This study investigates the suitability of the asynchronous ensemble Kalmanfilter (AEnKF) and a partitioned updating scheme for hydrologicalforecasting. The AEnKF requires forward integration of the model for theanalysis and enables assimilation of current and past observationssimultaneously at a single analysis step. The results of dischargeassimilation into a grid-based hydrological model (using a soil moistureerror model) for the Upper Ourthe catchment in the Belgian Ardennes show thatincluding past predictions and observations in the data assimilation methodimproves the model forecasts. Additionally, we show that elimination of thestrongly non-linear relation between the soil moisture storage andassimilated discharge observations from the model update becomes beneficialfor improved operational forecasting, which is evaluated using severalvalidation measures.
展开▼
