Probabilistic downscaling of EURO-CORDEX precipitation data for the assessment of future areal precipitation extremes of different durations

Abstract. This work presents a methodology to inspect the changing statistical properties of precipitation extremes with climate change. Data from regional climate models for the European continent (EURO-CORDEX 11) were used. The use of climate model data requires first an inspection of the data and a correction of the biases of the meteorological model. Both the correction of biases of the point precipitation and those of the spatial structure were performed. For this purpose, a quantile-quantile transformation of the point precipitation and a spatial recorrelation method were used. Once bias-corrected, the data from the regional climate model were downscaled to a finer spatial scale using a stochastic method with equally probable outcomes. This enables the assessment of the corresponding uncertainties. The downscaled fields were used to derive area-depth-duration-frequency (ADDF) curves, and area-reduction-factors (ARF) for selected regions in Germany. The estimated curves were compared to those derived from a reference weather radar data set. While the corrected and downscaled data show good agreement with the observed reference data over all temporal and spatial scales, the future climate simulations indicate an increase in the estimated areal rainfall depth for future periods. Moreover, the future ARFs for short durations and large spatial scales increase compared to the reference value, while for longer durations the difference is smaller.