Probabilistic Forecasting Based on Hydrometeorological Ensembles

Over the last two decades the paradigm in hydrometeorological forecasting has shifted from deterministic to probabilistic. Weather prediction and hydrological models are run increasingly as ensemble forecasting systems, which provide a finite sample of forecast scenarios. Ensemble forecasts are often biased and lack calibration. Different statistical methods to correct for bias and miscalibration are developed and applied to real-world problems within the framework of this thesis.