Climate Change Projection in the Upper Danube (European Alps) and the Upper Brahmaputra (Himalayas)

Coarse-grid global circulation models do not allow for regional estimates of the water balance or trends of extreme precipitation. This is especially true in complex terrain. Therefore, downscaling of the global simulations to generate regional precipitation is necessary. This paper discusses dynamical and statistical downscaling in two watersheds: (1) the upper Danube in the European Alps and (2) the upper Brahmaputra in the Himalayas. The discussion focuses on simulation of daily precipitation patterns and the advantages/disadvantages of the downscaling approaches in the different climate regions.

The global datasets to be downscaled are the ECMWF ERA40 re-analysis data and climate scenarios by the global climate model ECHAM5. Dynamical downscaling is performed with the regional climate model CLM (www.clm-community.eu) with horizontal grid spacing of 0.44° and 0.165°. Statistical downscaling is performed with a simple scaling approach using precipitation of the global data set as predictor. A combination of the dynamical downscaling method with the statistical approach is additionally proposed.

Evaluation of the simulated precipitation patterns indicate that both methods perform well in the data-rich European area and perform less well in the data-poor monsoonal Asian area. The statistical approach is not able to deal with the sparseness of precipitation events in the non-monsoon season. The CLM is developed in Europe and the transfer of the computational domain to South-Asia, especially with respect to the convection parameterization, is not without challenge. In the light of the discussed uncertainties this paper gives a careful discussion of simulated trends of daily precipitation and regional patterns.