brahmaputra(runoff) and Parana (runoff and ET) catchments. Also, pro-jected changes are not constrained for the Danube (ET) andYangtze (runoff) due to low projected mean changes andlarge GHM spreads. Consistent with Fig. 5, the GHM spreadof the ET changes is largest for most of the catchments (ex-cept for Mississippi and Parana), while for runoff the GCMspread prevails for 8 of the 12 catchments considered. Butfor the large area of the 6 largest Arctic rivers as well as for



30

1

2 3 Figure 3: Ensemble mean results for evapotranspiration (left column) and runoff (right 4 column) from 24 simulations (8 GHMs using output from 3 GCMs): Mean future changes 5 [mm/a] following the A2 scenario for 2071-2100 compared to 1971-2000 (a, b), Number of 6 simulations showing a positive change minus the number showing a negative change (c, d). 7

8 9

a) b)

c) d)

Fig. 3.Ensemble mean results for evapotranspiration (left column) and runoff (right column) from 24 simulations (8 GHMs using output from3 GCMs): mean future changes (mm a−1) following the A2 scenario for 2071–2100 compared to 1971–2000(a, b); number of simulationsshowing a positive change minus the number showing a negative change(c, d).

31

1

2 3 Figure 4: CV of evapotranspiration (left column) and runoff (right column) changes shown in 4 Fig. 2 from 3 GCMs (a, b) and CV from 8 GHMs (c, d). 5

6

a) b)

c) d)

Fig. 4. CVs of evapotranspiration (left column) and runoff (right column) changes shown in Fig. 2 from 3 GCMs(a, b) and from8 GHMs(c, d).



32

1

2

Figure 5: Areas where the largest spread in projected evapotranspiration (a) and runoff (b) 3

changes (2071-2100 compared to 1971-2000) is due to the choice of the GCM (blue), GHM 4

(red) or scenario (green). 5

6

a) b)

Fig. 5. Areas where the largest spread in projected evapotranspiration(a) and runoff(b) changes (2071–2100 compared to 1971–2000) isdue to the choice of the GCM (blue), GHM (red) or scenario (green).

33

1

2

3

Figure 6: (a) Ensemble mean runoff coefficient (R/P) for present day (1971-2000) and 4

scatterplots showing how the runoff coefficient relates to the A2 changes in (b) 5

evapotranspiration and (c) runoff obtained from the GCM-GHM ensemble. For panels b and 6

c, the colours show the number of scatter points. 7

8

9

a)

b) c)

Fig. 6. (a)Ensemble mean runoff coefficient (R/P ) for present day (1971–2000), and scatter plots showing how the runoff coefficient relatesto the A2 changes in(b) evapotranspiration and(c) runoff obtained from the GCM–GHM ensemble. For(b) and(c), the colours show thenumber of scatter points.



34

1

2

3

Figure 7: Scatterplots showing how the ensemble mean runoff coefficient (R/P) for present 4 day relates to the CV of A2 evapotranspiration (left column) and runoff (right column) 5 changes obtained from 3 GCMs (a, b) and to the CV from 8 GHMs (c, d). The colours show 6 the number of scatter points. 7

8

9

10

11

a) b)

c) d)

Fig. 7. Scatter plots showing how the ensemble mean runoff coefficient (R/P ) for present day relates to the CV of A2 evapotranspiration(left column) and runoff (right column) changes obtained from 3 GCMs(a, b) and to the CV from 8 GHMs(c, d). The colours show thenumber of scatter points.

the catchments of Amur, Baltic Sea and Yangtze, the GHMspread is largest, demonstrating the impact of uncertaintiesin the projected ET changes on the runoff changes over thehigh- and some midlatitude regions. The scenario spread isgenerally the smallest spread, especially for runoff, and it isalways smaller than the GHM spread, except for ET in theMurray catchment where it is largest. Catchments where thescenario spread is larger than the GCM spread usually com-prise areas where the GCM spread is rather low. It should benoted that only three GCMs were applied in this study, so thatthe uncertainty due to the choice of the GCM is likely some-what underrepresented, even though the chosen GCMs coversome of the range in projected precipitation change amongthe CMIP3models (see Sect. 2.1).

3.2 Comparison to direct GCM output

With regard to the projected changes in ET and runoff, itis interesting to compare the results from the GCM–GHMensemble with the uncorrected climate model output ob-tained directly from the three GCMs. The projected meanA2 changes and associated spreads among the 3 GCMs

are shown in Fig. 9. The large-scale patterns of the meanchanges are similar to those for the GCM–GHM ensemble(see Fig. S2a and b in the Supplement and Fig. 3a and b),but the absolute intensity of change is mostly lower in thedirect GCM output (Fig. 9a and b). This is supported byFig. 10, where the intensity of change (decreases and in-creases) is compared. Here, areas indicating larger changes inthe GCM–GHM ensemble exceed areas with larger changesin the original GCM output for both ET and runoff. Areaswhere the sign of change differs are relatively scarce. Inthis respect, noticeable larger areas are seen for the runoffchanges over Australia and east of the Caspian Sea. Con-sidering average changes over large catchments (Fig. 8), itcan be noted that the projected A2 changes from the origi-nal GCM output are often significantly lower, up to 50–70 %less, than the respective changes projected by the GCM–GHM ensemble. This may partly be due to the small sampleof 3 GCMs that, with regard to the calculation of ET and theirprojected changes, may not cover the full space of possiblemodel solutions.

The spread patterns behave differently than the corre-sponding mean changes. For both ET and runoff, the spread



35

1 2 Figure 8a: Mean projected A2 and B1 changes (2071-2100 compared to 1971-2000) in 3 evapotranspiration and associated standard deviations (Std.) due to the choice of the GCM 4 (Std. 3 GCMs) and GHM (Std. 8 GHMs) about the A2 mean (i.e. ±Std.) over selected large 5 catchments. The mean A2 change and Std. obtained directly from the three original GCM 6 output are also shown (A2 GCM org., Std. GCM org.). The 6 largest Arctic rivers comprise 7 the catchments of Mackenzie, Northern Dvina, Yenisei, Ob, Lena and Kolyma. 8

9

a)

36

1 Figure 8b: As Fig. 8a, but for runoff. 2

3

b)

Fig. 8. (a)Mean projected A2 and B1 changes (2071–2100 compared to 1971–2000) in evapotranspiration and associated standard deviations(Std.) due to the choice of the GCM (Std. 3 GCMs) and GHM (Std. 8 GHMs) about the A2 mean (i.e.±Std.) over selected large catchments.The mean A2 change and Std. obtained directly from the three original GCM outputs are also shown (A2 GCM org., Std. GCM org.). The6 largest Arctic rivers comprise the catchments of Mackenzie, Northern Dvina, Yenisei, Ob, Lena and Kolyma.(b) As (a), but for runoff.

in the original (uncorrected) GCM output tends to be largerthan the GCM spread in the GCM–GHM ensemble (Fig. 9cand d). This is the case for most parts of the globe (Fig. 11).For ET, the pattern of spread in the original GCM output(Fig. S2c in the Supplement) is rather similar to the spreaddue to the choice of the GCM in the GCM–GHM ensem-ble (Fig. 4a) over the Southern Hemisphere and the Tropics,but there are larger differences over the mid- and especiallyover the high northern latitudes. For runoff, both spreads(Fig. S2d in the Supplement, Fig. 4b) show only some sim-ilarities over South America and Central Africa. The higher

spread in the original GCM output seems to be a direct re-sult of the GCM-specific biases in precipitation and tem-perature. In the GCM–GHM ensemble, the bias correctionis not only reducing the spread (per definition) in the GCMdata for the control period, but also the spread in the climatechange signal. Note that the absolute standard deviations areshown in Fig. 8 while the spreads represented by the CV inthe Figs. 4, 7, 11 and S2 (in the Supplement) are relative val-ues. Thus, Fig. 8 shows catchment-averaged absolute stan-dard deviations of the original GCM output that are partiallysmaller than the corresponding standard deviations in the



37

1

2 3

4

5

Figure 9: Scatterplots to compare the results obtained directly from the output of the 3 GCMs 6

to those from the GCM-GHM ensemble: Mean future changes [mm/a] of evapotranspiration 7

(left column) and runoff (right column) following the A2 scenario for 2071-2100 compared to 8

1971-2000 (a, b), and the CV of these changes from the 3 GCMs (c, d). The colours show the 9

number of scatter points. 10

11

a) b)

c) d)

Fig. 9. Scatter plots to compare the results obtained directly from the output of the 3 GCMs to those from the GCM–GHM ensemble: meanfuture changes (mm a−1) of evapotranspiration (left column) and runoff (right column) following the A2 scenario for 2071–2100 comparedto 1971–2000(a, b), and the CV of these changes from the 3 GCMs(c, d). The colours show the number of scatter points.

38

1 2

Figure 10: Comparison of mean A2 evapotranspiration (left panel) and runoff (right panel) 3

changes (2071–2100 compared to 1971–2000) projected by the GCM-GHM ensemble and the 4

original uncorrected GCMs. Areas are indicated where the projected decreases and increases 5

are larger in the GCM-GHM ensemble (red and blue, respectively) than in the original GCM 6

output and vice versa (orange and turquoise, respectively), as well as areas where the sign of 7

projected change differs between them (green). 8

9

Fig. 10. Comparison of mean A2 evapotranspiration (left panel) and runoff (right panel) changes (2071–2100 compared to 1971–2000)projected by the GCM–GHM ensemble and the original uncorrected GCMs. Areas are indicated where the projected decreases and increasesare larger in the GCM–GHM ensemble (red and blue, respectively) than in the original GCM output and vice versa (orange and turquoise,respectively), as well as areas where the sign of projected change differs between them (green).

GCM–GHM ensemble due to the choice of the GCM, espe-cially for runoff. But as the projected mean changes in the di-rect GCM output are mostly weaker than for the GCM–GHMensemble (see above in the text), the associated relative val-ues become much larger for the direct GCM output. Thus, the

spreads represented by the CV are often larger for the directGCM output than their GCM–GHM counterparts (Fig. 11).

This means that even though the projected ET and runoffchanges of the original GCM output are fully consistent withthe other GCM variables, the associated spread and related



39

1 2

Figure 11: Comparison of the spreads associated with the mean A2 evapotranspiration (left 3

panel) and runoff (right panel) changes due to the choice of the GCM for the GCM-GHM 4

ensemble and the original uncorrected GCMs. Areas are indicated where the CV is larger for 5

the GCM-GHM ensemble (blue) or for the original GCM output (red). 6

7

8

Fig. 11.Comparison of the spreads associated with the mean A2 evapotranspiration (left panel) and runoff (right panel) changes due to thechoice of the GCM for the GCM–GHM ensemble and the original uncorrected GCMs. Areas are indicated where the CV is larger for theGCM–GHM ensemble (blue) or for the original GCM output (red).

40

1

2

3 4

Figure 12: A2 changes (2071-2100 compared to 1971-2000) in available water resources 5

projected by the 8 GHM ensemble averaged for all 3 GCMs (a), ECHAM (b), CNRM (c) and 6

IPSL (d). 7

8

9

10

a) b)

c) d)

Fig. 12.A2 changes (2071–2100 compared to 1971–2000) in available water resources projected by the 8 GHM ensemble averaged for all3 GCMs(a), ECHAM (b), CNRM (c) and IPSL(d).

uncertainties in these changes are larger than for the GCM–GHM ensemble where the consistency between variables isnot necessarily the case due to the bias correction. Thus, theGCM-specific biases in precipitation and temperature lead tolarger spreads in the projected changes of terrestrial compo-nents of the hydrological cycle. Consequently, these resultsshow a beneficial characteristic of the chosen model setupcompared to the direct use of GCM data for impact assess-

ment, which can be regarded in the on-going discussion onpros and cons of bias correction (see Sect. 2.1).

4 Impact on the available water resources

Based on the results from the 8 GHMs and 3 GCMs,catchment-based maps of changes in available water re-sources can identify areas that are vulnerable to projected



41

1

2

3 Figure 13: Ensemble seasonal mean future changes [mm/season] of runoff from 24 4 simulations (8 GHMs using output from 3 GCMs) following the A2 scenario for 2071-2100 5 compared to 1971-2000: (a) Boreal winter (DJF), (b) spring (MAM), (c) summer (JJA) and 6 (d) autumn (SON). 7

8

9

a) b)

c) d)

Fig. 13. Ensemble seasonal mean future changes (mm season−1) of runoff from 24 simulations (8 GHMs using output from 3 GCMs)following the A2 scenario for 2071–2100 compared to 1971–2000:(a) boreal winter (DJF),(b) spring (MAM), (c) summer (JJA) and(d) autumn (SON).

climate change with regard to water availability. In this re-spect, available water resources are defined as the total an-nual runoff (R) minus the mean environmental water require-ments. Using results of Smakhtin et al. (2004), environmentalwater requirements (EWR) for a catchment were defined as30 % of the total annual runoff. Assuming EWR will notchange significantly in the future, available water resources(1AW) can be estimated as

1AW = ((RScen− EWR) − (RC20 − EWR))/(RC20 − EWR)

= (RScen− RC20)/(RC20 − EWR) .

Here,RC20 andRScenare the mean annual runoff for the cur-rent climate (1971–2000) and future scenario (2071–2100)periods, respectively, and EWR = 0.3RC20.

Although for runoff there is large uncertainty induced bythe choice of GCM (see Figs. 4 and 5), most large catchments(e.g. Amazon, Parana, Nile, Congo, Ganges/Brahmaputra)show an increase in available water resources in the future(Fig. 12). There is some agreement between the 3 GCMsas to where the available water resources are expectedto decrease considerably (more than 10 %). These regionscomprise Central, Eastern and Southern Europe, the catch-ments of the Euphrates/Tigris in the Middle East, Missis-sippi in North America, Zhu Jiang in southern China, Mur-ray in SE Australia, and Okavango and Limpopo in southernAfrica. Here the projections based on the different GCMslargely agree. As only three GCMs are considered, poten-tial future significant reductions in available water resources

based only on one of these GCMs cannot be neglected forthis impact assessment. Examples of significant decreasesare seen in the following large catchments for individualGCMs (Fig. 12): Parana (more than−50 % for IPSL) andUruguay (more than−20 % for IPSL) in South America, Or-ange (more than−20 % for ECHAM) in South Africa, Sa-hel zone comprising Senegal, Niger, Volta and Chari (morethan−50 % for IPSL) and Central and Eastern Asia compris-ing the Ganges/Brahmaputra (more than−20 % for IPSL),Indus, Amudarja and Huang He (more than−10 % forECHAM).

The analysis presented above has been conducted on theannual scale, but in some regions available water resourcesare also affected by seasonal changes. Figure 13 shows theprojected changes in runoff per season that can be com-pared to the annual mean changes shown in Fig. 3b. Regionsthat might experience a seasonal reduction in runoff that ismore severe than in the respective annual mean are poten-tially affected by a seasonal reduction in available water re-sources. These regions comprise parts of southern Africa inDJF (Fig. 13a), central eastern South America, Eastern USand Eastern Europe in MAM (Fig. 13b), almost the whole ofEurope and western Siberia as well as Western US and south-ern and western Canada in JJA (Fig. 13c), and north-westernSouth America in SON (Fig. 13d).

It has to be noted that even if the long-term mean an-nual change in annual water resources may be quite smallfor some regions, they might be strongly affected by changes



in interannual variability and the occurrence of droughts. Ananalysis of these effects is beyond the scope of the presentstudy, but it is an important topic for future studies. In thisrespect, Prudhomme et al. (2013) investigated the impact ofclimate change on hydrological droughts.

Climate change has been identified as a major influenceon basin water balances. However, land use and water usepractices also play a role in the assessment of whether andhow strongly human societies are affected in those changingregions. The impact of reduced water availability on differentregions also depends on the total water demand and duringwhich season availability will change. These are necessarysubjects for future studies.

5 Conclusions and discussion

The climate modelling community has a long history of sys-tematic model intercomparison through the climate modelintercomparison projects (CMIPs; Meehl et al., 2000). Theresults of CMIP3 (Meehl et al., 2007) are the basis for the fu-ture climate change projections presented in the IPCC 4th as-sessment report (Solomon et al., 2007). The results of CMIPsare also used in many climate change impact assessmentsto quantify the uncertainties originating from climate mod-els and emission scenarios (Gosling et al., 2012; Osborneet al., 2013; Sperna Weiland et al., 2012). Most of the im-pact assessments, however, only use a single impact model.Here, for the first time a multi-model ensemble comprisingmultiple global climate (3) and global hydrology models (8)was used to assess future large-scale changes in land surfacewater fluxes and available water resources. The results pre-sented here clearly show that climate change impacts do notonly depend on emission scenarios and climate models, butthat different impact models give considerably different re-sults. In some regions the spread of the impact models islarger than that of the climate models.

This ensemble of many different simulations formed thebasis for a comparison of the uncertainty in the projectedchanges originating from the choice of the GCM, the GHMand the emission scenario (B1 and A2). We did not use thedirect output of GCMs but instead bias-corrected the GCMtime series of precipitation and temperature. Thus, we es-sentially removed differences between GCMs in those timeseries in the baseline period, which correspondingly reducesthe absolute spread of the GCMs over many regions. Notethat the bias correction also adds uncertainty to the projec-tions. Precipitation and temperature are corrected indepen-dently. Several studies, such as that of Berg et al. (2009),have shown that daily precipitation shows some scaling withtemperature so that future improvements of the bias correc-tion method may be achieved with multivariate approaches(such as presented by Piani and Haerter, 2012) that takethese dependencies into account. In addition, GCM variablesother than precipitation and temperature are not corrected,which potentially introduces inconsistencies between vari-

ables, e.g. between the near surface air humidity and temper-ature used by some of the GHMs as forcing. However, Had-deland et al. (2012) found that the relative values of projectedhydrological change are very similar if other GCM variablesare also bias corrected. Thus, it can be assumed that the im-pact of these inconsistencies is generally rather small. An-other uncertainty inherent to the chosen model setup is thatthe GHM ET does not feed back to the atmosphere, hence itdoes not impact GCM precipitation or near surface specifichumidity.

Despite these uncertainties and inconsistencies, there arecurrently not many alternatives to this approach for hydro-logical impact assessments. As mentioned in Sect. 2.1, out-put from the current generation of GCMs is generally not di-rectly applicable for impact studies, mainly due to the largebiases in precipitation and associated biases in surface hy-drology (runoff, ET). These biases impact the GCM signals,as do the different GCM parameterizations, and thus lead touncertainties in projected changes in terrestrial componentsof the hydrological cycle that are larger than in the modelsetup presented here. These differences can also lead to dif-ferent climate change signals which, in the present study, aregenerally weaker in the uncorrected GCM output, but thismay be a characteristic of the chosen 3 GCMs. Note that theusability of direct GCM output may change in the future asdevelopments are also being targeted at improving the watercycle in GCMs and there are some studies in the literatureimplying there is useful information obtainable directly fromGCMs (e.g. Falloon et al., 2011).

Future changes in runoff and ET generally follow theprojected changes in the bias-corrected GCM precipitation.These changes comprise projected increases over the highlatitudes and some midlatitude regions, while Southern Eu-rope, large parts of the Middle East, southern parts of Africaand the USA, eastern Australia and the north-eastern part ofSouth America will likely experience decreased runoff in thefuture compared to the control period. These changes aregenerally consistent with those in previous studies. The re-sults of Haddeland et al. (2011) indicate that, globally aver-aged, the majority of the interannual variation in precipita-tion feeds directly through to the runoff and that the evap-otranspiration is constrained by other atmospheric factorssuch as temperature, radiation, and humidity. The same isvalid for future changes in runoff and ET, while ET will alsobe affected by precipitation changes in transitional wet re-gions where the availability of soil moisture directly affectsthe evaporative fraction (Seneviratne et al., 2010). In the wethigh latitudes (see Fig. 6a) where ET is energy-limited, theprecipitation increase (Fig. 2a) is accompanied by a largerwarming that leads to more available energy which in turnresults in increased ET. Associated with reduced runoff, asignificant reduction in available water resources will occurin many catchments on the annual scale, but also for spe-cific seasons. For those regions, the projections based on thedifferent GCMs largely agree. Moreover, when considering



the large uncertainty associated with the choice of GCM, itis also possible that some regions might be affected by a sig-nificant future reduction in available water resources whenthis is projected only by one of the GCMs.

A deeper analysis of the cause of the projected changesis beyond the scope of the present study as the parame-terizations of ET and runoff vary substantially between theGHMs (Table 1 and Haddeland et al., 2011), and the com-plicated interactions between the various processes make itinfeasible to explain the causes of many simulation differ-ences in detail, as noted in previous model intercomparisons(e.g. Koster and Milly, 1997). However, some comments canbe made. The results presented here show that the uncer-tainty in projected changes of land surface water fluxes dueto the choice of the GHM cannot be neglected over many re-gions of the earth. This uncertainly mainly arises from thedifferent model formulations used to represent hydrologicalprocesses in the GHMs. Haddeland et al. (2011) found thatsignificant differences between simulations by land surfacemodels (LSMs; models that calculate the land surface energybalance) and “pure” GHMs (without energy balance calcula-tion) for the current climate are partly caused by the snowscheme applied. In that study, which included all the hy-drological models that are included in this study, the phys-ically based energy balance approach used by LSMs gen-erally resulted in lower snow water equivalent values thanthe conceptual degree-day approach used by most GHMs.Some differences in simulated runoff and evapotranspira-tion are explained by model parameterizations, such as thedifferent treatment of soil moisture and evapotranspiration(Hagemann et al., 2011; Haddeland et al., 2011), althoughthe processes included and parameterizations used are notdistinct to either LSMs or pure GHMs. The present studyindicates that large differences in the projected changes be-tween the GHMs may be attributed to the different modelformulations of evapotranspiration. This becomes especiallyobvious if the projected changes in evapotranspiration areconsidered for which the uncertainty related to the choiceof the GHM is larger than due to the choice of the GCMover many regions. Uncertainties in projected evapotranspi-ration changes are generally shown to be due to the choice ofthe impact model, whereas the choice of the climate modelprevails for the future projections of runoff, except for thoseareas where the evapotranspiration is strongly affecting thefuture changes in runoff and, thus, the GHM uncertainty ismore pronounced.

It should be noted that the response of stomata to CO2 andthe related effect on evapotranspiration is neither accountedfor by the GCMs nor by the GHMs (except for 1 GHM:LPJmL). It has been shown that the stomatal response leadsto a physiological forcing on runoff (e.g. Betts et al., 2007)and is also important for large-scale precipitation over land,so that similar GCM–GHM exercises (see below) with theforthcoming CMIP5 results are recommended to address as-sociated uncertainties. Note also that individual catchment

properties can affect the magnitude of hydrological response(see e.g. Arora, 2002; Roderick and Farquhar, 2011; Rennerand Bernhofer, 2012). Thus, an application of the frameworkof Renner and Bernhofer (2012) may be a valuable future ex-tension of our study by linking some of these changes to thearidity index.

The impact of the bias correction (see above) on theprojected changes is probably smaller than that caused bythe different GHMs. A direct comparison of the simulatedGHM changes in ET and runoff from uncorrected and bias-corrected GCM output cannot be made for the whole GHMensemble as most of the GHMs did not produce simula-tions with the uncorrected GCM output. An indication of thesize of the effect can be found by comparing the changesfor two of the GHMs in Hagemann et al. (2011) for which,for most of the large catchments considered, the two annualmean GHM climate change signals in ET and runoff differmore than the mean signals obtained with and without biascorrection.

Future analyses of global climate change impacts to beused in, for example, the IPCC assessments should not bebased on the output of a single impact model. Well coordi-nated model intercomparison activities are not only neededfor climate models but also for the important impacts. Ourresults show a clear need for intercomparison activities suchas ISI-MIP (http://www.isi-mip.org), AgMIP (http://www.agmip.org/) and WaterMIP (Haddeland et al., 2011;http://www.eu-watch.org/watermip). A major obstacle to the useby policy makers of the results of such model intercompar-ison projects is the large amount of data and scenarios pro-duced by the different modelling exercises, which have to bereduced to a demonstrative and meaningful gist. There is aneed to develop tools and methods which allow for the quan-tification of uncertainty and assessment of robustness of cli-mate change impacts using multi-model ensembles withoutgenerating too much data, and which may be used to demon-strate the associated results in a relatively simple way.

Supplementary material related to this article isavailable online at:http://www.earth-syst-dynam.net/4/129/2013/esd-4-129-2013-supplement.pdf.

Acknowledgements.This study was supported by funding from theEuropean Union within the WATCH project (contract No. 036946).Andrew Wiltshire was partly supported by the Joint DECC/DefraMet Office Hadley Centre Programme (GS01101). The authorswould like to thank Tobias Stacke (MPI-M) for implementing sev-eral modifications into MPI-HM, Alexander Schroder (MPI-M) forsome data processing and plotting, and Richard Gilham (UKMO)for data transfer and disaggregation work regarding JULES data.The GCM data were obtained from the CERA database at the Ger-man Climate Computing Center (DKRZ) in Hamburg. Additionaldata were thankfully provided by Nathalie Bertrand from IPSL.

Edited by: A. Kleidon


http://www.isi-mip.org

http://www.agmip.org/

http://www.agmip.org/

http://www.eu-watch.org/watermip

http://www.eu-watch.org/watermip

http://www.earth-syst-dynam.net/4/129/2013/esd-4-129-2013-supplement.pdf

http://www.earth-syst-dynam.net/4/129/2013/esd-4-129-2013-supplement.pdf


The service charges for this open access publicationhave been covered by the Max Planck Society.

References

Arora, V.: The use of the aridity index to assess climate change ef-fect on annual runoff, J. Hydrol., 265, 164–177, 2002.

Berg, P., Haerter, J. O., Thejll, P., Piani, C., Hagemann, S., andChristensen, J. H.: Seasonal characteristics of the relationshipbetween daily precipitation intensity and surface temperature, J.Geophys. Res., 114, D18102,doi:10.1029/2009JD012008, 2009.

Betts, R. A., Boucher, O., Collins, M., Cox, P. M., Falloon, P. D.,Gedney, N., Hemming, D. L., Huntingford, C., Jones, C. D., Sex-ton, D. M., and Webb, M. J.: Projected increase in continentalrunoff due to plant responses to increasing carbon dioxide, Na-ture, 448, 1037–1041, 2007.

Brutsaert, W. and Parlange, M. B.: Hydrologic cycle explains theevaporation paradox, Nature, 396, 30,doi:10.1038/23845, 1998.

Doll, P., Kaspar, F., and Lehner, B.: A global hydrological modelfor deriving water availability indicators: model tuning and vali-dation, J. Hydrol., 270, 105–134, 2003.

Dosio, A. and Paruolo, P.: Bias correction of the ENSEMBLEShigh-resolution climate change projections for use by impactmodels: Evaluation on the present climate, J. Geophys. Res., 116,D16106,doi:10.1029/2011JD015934, 2011.

Ehret, U., Zehe, E., Wulfmeyer, V., Warrach-Sagi, K., and Liebert,J.: HESS Opinions “Should we apply bias correction to globaland regional climate model data?”, Hydrol. Earth Syst. Sci., 16,3391–3404,doi:10.5194/hess-16-3391-2012, 2012.

Falloon, P., Betts, R., Wiltshire, A., Dankers, R., Mathison, C., Mc-Neall, D., Bates, P., and Trigg, M.: Validation of river flows inHadGEM1 and HadCM3 with the TRIP river flow model, J. Hy-drometeorol., 12, 1157–1180, 2011.

Gosling, S. N. and Arnell, N. W.: Simulating current global riverrunoff with a global hydrological model: model revisions, vali-dation, and sensitivity analysis, Hydrol. Process., 25, 1129–1145,2011.

Gosling, S. N., Taylor, R. G., Arnell, N. W., and Todd, M. C.: Acomparative analysis of projected impacts of climate change onriver runoff from global and catchment-scale hydrological mod-els, Hydrol. Earth Syst. Sci., 15, 279–294,doi:10.5194/hess-15-279-2011, 2011.

Gosling, S. N., McGregor, G. R., and Lowe, J. A.: The benefitsof quantifying climate model uncertainty in climate change im-pacts assessment: an example with heat-related mortality changeestimates, Climatic Change, 112, 217–231,doi:10.1007/s10584-011-0211-9, 2012.

Haddeland, I., Clark, D. B., Franssen, W., Ludwig, F., Voß, F.,Arnell, N. W., Bertrand, N., Best, M., Folwell, S., Gerten, D.,Gomes, S., Gosling, S. N., Hagemann, S., Hanasaki, N., Harding,R., Heinke, J., Kabat, P., Koirala, S., Oki, T., Polcher, J., Stacke,T., Viterbo, P., Weedon, G. P., and Yeh, P.: Multi-model estimateof the global terrestrial water balance: setup and first results,J. Hydrometeorol. 12, 869–884,doi:10.1175/2011JHM1324.1,2011.

Haddeland, I., Heinke, J., Voß, F., Eisner, S., Chen, C., Hagemann,S., and Ludwig, F.: Effects of climate model radiation, humidity

and wind estimates on hydrological simulations, Hydrol. EarthSyst. Sci., 16, 305–318,doi:10.5194/hess-16-305-2012, 2012.

Haddeland, I., Biemans, H., Eisner, S., Fekete, B., Florke, M.,Hanasaki, N., Heinke, J., Ludwig, F., Schewe, J., Stacke, T.,Wada, Y., and Wisser, D.: A global multi-model view on waterbalance alterations caused by human interventions versus climatechange, P. Natl. Acad. Sci., submitted, 2013.

Haerter, J. O., Hagemann, S., Moseley, C., and Piani, C.: Cli-mate model bias correction and the role of timescales, Hy-drol. Earth Syst. Sci., 15, 1065–1079,doi:10.5194/hess-15-1065-2011, 2011.

Hagemann, S., Berg, P., Christensen, J. H., and Haerter, J.O.: Analysis of existing climate model results over Eu-rope, WATCH Technical Rep. 7,http://www.eu-watch.org/publications/technical-reports(last access: 2 May 2013), 2008.

Hagemann, S., Gottel, H., Jacob, D., Lorenz, P., and Roeckner, E.:Improved regional scale processes reflected in projected hydro-logical changes over large European catchments, Clim. Dynam.,32, 767–781,doi:10.1007/s00382-008-0403-9, 2009.

Hagemann, S., Chen, C., Haerter, J. O., Gerten, D., Heinke,J., and Piani, C.: Impact of a statistical bias correction onthe projected hydrological changes obtained from three GCMsand two hydrology models, J. Hydrometeorol., 12, 556–578,doi:10.1175/2011JHM1336.1, 2011.

Hansen, J. W., Challinor, A., Ines, A., Wheeler, T., and Moronet, V.:Translating forecasts into agricultural terms: advances and chal-lenges, Climate Res., 33, 27–41, 2006.

Koster, R. D. and Milly, P. C. D.: The interplay between transpira-tion and runoff formulations in land surface schemes used withatmospheric models, J. Climate, 10, 1578–1591, 1997.

Masson, D. and Knutti, R.: Climate model genealogy, Geophys.Res. Lett., 38, L08703,doi:10.1029/2011GL046864, 2011.

Meehl, G. A., Boer, G. J., Covey, C., Latif, M., and Stouffer, R.J.: The Coupled Model Intercomparison Project (CMIP), B. Am.Meteorol. Soc., 81, 313–318, 2000.

Meehl, G. A., Covey, C., Delworth, T., Latif, M., McAvaney,B., Mitchell, J. F. B., Stouffer, R. J., and Taylor, K. E.: TheWCRP CMIP3 multi-model dataset: A new era in climate changeresearch, B. Am. Meteorol. Soc., 88, 1383–1394, 2007.

Nakicenovic, N., Alcamo, J., Davis, G., de Vries, B., Fenhann,J., Gaffin, S., Gregory, K., Grubler, A., Jung, T. Y., Kram, T.,La Rovere, E. L., Michaelis, L., Mori, S., Morita, T., Pepper,W., Pitcher, H., Price, L., Raihi, K., Roehrl, A., Rogner, H.-H.,Sankovski, A., Schlesinger, M., Shukla, P., Smith, S., Swart, R.,van Rooijen, S., Victor, N., and Dadi, Z.: IPCC Special Report onEmissions Scenarios, Cambridge University Press, Cambridge,UK and New York, NY, USA., 2000.

Nijssen, B., O’Donnell, G. M., Hamlet, A. F., and Lettenmaier, D.P.: Hydrologic sensitivity of global rivers to climatic change, Cli-matic Change, 50, 143–175, 2001.

Oki, T., Agata, Y., Kanae, S., SaruhashiI, T., and Musiake, K.:Global water resources assessment under climatic change in2050 using TRIP, Water Resources Systems – Water availabil-ity and global change, Proceedings of symposium HS02a heldduring IUGG2003 at Sapporo, July 2003, IAHS Publ., 280, 124–133, 2003.

Osborne, T., Rose, G. and Wheeler, T.: Variation in the global-scaleimpacts of climate change on crop productivity due to climatemodel uncertainty and adaptation, Agr. Forest Meteorol., 170,


http://dx.doi.org/10.1029/2009JD012008

http://dx.doi.org/10.1038/23845

http://dx.doi.org/10.1029/2011JD015934

http://dx.doi.org/10.5194/hess-16-3391-2012



http://dx.doi.org/10.1007/s10584-011-0211-9

http://dx.doi.org/10.1007/s10584-011-0211-9

http://dx.doi.org/10.1175/2011JHM1324.1




http://www.eu-watch.org/publications/technical-reports

http://www.eu-watch.org/publications/technical-reports

http://dx.doi.org/10.1007/s00382-008-0403-9


http://dx.doi.org/10.1029/2011GL046864


183–194.doi:10.1016/j.agrformet.2012.07.006, 2013.Piani, C. and Haerter, J. O.: Two dimensional bias correction of tem-

perature and precipitation copulas in climate models, Geophys.Res. Lett.,doi:10.1029/2012GL053839, in press, 2012.

Piani, C., Haerter, J. O., and Coppola, E.: Statistical bias correctionfor daily precipitation in regional climate models over Europe,Theor. Appl. Climatol., 99, 187–192, 2010a.

Piani, C., Weedon, G. P., Best, M., Gomes, S., Viterbo, P., Hage-mann, S., and Haerter, J. O.: Statistical bias correction of globalsimulated daily precipitation and temperature for the applicationof hydrological models, J. Hydrol., 395, 199–215, 2010b.

Prudhomme, C., Giuntoli, I., Robinson, E. L., Clark, D. B., Ar-nell, N. W., Dankers, R., Fekete, B., Franssen, W., Gerten, D.,Gosling, S. N., Hagemann, S., Hannah, D. M., Kim, H., Masaki,Y., Satoh, Y., Stacke, T., Wada, Y., and Wisser, D.: Drought inthe 21st century: a multi-model ensemble experiment to assessglobal change, quantify uncertainty and identify “hotspots”, P.Natl. Acad. Sci., submitted, 2013.

Randall, D. A., Wood, R. A., Bony, S., Colman, R., Fichefet, T.,Fyfe, J., Kattsov, V., Pitman, A., Shukla, J., Srinivasan, J., Stouf-fer, R. J., Sumi, A., and Taylor, K. E.: Climate Models and TheirEvaluation. In: Climate Change 2007: The Physical Science Ba-sis. Contribution of Working Group I to the Fourth AssessmentReport of the Intergovernmental Panel on Climate Change, editedby: Solomon, S., Qin, D., Manning, M., Chen, Z., Marquis, M.,Averyt, K. B., Tignor, M., and Miller, H. L., Cambridge Uni-versity Press, Cambridge, United Kingdom and New York, NY,USA, 2007.

Renner, M. and Bernhofer, C.: Applying simple water-energy bal-ance frameworks to predict the climate sensitivity of streamflowover the continental United States, Hydrol. Earth Syst. Sci., 16,2531–2546,doi:10.5194/hess-16-2531-2012, 2012.

Roderick, M. and Farquhar, G.: A simple framework for relat-ing variations in runoff to variations in climatic conditionsand catchment properties, Water Resour. Res., 47, W00G07,doi:10.1029/2010WR009826, 2011.

Rojas, R., Feyen, L., Dosio, A., and Bavera, D.: Improving pan-European hydrological simulation of extreme events through sta-tistical bias correction of RCM-driven climate simulations, Hy-drol. Earth Syst. Sci., 15, 2599–2620,doi:10.5194/hess-15-2599-2011, 2011.

Scanlon, T. S., Caylor, K. K., Levin, S. A., and Rodriguez-Iturbe,I.: Positive feedbacks promote power-law clustering of Kalaharivegetation, Nature, 449, 209–212, 2007.

Schewe, J., Heinke, J., Gerten, D., Haddeland, I., Arnell, N. W.,Clark, D., Dankers, R., Eisner, S., Fekete, B., Colon-Gonzalez,F. J., Gosling, S., Kim, H., Liu, X., Masaki, Y., Portmann, F.T., Satoh, Y., Stacke, T., Tang, Q., Wada, Y., Wisser, D., Al-brecht, T., Frieler, K., Piontek, F., Warszawski, L., and Kabat, P.:Multi-model assessment of water scarcity under climate change,P. Natl. Acad. Sci., submitted, 2013.

Seneviratne, S. I., Corti, T., Davin, E. L., Hirschi, M.,Jaeger, E. B., Lehner, I., Orlowsky, B., and Teuling,A. J.: Investigating soil moisture–climate interactions in achanging climate: A review, Earth-Sci. Rev., 99, 125–161,doi:10.1016/j.earscirev.2010.02.004, 2010.

Sharma, D., Das Gupta, A., and Babel, M. S.: Spatial disaggrega-tion of bias-corrected GCM precipitation for improved hydro-logic simulation: Ping River Basin, Thailand, Hydrol. Earth Syst.Sci., 11, 1373–1390,doi:10.5194/hess-11-1373-2007, 2007.

Smakhtin, V., Revenga, C., and Doll, P.: A Pilot Global assessmentof environmental water requirements and scarcity, Water Int., 29,307–317, 2004.

Solomon, S., Qin, D., Manning, M., Marquis, M., Averyt, K., Tig-nor, M. M. B., Miller Jr., H. L., and Chen, Z. (Eds.): ClimateChange 2007: The Physical Science Basis, Cambridge Univer-sity Press, Cambridge, 996 pp., 2007.

Sperna Weiland, F. C., van Beek, L. P. H., Kwadijk, J. C. J.,and Bierkens, M. F. P.: Global patterns of change in dischargeregimes for 2100, Hydrol. Earth Syst. Sci., 16, 1047–1062,doi:10.5194/hess-16-1047-2012, 2012.

Weedon, G. P., Gomes, S., Viterbo, P., Shuttleworth, W. J., Blyth,E., Osterle, H., Adam, J. C., Bellouin, N., Boucher, O., andBest, M.: Creation of the WATCH forcing data and its use toassess global and regional reference crop evaporation over landduring the twentieth century, J. Hydrometeorol., 823–848, 12,doi:10.1175/2011JHM1369.1, 2011.

Wood, A. W., Leung, L. R., Shridhar, V., and Lettenmaier, D. P.:Hydrologic implications of dynamical and statistical approachesto downscaling climate outputs, Climatic Change, 62, 189–216,2004.


http://dx.doi.org/10.1016/j.agrformet.2012.07.006

http://dx.doi.org/10.1029/2012GL053839


http://dx.doi.org/10.1029/2010WR009826



http://dx.doi.org/10.1016/j.earscirev.2010.02.004




climate change impact on available water resources...

Documents