Improving Flood Forecast Capability by Coupling Radar-based Precipitation Estimation with Distributed Hydrological Model

This paper explored the potential of improving flood forecast capability by coupling radar-based precipitation estimation with distributed hydrological model and did some validations in Southern China. CINRDA radar is China's new generation digital weather radar that has been deployed in whole China and has been put into real-time operation. CINRDA has a cycle coverage with a radius of 230km at an 5-6 minutes, the spatial resolution is 1-4km2. CINRDA radar will provide a new measurement for China's flood warning and forecasting. To validate the CINRDA estimated precipitation, this paper first proposed a data quality control and precipitation estimation procedure for CINRDA and was tested in the Guangzhou CINRDA in Southern China, the results were compared with that interpolated with rain gauges, and a composited gridded precipitation was formulated based on the above research and is capable of coupling with the distributed hydrological model. Physically based distributed hydrological models are regarded to be able to better represent the hydrological processes, thus have the potential to improve the flood forecast capability. Liuxihe Model is a physically-based distributed hydrological model mainly proposed for watershed flood forecasting, and was employed in this study. Liuxihe Model divides the studied basin into a number of cells horizontally, which are further divided into 3 layers vertically. All cells are classified as hillslope cells, river cells and reservoir cells according to their flow accumulation while having its own properties and model parameters. The saturation excess mechanism is employed to determine the surface runoff while the interflow is calculated using Campbell's equation. The runoff routing is divided into hillslope routing and river routing. The model parameters are categorized as unadjustable and adjustable parameters. The unadjustable parameters are derived directly from the cell's properties, while the adjustable parameters are calibrated to improve the model performance with three methods tested including the SCE-UA algorithm, the SP-UCI algorithm and the Particle Optimization method. The Liuxihe Model could be easily coupled with the composited gridded precipitation estimated main based on radar observation. Several river basins in southern China with the basin areas ranging from several hundred to ten thousand square kilometers were case studied. The results suggest that by coupling radar-based precipitation estimation with distributed hydrological model, the flood forecasting capability could be improved both in forecast accuracy and lead time, and could be employed in real-time flood forecasting. Also several issues need to be further explored such as parameter sensitivity and uncertainty, precipitation uncertainty and its propagation in hydrological model, data assimilation for distributed hydrological modeling, parallel computation for application in large scale river basins.