| description abstract | his study presents a method to correct regional climate model (RCM) outputs using observations from automatic weather stations. The correction applies a nonlinear procedure, which recently appeared in the literature, to both precipitation and temperature on a monthly basis in a region of complex orography. To assess the temporal stability of such a correction, the correcting parameters of each variable are investigated using different time periods within the observational record. The RCM simulations used in this study to evaluate the bias-correction method are the publicly available ?Reg-CM3? experiments from the Ensemble-Based Predictions of Climate Changes and Their Impacts (ENSEMBLES) project. They provide daily precipitation and temperature time series on a raster with spatial resolution of 0.22°. The analysis is performed in the Rhone catchment, located in southwestern Switzerland and characterized by highly complex orography. The results show that the nonlinear bias correction increases dramatically the accuracy not only of the RCM mean daily precipitation and temperature but also of values across the entire domain of the probability distribution. Moreover, the correction parameters seem to be reasonably independent from the sample used for their calibration, especially in the case of temperature. The good performance of the method over the considered mountainous region during the evaluation period points to the suitability of this technique for correcting RCM biases regardless of the stationarity of the climate and, therefore, also for future climate and in regions characterized by marked orography. | |
