Technical Report No. 45 Projected hydrological changes in the 21st century and related uncertainties obtained from a multi-model ensemble Author names: Chen, C., Hagemann, S., Clark, D., Folwell, S., Gosling, S., Haddeland, I., Hanasaki, N., Heinke, J., Ludwig, F., Voβ, F. and Wiltshire, A Date: 31.07.2011 WATCH is an Integrated Project Funded by the European Commission under the Sixth Framework Programme, Global Change and Ecosystems Thematic Priority Area (contract number: 036946).
Transcript
Technical Report No. 45
Projected hydrological changes in the 21st century and
related uncertainties obtained from a multi-model ensemble
Author names: Chen, C., Hagemann, S., Clark, D., Folwell, S., Gosling, S., Haddeland, I., Hanasaki, N., Heinke, J., Ludwig, F., Voβ, F. and Wiltshire, A Date: 31.07.2011 WATCH is an Integrated Project Funded by the European Commission under the Sixth Framework Programme, Global Change and Ecosystems Thematic Priority Area (contract number: 036946).
Technical Report No. 45 - 2 -
The WACH project started 01/02/2007 and will continue for 4 years.
Title: Projected hydrological changes in the 21st century and related uncertainties obtained from a multi-model ensemble
Authors:
Cui Chen1, Stefan Hagemann1, Douglas Clark2, Sonja Folwell3, Simon Gosling4, Ingjerd Haddeland5, Naota Hanasaki6, Jens Heinke7, Fulco Ludwig8, Frank Voβ9, F. and Andy Wiltshire2
Organisations: 1 Max Planck Institute for Meteorology, Hamburg, Germany 2 UK Met Office, United Kingdom 3 Centre for Ecology and Hydrology, Wallingford, United Kingdom 4 School of Geography, University of Nottingham, United Kingdom 5 Norwegian Water Resources and Energy Directorate, Oslo, Norway 6 National Institute for Environmental Studies, Tsukuba, Japan 7 Potsdam Institute for Climate Research, Potsdam, Germany 8 Wageningen University and Research Centre, Wageningen, Netherlands 9 Center for Environmental Systems Research, University of Kassel, Kassel, Germany.
Submission date: 31.07.2011
Function: This report is an output from Work Block 3; task 3.1.3. and 3.1.4
Deliverable WATCH deliverable 3.1.4 and 3.1.5
Summary 21st century climate change is likely to have a significant impact on the hydrological cycle and thus has the potential to impose additional water stress in several regions. Thus, this study focuses on the assessment of the implications of climate change for global hydrological regimes and related water resources states for the 21st century. Different climate and hydrological models show quite different projected changes with a large variation of uncertainty within the climate – hydrology modelling chain. Therefore, multiple climate and hydrological models were used within the European project "Water and Global Change" (WATCH) to assess the hydrological response to climate change and to project the future state of global and large scale water resources. Climate model data were taken from projections of three coupled atmosphere-ocean General Circulation Models (GCMs) (ECHAM5/MPIOM, CNRM-CM3, LMDZ-4) following the A2 and B1 emission scenarios. Due to the systematic errors of climate models, their output has been corrected with a statistical bias correction method and then the output was used directly to force global hydrological models (GHMs) (MPI-HM, LPJmL, WaterGAP, VIC, MacPDM, H08, GWAVA, JULES) to calculate the corresponding changes in hydrological fluxes. The analyses focus on the changes in the hydrological characteristics for twelve large, continental river basins without taking into account direct anthropogenic influences in the hydrological simulations. The hydrological cycle was evaluated and multiple-model based projections were analysed for the terrestrial components of the hydrological cycle focusing on the time period of 2071-2100. Global maps are constructed to identify regions where the water cycle and associated water resources are significantly impacted by climate change, and which regions are vulnerable to these changes in terms of e.g. water availability. The uncertainties due to the choice of GCM and GHM are also assessed.
Technical Report No. 45 - 3 -
Evaluation of projected hydrological changes in the 21st century obtained from a multi-model ensemble
1. Introduction
Global warming leads to changes of water resources distribution over several regions and the global and regional hydrological cycles have been greatly influenced by climate change in the past century (Brutsaert and Parlange, 1998; Scanlon et al., 2007; IPCC, 2007) Following the green house emission scenarios for the 21st century (Nakićenović et al., 2000), climate change will cause increased temperatures and changes in precipitation that vary globally. Hydrological models have been used widely for water resource assessments, especially for studying the impact of climate change. A lot of studies have tried to assess the impact of climate change on the past and future water studies for some global locations using certain hydrological models (Christensen, 2004; Fu 2004; Gosling and Arnell, 2010; Oki, 2003; Wang, 2009; Nijssen et al., 2001; Doell et al., 2003). But in each off these studies only one hydrology model was applied to consider climate change impacts on hydrology. The recent study of Haddeland et al. (2011a) suggests that using multiple impact models (Global Hydrology Models (GHMs) or Land Surface Models (LSMs)) is necessary for climate change impact studies in the future period. In Haddeland et al. (2011a) the results show that the differences between models are a major source of uncertainty. Climate change impact studies thus need to use not only multiple climate models, but also some other measure of uncertainty, e.g. multiple impact models. Here, in this study, climate projections from three state-of-the-art coupled atmosphere-ocean general circulation models (GCMs) and eight GHMs are used to assess the hydrologic response to climate change of global river basins. These regions represent a range of climate and geographic regions globally. Three GCMs (ECHAM5/MPIOM, CNRM-CM3 and LMDZ-4) are considered in this study. The eight GHMs that calculated the future water flux changes are MPI-HM, LPJmL, WaterGAP, VIC, MacPDM, H08, GWAVA and JULES. The climate system is less predictable for its natural climate variability. Each projection introduces a certain magnitude of uncertainty and the causes are different. It is well known that these cannot be directly derived from GCM simulations of future climate, which are significantly affected by errors, and results from a forced hydrological simulation will be unrealistic and of little use (Sharma et al. 2007; Hansen et al. 2006). Therefore, the statistical bias correction method of Piani et al. (2010) was applied to the GCM data for the control period from 1960-2000 and the future period from 2001-2100.
2. Models and Methods GCMs Three coupled atmosphere-ocean General Circulation Models (GCMs) are used in this study to provide quantitative estimates of future climate projections, particularly at continental and larger scales. Table 2 shows the details of the model characteristics. For the past climate, observed concentrations of greenhouse gases and aerosols were prescribed. For the future climate, these concentrations were prescribed according to the two IPCC scenarios A2 and B1 (IPCC Special Report on Emission Scenarios, Nakićenović et al., 2000). Some additional information about these GCMs used for this study is described below:
1. ECHAM The coupled atmosphere/ocean GCM ECHAM5/MPIOM (denoted as ECHAM5 henceforth, Roeckner et al. 2003, Jungclaus et al. 2006) of the Max Planck Institute for Meteorology (MPI-M), has been used to generate an ensemble of climate simulations of which three are for the past century covering the period 1860-2000 and nine are for the future climate from 2001-2100. In this study, the third initial condition
Technical Report No. 45 - 4 -
ensemble member is used. The coupled model was run at T63 (about 1.9° or 200 km grid size) horizontal resolution and 31 vertical levels in the atmosphere, and about 1.5° horizontal resolution and 40 vertical layers in the oceans. For the past climate (1860-2000), observed concentrations of CO2, CH4, N2O, CFCs, O3 and sulphate aerosols were prescribed, thereby considering the direct and first indirect aerosol effect.
2. IPSL The coupled model IPSL (Hourdin et al. 2006; Madec et al. 1998; Fichefet and Morales Maqueda 1997; Goosse and Fichefet 1999) of Institute Pierre Simon Laplace was run at 2.5° x 3.75° horizontal resolution with 19 vertical levels in the atmosphere, and on a quasi-isotrope tri-polar grid with 31 vertical levels in the ocean (2 poles in the northern hemisphere, one over 0Canada and the other over Siberia), thereby using a 2° resolution Mercator grid with enhanced meridional resolution in the vicinity of the equator and in Mediterranean and Red seas (1°).
3. CNRM The coupled GCM CNRM (Déqué et al. 1994; Déqué and Piedelièvre 1995; Royer et al. 2002; Madec et al. 1998; Salas-Mélia 2002) of Centre National de Recherches Météorologiques, Météo-France contains triangular truncation T63 with a linear reduced Gaussian grid equivalent to T42 quadratic grid horizontal resolution (2.8° ~ 300 km) and a progressive hybrid sigma-pressure vertical coordinate with 45 layers in the atmosphere. The ocean has a resolution of about 2° in longitude, a resolution varying in latitude from near 0.5° at the equator to roughly 2° in Polar regions, and 31 vertical levels. The distribution of marine, desert, urban aerosols, and sulphate aerosols were specified, whereas for aerosols, only the direct effect of anthropogenic sulphate aerosols was taken into account. GHMs Global hydrological models (GHMs) simulate the land surface hydrologic dynamics of large scale river basins. In this study, eight global hydrological models are used. Table 1(Haddeland et al., 2011a) gives an overview for the model characteristics and the physical processes. In the following, a detailed description about each model is given.
1. MPIHM MPI-HM consists of the Simplified Land surface (SL) scheme (Hagemann and Dümenil Gates 2003), which computes vertical water fluxes, and the Hydrological Discharge (HD) model (Hagemann and Dümenil 1998) that globally simulates the lateral freshwater fluxes at the land surface. The latter is a state-of-the-art discharge model that is applied and validated on the global scale, and it is also part of the coupled atmosphere-ocean GCM ECHAM5/MPIOM. The SL scheme incorporates the main components of the hydrological cycle at the land surface and primarily uses relations that are functions of temperature and precipitation. The soil is represented by a single soil layer, and major process representations comprise the separation of throughfall into surface runoff and infiltration according to the improved Arno scheme (Hagemann and Dümenil Gates 2003), the separation of precipitation into rain and snow according to Wigmosta et al. (1994), snowmelt using a daily degree formula according to Bergström (1992), and potential evapotranspiration using the Thornthwaite formula (Chebotarev 1977) that is purely based on temperature. For the current study, slight modifications compared to Hagemann and Dümenil Gates (2003) were implemented. Land sea mask, glacier mask, total (field capacity) and plant-available soil water capacity were taken from the LSP2 dataset (Hagemann 2002). Lake and wetland fractions were obtained from the global lake and wetlands database (Lehner and Döll 2004), and the lake and wetland evaporation at the potential rate was modified in both components of MPI-HM (T. Stacke, pers. comm.).
Technical Report No. 45 - 5 -
2. H08 H08 is a global water resources model which deals with both natural hydrological processes and major human activities related to water use. Detailed description of the model formulations and results of validation can be found in Hanasaki et al. (2008a, b). H08 consists of six sub-models: land surface hydrology, river routing, crop growth, reservoir operation, water withdrawal, and environmental flow requirement sub models. The land surface hydrology sub model is based on the bucket model (Manabe, 1969; Robock et al., 1995). Simple sub surface flow process which is similar to Gerten et al. (2004) is included. This sub model assumes uniform vegetation cover and soil properties all over the world; exceptionally the parameters for sub surface flow are set by Köppen-Geiger climatic zones. The river routing sub model is identical to Oki et al. (1999). The effective flow velocity is set at globally uniform 0.5 m s-1.
3. LPJmL
The global ecohydrological model LPJmL (Bondeau et al. 2007, Rost et al. 2008) simulates at 0.5° resolution the growth, production and phenology of natural and agricultural vegetation in direct coupling with the carbon and water cycling. Atmospheric CO2 concentration is simulated to affect plant transpiration and biomass production via both physiological and structural plant responses (Gerten et al. 2004). The establishment and dynamic distribution of natural vegetation and the seasonal phenology of natural and agricultural vegetation are simulated based on long-term average climate. The model distinguishes two soil layers with fixed thickness (upper, 50 cm; lower, 100 cm). Soil moisture of each layer is updated daily, according to the balance between the amount of water infiltrating into the soil (throughfall minus surface runoff) and that removed from the soil layers through sub-surface runoff, percolation, soil evaporation and plant transpiration. Evapotranspiration is calculated from radiation and temperature using the Priestley-Taylor formulation (with a modification for plant transpiration to mimic boundary-layer effects). Runoff is generated when field capacity of the upper and/or lower soil layer is surpassed. Snowmelt is modeled following a degree-day approach. For details on the hydrological scheme see Gerten et al. (2004) and Rost et al. (2008); for the most recent description of the here used model version and the land use input datasets see Gerten et al. (2011).
4. MacPDM Macro-scale-Probability-Distributed Moisture model (MacPDM) (Arnell, 1999 and 2003, Gosling and Arnell, 2010) is a GHM designed to simulate river runoff on a gridded basis across a large spatial domain. The model is driven by inputs of either monthly climate data (some variables are disaggregated within the model to a daily time step), or daily data. The model is a daily water balance model, which assumes that the soil moisture storage capacity varies statistically across the catchment: all other catchment properties, and climatic inputs, are assumed constant across the catchment. Model parameters are derived, directly or indirectly, from digital spatial data bases (Gosling, et al 2010).
5. VIC Variable infiltration capacity (VIC) (Liang et al 1994; Nijssen et al., 1997, 2001) model is a land surface model based on the fundamental hydrological processes which include interaction of the atmosphere with underlying vegetation and soils, where the dynamic water and energy fluxes are considered. One distinguishing characteristic of the VIC model is that it represents the sub-grid spatial heterogeneity of precipitation with sub-grid spatial variability of soil infiltration capacity. The land surface is divided into different land cover types horizontally, whereas soils are partitioned into three vertical layers. Quick bare soil evaporation following short-duration summer rainfall events happens in the top-most soil layer; the upper soil layer is designed to represent the dynamic change of soil moisture and the production of run-off in response to rainfall events. Soil moisture changes and contributions to baseflow mainly occur in the lower soil layer of the model. A variable infiltration curve is used to represent the sub-grid variability of soil infiltration capability under different land cover and soil types (Zhao et al., 1980a, 1980b). Three
Technical Report No. 45 - 6 -
types of evaporation are considered in the model: evaporation from the canopy layer of each vegetation class, transpiration from each of the vegetation classes, and bare soil evaporation. Evapotranspiration from each vegetation type is calculated using the Penman–Monteith formulation (Liang et al., 1994). Total evapotranspiration over a grid cell is computed as the sum of the above components, weighted by the respective surface cover area fractions. For more information on the VIC model and its application in different catchments, the reader is referred to Su et al. (2008), Xie et al. (2007), and the VIC website, http://www.hydro.washington.edu/ (Wang, et al 2010).
6. WaterGAP Water - Global Assessment and Prognosis (WaterGAP) (Alcamo, 2003; Döll, 2003) has been developed at the Center for Environmental Systems Research at the University of Kassel in Germany. WaterGAP comprises two main components, a Global Hydrology Model and a Global Water Use Model. The WaterGAP Global Hydrology Model calculates a daily vertical water balance for both the land area and the open water bodies at each of the 0.5° cells. The vertical water balance for the land fraction in a cell consists of a canopy water balance and a soil water balance. These are calculated as functions of land cover, soil water capacity, and daily climate variables (i.e. temperature, radiation, and precipitation). The canopy water balance determines which part of the precipitation is intercepted by the canopy and directly evaporates, and which part reaches the soil as throughfall. At this level, the soil water balance subdivides the throughfall into evapotranspiration and total runoff. A different vertical water balance for open water bodies is applied to lakes, reservoirs, and wetlands (based on a global 1-minute wetlands, lakes, and reservoirs map by Lehner and Döll (2001)), where the runoff is computed as the difference between precipitation and open water evaporation. The sum of the runoff produced within a cell and the discharge flowing into a cell from upstream is transported through a series of storages that represent groundwater, lakes, reservoirs, wetlands, and the river itself. Finally, the total cell discharge is routed to the next downstream cell following a global drainage direction map (Döll and Lehner 2002) to compute river discharge (www.usf.uni-kassel.de).
7. Gwava Global Water Availability Assessment model (Gwava) a hydrological model which incorporates additional water resource components such as reservoirs, abstractions, and water transfers that modify water quantity and flow regime. For more information about the Gwava see slso: http://www.ceh.ac.uk/sci_programmes/Water/GWAVA.html.
8. JULES JULES is the Joint UK Land Environment Simulator. JULES has a tiled model of sub-grid heterogeneity with separate surface temperatures, shortwave and long-wave radiative fluxes, sensible and latent heat fluxes, ground heat fluxes, canopy moisture contents, snow masses and snow melt rates are computed for each surface type in a grid-box. For more information about JULES model, see also: http://www.jchmr.org/jules/. Statistical bias-correction method It is well known that climate model output data contain systematic errors and can not be used directly in the hydrological simulations. Thus, their output data were corrected by using a statistical bias-correction method. The statistical bias correction function (Piani et al. 2010) was applied to the control period from 1960-1999 as well as to two scenario periods (A2, B1) from 2000-2100 of the GCM simulations. The statistical bias correction methodology for global climate simulations was applied to daily land precipitation and mean, minimum and maximum daily land temperatures. The bias correction is based on a fitted histogram equalization function. This function is defined daily, as opposed to earlier published versions in which they were derived yearly or seasonally at best, while conserving properties of
Technical Report No. 45 - 7 -
robustness and eliminating unrealistic jumps at seasonal or monthly transitions. Bias corrections factors are derived from 1960 to 1999 from observed and simulated data and then applied to 1960-2100 simulations. (Piani et al, 2010). There are several studies done for the assessment of this method in the hydrological projections. The impact of bias correction on the global hydrological cycle was studied. It is found that the bias correction has an impact on the climate change signal for specific locations and months, thereby adding another level of uncertainty in the modelling chain from the GCM to the simulated changes calculated by the hydrology models. This uncertainty may be of the same order of magnitude as uncertainty related to the choice of the GCM or GHM (Hagemann, et al 2011). In Hagemann et al. (2011) the benefits of bias correction are weighted against the apparent addition of uncertainty in the resulting simulated future hydrological fields due to bias correction parameter choice. Bias correction of projected scenario forcing fields is known to affect not only the projected hydrological fields, but also the climate change signal (Haerter et al., 2011). Contribution of bias correction and other sources to the uncertainty in the projected hydrological cycle was studied as well. The comparative analyses of uncertainties in global hydrological model simulations showed that uncertainties due to the choice of forcing global climate model, future CO2 emission scenario and bias correction calibration period are inter-compared along with the inter-annual variability. Discharge and integrated runoff and precipitation over several large scale catchment areas, representative of the entire globe, are considered. Results are similar for all catchments, all hydrological variables and for most periods of the year. The choice of different decadal periods over which to conduct the calibration of the bias correction parameters is only a minor contributor to the total uncertainty, while other contributors play more significant roles. (Chen, et al 2011, submitted)
3. Simulation setup
In this study, to evaluate the projected hydrological cycle obtained from the multi-model ensemble, the spread of the mean and the uncertainty distribution are calculated. For each hydrological variable, 24 different datasets of the same time series are obtained from 8 GHMs which are forced by 3 GCMs with A2 and B1 emission scenario, respectively. The ensemble mean of the hydrological variables are calculated in the control period from 1971-2000 and in future period from 2071-2100. The changes in the future period relative to the control period are expressed by absolute changes and relative changes given in percentage. The uncertainty distribution is calculated from the normalized standard deviation (or CV, coefficient of variation). The relative differences between models can be expressed by the coefficient of variation. The hydrological variables such as total runoff, discharge and evapotranspiration are explored in this study. Water resources depend heavily on the available total runoff (surface runoff and subsurface runoff) and associated discharge.
4. Results Study area 12 global large scale basins (Fig 1) are mainly considered in this study as they represent different climate zones varying globally. These basins are: Amazon, Amur, Mackenzie, Lena, Congo, Danube, Ganges, Mississippi, Murray, Nile, Parana and Yangtze. In the following, we concentrate on the results obtained from simulations forced by the GCMs under the assumption of the A2 scenario. According to the climate zone definition of Koeppen (1923), these 12 basins are categorised into the following climate zones: Climate type A - tropical: Amazon, Congo, Ganges; Climate type B - dry: Nile; Climate type C – temperate: Danube, Mississippi, Murray, Parana, Yangtze; Climate type D – continental: Lena, Amur, Mackenzie. The response of hydrological dynamics to climate change in the climate zone behaves differently.
Technical Report No. 45 - 8 -
Annual mean
1. Precipitation Figure 2 shows the global map of annual ensemble mean precipitation from 3 GCMs - ECHAM, IPSL and CNRM in the control period 1971-2000, the precipitation change projected by the 3 GCMs in the future period from 2071-2100 compared to the control period, and standard deviation of changes from the 3 GCMs. In the baseline simulation of precipitation, basically the 3 GCMs agree quite well with each other due to the bias correction, except for the dry region of North Africa. In this dry area, the precipitation is very small during the most of the time period so that slight absolute deviations from these small numbers can lead to large relative changes. If we look at the future precipitation change, most of the area located in the higher latitudes and some part of the areas in the middle latitudes will get increase of precipitation. Part of the Middle East, south part of Europa, south part of North America, north part of Africa and the south part of Africa including south east part of Australia will receive less precipitation.
2. Runoff Figure 3 shows the multi-model annual changes of precipitation and temperature in the near future (2020-2050, centred at 2035) and far future (2071-2100, centred at 2085) compared to the control period (1971-2000). The temperature is projected to continuously increase for all the 12 river basins under both A2 and B1 scenarios. Several river basins such as Amazon, Amur, Congo, Lena, Mackenzie and Nile will experience a continuous increase of precipitation and Danube and Murray a continuous decrease of precipitation in the 21st century. For Ganges and Yangtze, the A2 scenario projects a decrease of precipitation in the near future but increase in the far future, but in the B1 scenario a continuous increase of precipitation is projected. Figure 4 shows the ensemble mean runoff from the 24 simulations (8 GHMs and 3 GCMs with A2 scenario) for the control period 1971-2000 and the projected ensemble mean runoff changes in the future period from 2071-2100 compared to control period. In addition, it shows the associated standard deviations of mean runoff and projected runoff change yielded by the 8 GHMs and by the 3 GCMs. The east part of Australia, south part of Africa, south part of United States, north east part of South America, south part of Europe, and large part of the Middle East will experience a decrease in runoff in the future compared to the control period. How strongly these areas would have to suffer from problems, such as water scarcity in the 21st century, can not be quantified from the naturalized simulations alone. Climate change has been identified to have a major influence on basin water balances. However, land use and water use practices also play role in the assessment whether and how strongly human societies are affected in those changing regions. For an estimation of combined anthropogenic and climate change effects, water use and further direct anthropogenic impacts on hydrology have to be taken into account. Generally, the uncertainty due to the choice of GCM is comparable to the choice of GHM for runoff simulation. The uncertainty of the future runoff change is larger than the uncertainty of the runoff in the control period.
3. Evapotranspiration Figure 5 shows ensemble mean evapotranspiration (ET) from 24 simulations (8 GHMs and 3 GCMs with A2 scenario) in the control period from 1971-2000 and the projected ET changes (in 2071-2100 compared to 1971-2000) together with the associated standard deviations yielded by the 8 GHMs and the 3 GCMs. Generally, the uncertainty due to the choice of GCM is smaller than due to the choice of GHM for the ET simulation. The uncertainty of the future ET change is larger than the uncertainty of the ET in the control period. Monthly mean
Technical Report No. 45 - 9 -
1. Meteorological forcing
Figure 6 shows the monthly mean precipitation over several river basins in the control period and its change in the near and far future for the A2 scenario. Due to the bias correction of daily precipitation, the 3 GCMs agree quite well with the WFD (Watch Forcing Data, Weedon et al., 2011) results in the control period (see also Hagemann et al. 2011). Analogous, the temperature also shows good agreement between the models and WFD (not shown). From this point of view, the 3 GCMs give robust meteorological input into the hydrological models for the baseline period. How much the uncorrected GCM variables play a role for the hydrological simulations is currently being investigated in a study of Haddeland et al. (2011b).
2. Discharge Figure 7 shows the monthly mean discharge in the control period (1971-2000) and the changes in the future period from 2071-2100 compared to the control period for several river basins. The 8 GHMs can give quite different results compared to the observation. For Amazon and Danube catchment, most of the models show a similar seasonality but the timing and value of the monthly peak are different from model to model. For the Nile, the GHMs deviate largely from the observations. This can be explained by fact that the naturalized hydrological simulations do not include direct anthropogenic effects on hydrology, e.g. irrigation and dams, whereas the Nile river is heavily influenced by these effects. In high latitudinal continental areas, such as the Lena basin, there is a projected increase of discharge during most time in a year except for a slight decrease in the summer season. For many tropical areas (except Amazon, see below), such as the Nile basin, most of the models show a projected increase of discharge in the whole year except MPI-HM. This can be explained by the potential evapotranspiration scheme in MPI-HM model, which depends only on temperature. It seems that especially in tropical and subtropical areas, this formulation reacts quite sensitively to future increases in temperature, and thereby causing overly large increases in total evapotranspiration under global warming conditions (see Hagemann et al. 2011). The too enhanced increase of ET leads to less discharge according to the closure of the water balance. For the Amazon basin, the future changes have a seasonal cycle. Here, most models show an increase of discharge in the boreal winter and spring time, and decrease in the summer time. This corresponds to a future moistening in the wet season and a drying in the dry season. Again, MPI-HM likely underestimates the discharge in the future period due to its too enhanced increase in ET. In the temperate areas, such as the Danube basin, a similar seasonal behaviour as for the Amazon is found with an increase of discharge in the wet winter/spring time, but a pronounced decrease in the dry summer season.
3. Total runoff The total runoff is the sum of surface runoff and subsurface runoff. For the multi-model ensemble monthly mean in the control period it is found that the uncertainties due to the choice of the GCM are comparable to the uncertainties due to the choice of the GHM. If the absolute and relative changes in the future period compared to the control period are considered, the uncertainties due to the choice of the GHM are generally smaller than those due to the choice of the GCM. Here, the choice of the GCM can lead to quite different change signals for some river basins. In order to illustrate this behaviour, several rivers are selected in figure 8 that shows the multi-model ensemble mean during the control period, the absolute and relative projected A2 changes as well as the GHM model differences expressed as standard deviation using each GCM, separately. The red, blue and green lines represent the multi-model ensemble monthly mean values (solid lines) and standard deviations (dashed lines) of total runoff calculated from the 8 GHM simulations with ECHAM, IPSL and CNRM forcing, respectively. For the Parana basin, runoff is projected to increase using the ECHAM and CNRM forcing, but it shows a decrease signal for the IPSL forcing. For the Yangtze, the projected changes in runoff differ largely between the three GCMs. Differences in the projected change signal due to the choice of the GCM
Technical Report No. 45 - 10 -
seem to be similar, independent of whether absolute or relative changes are considered. The uncertainties due to the different GHMs are smaller for the relative changes than for absolute changes.
4. Evapotranspiration Opposite to runoff, for evapotranspiration, the uncertainty due to the choice of GCM is smaller than due to the choice of GHM. This is true for the ensemble monthly mean in the control period as well as for the absolute and relative changes in the future. Here, the uncertainties for the absolute and relative changes due to the choice of GHM can be even much larger than the ones due to the choice of GCM. For example, figure 9 shows the multi-model ensemble monthly means of ET during the control period and the projected A2 changes in the future together with the uncertainties due to the choice of GHM expressed as standard deviation for the rivers basins Murray, Parana and Yangtze. The large differences in GHM simulated ET can be explained by the different GHM schemes to calculate ET. These schemes strongly vary between the different GHMs (see table1). This leads to larger differences in ET for the control period and to larger uncertainties in the future period as well. Changes in the available water resources From these results obtained by the 8 GHMs and 3 GCMs, catchment based maps of changes in available water resources can be derived to identify areas that are vulnerable to projected climate changes with regard to water availability. In this respect available water resources are defined for various catchments around the globe as the total annual runoff (R) minus the mean environmental water requirements. Adopting results of Smakhtin et al. (2004), environmental water requirements (EWR) for a specific catchment can be roughly approximated by 30% of the total annual catchment runoff. Let us assume that these requirements obtained from the current climate simulations (1971-2000) will not significantly change until the end of the 21st century, and then the projected change in available water resources (∆AW) can be determined as:
∆AW = (RScen – EWR) – (RC20 – EWR) / (R C20 – EWR) = (RScen – RC20) / (R C20 – EWR) Here, RC20 and RScen are the mean annual runoff for the current climate (1971-2000) and future scenario periods, respectively, and EWR = 0.3 RC20. Figure 10 shows ∆AW for the period 2071-2100 according to the A2 scenario for a selection of about 90 catchments around the globe. For Fig. 10 – upper left panel, ∆AW was calculated from the multi-model ensemble mean runoff values averaged over the simulations from the 8 GHMs and the 3 GCMs, i.e. 24 simulations for the current and future climate each. The other panels show ∆AW calculated for each GCM separately, but using the averaged runoff from the simulations of the 8 GHMs. Several regions can be identified were the available water resources are expected to significantly decrease (more than 10%). These regions comprise Central, Eastern and Southern Europe, the catchments of Euphates/Tigris in the Middle East, Missisippi in North America, Xun Jiang in Southern China, Murray in Australia, and Okawango and Limpopo in Southern Africa. Here, the projections based on the different GCMs largely agree. But giving the large uncertainty induced by the choice of a GCM, it can not be neglected that some regions might be affected by a significant future reduction in available water resources if this is even projected based on only one GCM. Here, especially the following catchments can be noted: Parana (more than -50% for IPSL) and Uruguay (more than -20% for IPSL) in South America, Orange (more than -20% for ECHAM) in South Africa, Sahel zone comprising Senegal, Niger, Volta and Chari (more than -50% for IPSL) and Central and Eastern Asia comprising the Ganges/Brahmaputra (more than -20% for IPSL), Indus, Amudarja and Huang He (more than -10% for ECHAM).
Technical Report No. 45 - 11 -
5. Discussion and Conclusion In Haddeland et al. (2011a) the significant differences between land surface model (models that calculate the land surface energy balance) and global hydrological model (without energy balance calculation) simulations are found to be caused by the snow scheme employed. The physically-based energy balance approach used by Land Surface Models (LSMs) generally results in lower snow water equivalent values than the conceptual degree-day approach used by GHMs. Some differences in simulated runoff and evapotranspiration are explained by model parameterizations, although the processes included and parameterization used is not distinct to either LSMs or GHMs. Note that in our study we didn’t distinguish between LSMs and GHMs. With regard to future projections, results of the present study yielded that those large differences in the projected changes between the 8 GHMs may be attributed to the different model formulations of evapotranspiration. This becomes especially obvious if the projected changes in evapotranspiration are considered where the uncertainty related to the choice of the GHM is even larger than due to the choice of the GCM. Previous estimates of global water resources typically come from only 1-2 hydrological models, for instance. Those assessments may have considered applying several GCMs, but such studies typically only apply one or two GHMs, and so completely over-look the notion of uncertainty that arises from using different impacts models in climate change impacts assessment. From the analysis of the multiple GCM-GHM simulations in the present study, it can be further concluded that:
1. For the projected hydrological changes, uncertainties due to the choice of GHM are generally larger than due to the choice of the GCM. For the control period, differences due to the choice of GCM are expected to be rather low. As GCM precipitation and temperature are bias corrected, these differences are mainly caused by differences of uncorrected GCM forcing variables, which for many regions are likely less important than precipitation and temperature for the hydrological simulations. How much the uncorrected GCM variables play a role for the hydrological simulations is currently being investigated in a study of Haddeland et al. (2011b)
2. The uncertainties due to the GHM/GCM model choice are generally larger for the projected future hydrological changes than for the simulation of present day conditions in the control period.
3. The east part of Australia, south part of Africa, south part of United States, north east part of South America, south part of Europe, and large parts of the Middle East will likely experience a decreased runoff in the future compared to the control period.
4. Associated with these runoff decreases, a significant reduction in available water resources will occur over many catchments in the respective regions.
5. Only three GCMs are applied in this study, so that the uncertainty due to the choice of the GCM is likely under-estimated, however. Note that the selection of GCMs for this study was imposed by the availability of daily climate model data that are necessary to force the ensemble of GHMs. A respective analysis of the original GCM results over Europe was provided by Hagemann et al. (2008).
Note that we have identified regions whose water resources are vulnerable to the impact of future climate change alone by analysing a multi-model ensemble comprising three GCMs and 8 GHMs. But in these so-called naturalized runs the direct human impact on hydrology has been neglected. With increasing population and improvements in technology in the future, changes in human water requirements and withdrawal will occur that may effect the environment as well as the vulnerability of the different regions to the projected climate change. An assessment of these human impacts is currently being prepared within WATCH by Ludwig et al (2011).
Technical Report No. 45 - 12 -
Acknowledgements. This study was supported by funding from the European Union within the WATCH project (contract No. 036946). The authors would like to thank Tobias Stacke (MPI-M) for implementing several modifications into MPI-HM, and Richard Gilham (UKMO) for data transfer and disaggregation work regarding JULES data. The GCM data were obtained from the CERA database at the German Climate Computing Center (DKRZ) in Hamburg. Additional data were thankfully provided by Nathalie Bertrand from IPSL. References Alcamo, J.; Döll, P.; Henrichs, T.; Kaspar, F.; Lehner, B.; Rösch, T.; Siebert, S. (2003)
Development and Testing of the WaterGAP 2 Global Model of Water Use and Availability. Hydrol. Sci.
Piani, C., Weedon, G. P., Best, M., Gomes, S., Viterbo, P., Hagemann, S., and Haerter, J. O., 2000:
Statistical bias correction of global simulated daily precipitation and temperature for the application of
hydrological models., J. Hydrol., 395, 199–215.
Robock, A., Vinnikov, K.Y., Schlosser, C.A., Speranskaya, N.A., Xue, Y.K., 1995. Use of Midlatitude
Soil-Moisture and Meteorological Observations to Validate Soil-Moisture Simulations with Biosphere and
Bucket Models. J. Climate 8(1): 15-35.
Technical Report No. 45 - 16 -
Roeckner, E., G. Bäuml, L. Bonaventura, R. Brokopf, M. Esch, M. Giorgetta, S. Hagemann, I. Kirchner,
L. Kornblueh, E. Manzini, A. Rhodin, U. Schlese, U. Schulzweida, A. Tompkins, 2003: The atmospheric
general circulation model ECHAM5. Part I: Model description. Max Planck Institute for Meteor. Rep.,
349, 127 pp. [available from MPI for Meteorology, Bundesstr. 53, 20146 Hamburg, Germany].
Royer, J.-F., D. Cariolle, F. Chauvin, M. Déqué, H. Douville, R.M. Hu, S. Planton, A. Rascol, J.-L.
Ricard, D. Salas y Mélia, F. Sevault, P. Simon, S. Somot, S. Tytéca, L. Terray, S. Valcke, 2002:
Simulation des changements climatiques au cours du 21-ième siècle incluant l'ozone stratosphérique.
C. R. Geophys., 334, 147-154.
Rost, S., D. Gerten, A. Bondeau, W. Lucht, J. Rohwer and S. Schaphoff, 2008: Agricultural green and
blue water consumption and its influence on the global water system. Water Resour. Res., 44, W09405,
doi:10.1029/2007WR006331.
Salas-Mélia, D., 2002: A global coupled sea ice-ocean model. Ocean Modelling, 4, 137-172.
Scanlon, T. S., K. K. Caylor, S. A. Levin, and I. Rodriguez-Iturbe, 2007. Positive feedbacks promote
power-law clustering of Kalahari vegetation. Nature 449:209–212.
Sharma, D., A. Das Gupta, and M. S. Babel, 2007: Spatial disaggregation of Bias corrected GCM
precipitation for improved hydrologic simulation: Ping river basin, Thailand. Hydrol. and Earth Sys. Sci.,
11(4): 1371-1390.
Smakhtin, V., C. Revenga and P. Döll, 2004: A Pilot Global Assessment of Environmental Water
Requirements and Scarcity. Water International 29 (3): 307–317.
Su F, Hong Y, Lettenmaier DP. 2008. Evaluation of TRMM multisatellite precipitation analysis (TMPA)
and its utility in hydrologic prediction in the La Plata basin. J. Hydrometeor. 9: 622–640.
Wang J., Hong Y, Gourley J, Adhikari P, Li L, Su F. 2010. Quantitative assessment of climate change
and human impacts on long-term hydrologic response: a case study in a sub-basin of the Yellow River,
China. Int. J. Climatol. 30 (14): 2130–2137.
Weedon, G. P., et al. (2011), Creation of the WATCH Forcing Data and its use to assess global and
regional reference crop evaporation over land during the twentieth century, J. Hydrometeol.
Dol:10.1175/2011JHM1369.1.
Wigmosta, M.S., L. Vail and D.P. Lettenmaier, 1994: A distributed hydrology vegetation model for
complex terrain. Water Resour. Res., 30, 1665-1679.
Xie Z, Yuan F, Duan Q, Zheng J, Liang M, Chen F. 2007. Regional parameter estimation of the VIC
land surface model: methodology and application to river basins in China. J. Hydrometeor. 8: 447–468,
DOI: 10.1175/JHM568.1.
Zhao R, Zhang Y, Fang L, Liu X, Zhang Q. 1980a. The Xinanjiang Model, Hydrological Forecasting
(Proceeding of the Oxford Symposium, April 1980): IAHS-Publ. no. 129, 351–356.
Technical Report No. 45 - 17 -
Zhao ZC, Ding YH, Xu Y, Zhang J. 1980b. Detection and prediction of climate change for the 20th and
21st century due to human activity in northwest China. Climatic and Environmental Research 8(1): 27–
34.
Technical Report No. 45 - 18 -
List of Tables Table 1: GHMs considered in this study and their main characteristics (Haddeland et al. 2011a)
Model name1
Model time step
Meteorological forcing variables2
Energy balance
Evapotranspiration scheme3
Runoff scheme4 Snow scheme
Reference(s)
GWAVA Daily P, T, W, Q, LW, SW, SP
No Penman-Monteith
Saturation excess / Beta function
Degree day
Meigh et al. (1999)
H08 6 h R, S, T, W, Q, LW, SW, SP
Yes Bulk formula Saturation excess / Beta function
Energy balance
Hanasaki et al. (2008a)
JULES 1 h R, S, T, W, Q, LW, SW, SP
Yes Penman-Monteith
Infiltration excess / Darcy
Energy balance
Cox et al. (1999), Essery et al. (2003)
LPJmL Daily P, T, LWn, SW No Priestley-Taylor
Saturation excess Degree day
Bondeau et al. (2007), Rost et al. (2008), Fader et al. (2010)
MacPDM Daily P, T, W, Q, LWn, SW
No Penman-Monteith
Saturation excess / Beta function
Degree day
Arnell (1999), Gosling and Arnell (2010)
MPI-HM Daily P, T No Thornthwaite Saturation excess / Beta function
Degree day
Hagemann and Dümenil Gates (2003), Hagemann and Dümenil (1998)
VIC Daily/3hP, Tmax, Tmin, W, Q, LW, SW, SP
Snow season
Penman-Monteith
Saturation excess / Beta function
Energy balance
Liang et al. (1994)
WaterGAP Daily P, T, LWn, SW No Priestley-Taylor
Beta function Degree day
Alcamo et al. (2003)
Table 2 GCMs Centre GCMs Original GCM grid resolution MPI-M ECHAM5/MPIOM T63 ~ 1.9° ~ 200 km
CNRM CNRM-CM3 T42 ~ 2.8° ~ 300 km
IPSL LMDZ-4 3.75° x 2.5° ~ 300 km
Technical Report No. 45 - 19 -
List of Figures
Figure 1: 12 global large scale catchments
Technical Report No. 45 - 20 -
Figure 2: The global map of ensemble mean precipitation from 3 GCMs - ECHAM, IPSL and CNRM in the control period 1971-2000 (upper left figure), precipitation change from 3 GCMs in the future period from 2071-2100 (upper right figure) compared to the control period and standard deviation of change from 3 GCMs (lower left figure).
Technical Report No. 45 - 21 -
Figure 3: Ensemble mean precipitation (upper panel) and temperature (lower panel) changes in the near (2021-2050) and far future (2071-2100) compared to control period (1971-2000) over the 12 large-scale catchments.
Technical Report No. 45 - 22 -
Figure 4: Ensemble mean of runoff from 24 simulation (8 GHMs and 3 GCMs) (left upper figure), standard deviation of runoff from 3 GCMs (left middle figure) and standard deviation of runoff from 8GHM (left lower figure) in the control period from 1971-2000. Ensemble runoff change from 24 simulations (8 GHMs and 3 GCMs) (right upper figure), standard deviation of runoff change from 3 GCMs (right middle figure) and standard deviation of runoff change from 8 GHMs (right lower figure) in the future period from2071-2100 compared to 1971-2000.
Technical Report No. 45 - 23 -
Figure 5: Ensemble mean of Evapotranspiration from 24 simulation (8 GHMs and 3 GCMs) (left upper figure), standard deviation of ET from 3 GCMs (left middle figure) and standard deviation from 8GHM (left lower figure) in the control period from 1971-2000. Ensemble ET change from 24 simulations (8 GHMs and 3 GCMs) (right upper figure), standard deviation of ET change from 3 GCMs (right middle figure) and standard deviation of ET change from 8 GHMs (right lower figure) in the future period from2071-2100 compared to 1971-2000.
Technical Report No. 45 - 24 -
Figure 6: Observed (WFD data) and bias corrected precipitation in the control period and its projected A2 changes in the near and far future over the catchments of Amazon (upper panels), Danube (middle panels) and Nile (lower panels).
Technical Report No. 45 - 25 -
Figure 7: Observed and simulated discharge based on ECHAM forcing in the control period and its projected A2 changes in the far future over the catchments of Lena (upper panels), Danube (2nd row), Amazon (3rd row), and Nile (lower panels).
Technical Report No. 45 - 26 -
Figure 8: Ensemble mean runoff and standard deviations across the 8 GHMs in the control period (left column) and its projected absolute (middle column) and relative (right column) A2 changes in the far future over the catchments of Murray (upper panels), Parana (middle panels) and Yangtze (lower panels).
Technical Report No. 45 - 27 -
Figure 9: Ensemble mean evapotranspiration and standard deviations across the 8 GHMs in the control period (left column) and its projected absolute (middle column) and relative (right column) A2 changes in the far future over the catchments of Murray (upper panels), Parana (middle panels) and Yangtze (lower panels).
Technical Report No. 45 - 28 -
Figure 10: A2 changes (2071-2100 compared to 1971-2000) in available water resources projected by the 8 GHM ensemble averaged for all 3 GCMs (upper left), ECHAM (upper right), CNRM (lower left) and IPSL (lower right). White areas are not considered.
