Ensemble Prediction Systems (EPSs) are increasingly applied for rainfall forecasts and flooding warning systems. In this paper, these forecasts and their skills are evaluated through relevant criteria, particularly by considering forecast performances for different lead times. Furthermore, to enhance their performance, we propose to preprocess the EPS forecasts’ output using bias correction methods. For this aim, forecasts for different ranges of precipitation as well as various climatic conditions are evaluated, which is particularly important for extreme events that can lead to flooding. The Karun River basin in Iran is used as case study, a large area including various climate conditions. The results showed that the performance of European Center for Medium-Range Weather Forecasts (ECMWF) forecasts vary with sub-basin properties, e.g., area and skewness of daily precipitation, and vary across dry to humid regions, between flooding and non-flooding seasons, and different lead times and show the effect of different methods for bias correction on the forecast skill. The forecast skill is decreasing from wet regions towards dry regions and the bias correction was more effective in the flooding season, for which the skill was increased by 40% based on continuous ranked probability skill score (CRPSS). When the precipitation thresholds were increased towards extreme values, the forecast performance of ECMWF became better.
- 22/2 OA procedure