Ascertainment of probabilistic runoff forecasts considering censored data (HyWa) – 2014

When forecasting water levels and river flow, ensemble weather forecasts are frequently used as meteorological input to hydrologic process models. In hydrologic forecasting systems, data below or above certain threshold values are subject to increased uncertainty. In this study, two sub-catchments of the River Rhine are considered. The forecasting systems associated with these two catchments use runoff data that are censored below a predefined threshold value. On the basis of Ensemble Model Output Statistics (EMOS), a censored EMOS method is developed that is able to cope with censored data. Censored EMOS will then be verified over the entire forecast time horizon of 1 to 114 hours using several different statistical measures.