Improved Flood Forecasting using Numerical Weather Prediction Meteorological Forecasts with Integrated Rainfall-Runoff and Hydrodynamic Model

This study develops an improved daily flood forecasting system using the Numerical Weather Prediction (NWP) model forecasts. Two different hydrological model setups are developed integrating rainfall-runoff and river-channel routing. The first model setup represents the existing operational flood forecasting network with limited number of base stations. The existing setup is improvised in this study considering the river network receiving contributions from the entire catchment area to obtain the second model setup. To analyze the daily discharge forecasting abilities of the existing and improvised model setups, they are forced with the observed hydro-meteorological inputs in the hindcast period only i.e., Case-I and Case-II, respectively. To further improve the performance of the improvised model setup, it is forced with the NWP forecasts in the forecast period along with observed hydro-meteorological inputs in the hindcast period (Case-III). While the existing system (Case-I) provides reasonable flood forecasts up to 1-day lead-time only, Case-II shows acceptable results for up to 3-days lead-time. Using the NWP forecasts in Case-III improved flood forecasting of the improvised model setup and provided reasonably accurate overall time-series forecasts up to 5-days lead-time. However, Cases II-III are able to predict the peak flows reasonably well up to 2-days and 3-days lead-time, respectively. The overall findings suggest the robustness of the use of integrated rainfall-runoff and hydrodynamic model along with NWP forecasts for providing reliable flood forecasts at higher lead-times.