Spatially differentiated management-revised discharge scenarios for an integrated analysis of multi-realisation climate and land use scenarios for the Elbe River basin

A spatially differentiated, management-revised projection of natural water availability up to 2053 was requested for a basin-wide scenario study about the impact of global change in the Elbe River basin. Detailed discharge and weather information of the recent years 1951-2003 were available for model calibration and validation. However, the straightforward "classic" approach of calibrating a hydrological model on observed data and running it with a climate scenario could not be taken, because most observed river runoffs in Central Europe are modified by human management. This paper reports how the problem was addressed and how a major projection bias could be avoided. The eco-hydrological model SWIM was set up to simulate the discharge dynamics on a daily time step. The simulation area of 134,890 kmA(2) was divided into 2,278 sub-basins that were subdivided into more than 47,500 homogeneous landscape units (hydrotopes). For each hydrotope, plant growth and water fluxes were simulated while river routing calculation was based on the sub-basin structure. The groundwater module of SWIM had to be extended for accurate modelling of low flow periods. After basin-scale model calibration and revisions for known effects of lignite mining and water management, evapotranspiration and groundwater dynamics were adjusted individually for more than 100 sub-areas largely covering the entire area. A quasi-natural hydrograph was finally derived for each sub-area taking into account management data for the years 2002 (extremely wet) and 2003 (extremely dry). The validated model was used to access the effect of two climate change scenarios consisting of 100 realisations each and resembling temperature increases of 2 and 3 K, respectively. Additionally, four different land use scenarios were considered. In all scenario projections, discharge decreases strongly: The observed average discharge rate in the reference period 1961-1990 is 171 mm/a, and the scenario projections for the middle of the twenty-first century give 91-110 mm/a, mainly depending on the climate scenario. The area-averaged evapotranspiration increases only marginally within the scenario period, e.g., from about 570 to about 580 mm/a for the temperature increase of 2 K, while potential evapotranspiration increases considerably from about 780 to more than 900 mm/a. Both discharge and evapotranspiration changes vary strongly within the basin, correlating with elevation. The runoff coefficient that globally decreases from 0.244 to 0.160 in the 2 K scenario is locally governed primarily by land use; 68% of the variance of the decreases can be attributed to this factor.