Assimilating streamflow data to update water table positions in rainfall-to-runoff models based on topmodel concepts

To improve the performance of forecasting models, real-time data can be assimilated to update the model states for each new forecast. This paper reports on a non-statistical way to update the groundwater state of rainfall runoff models based on TOPMODEL concepts. The updating procedure uses measured flows and an analytical relationship between water table position and baseflow. To apply the procedure it is desirable to define periods when there is no surface runoff to the stream network. An innovative and effective model-based way to define these periods is described. By assimilating measured flow data into a TopNet flood forecasting model, the forecasts of both high and low flows are improved, with the changes in water table position being consistent with the corrections to the forecast hydrograph.