Resampling of regional climate model output for the simulation of extreme river flows

Abstract

The objective of this paper is to investigate whether resampling of the output from a regional climate model (RCM) can provide realistic long-duration sequences of precipitation and temperature for the simulation of extreme river flows. This is important to assess the impact of climate change on river flooding. Daily streamflows of the river Meuse in western Europe are considered. Resampling is performed with a nearest-neighbour technique that was already successfully applied to the observed daily precipitation and temperature in the river basin. Streamflows are simulated with the semi-distributed HBV rainfall-runoff model. Two simulations of the KNMI regional climate model RACMO are considered. One of these simulations is driven by the global atmospheric model HadAM3H of the UK Meteorological Office for the period 1961-1990 and the other by ERA40 re-analysis data. Much attention is given to the bias correction of RACMO precipitation. It was found that a relatively simple nonlinear correction adjusting both the biases in the mean and variability led to a better reproduction of observed extreme daily and multi-day precipitation amounts than the commonly used linear scaling correction. This also resulted in more realistic discharge extremes, suggesting that a correct representation of the variability of precipitation is important for the simulation of extreme flood quantiles. For the Meuse basin it is further shown that it is advantageous to correct for the variability of the 10-day precipitation amounts rather than that of the daily amounts. Despite the remaining biases in the RCM data, the simulated extreme flood quantiles correspond quite well with those obtained using observed precipitation and temperature.