A pragmatic approach for downscaling precipitation in alpine-scale complex terrain

Abstract

A statistical method is presented to downscale precipitation from a mesoscale atmospheric model simulation. The algorithm consists of two steps. First, local subscale variability is estimated based on a high resolution observed climatology. Second, there is a bias correction, which constrains the downscaled model climatology to be equal to the observed climatology on the coarse grid. Combining both steps results in a local scaling factor for each day of the climatological year. The method is applied to the upper Danube catchment which encompasses part of the European Alps and which is characterized by highly complex orography. The subgrid-scale variability described by the first part of the algorithm partly reflects the underlying orography, especially the narrow alpine valleys. The bias correction leads to a redistribution of precipitation on the catchment scale and accounts for the model deficiency producing too much precipitation in the inner alpine regions and too little at the edges of the Alps. An evaluation with regard to the simulated and observed daily precipitation indicating the significant potential of the method is presented.