A high-resolution flood forecasting and monitoring system for China using satellite remote sensing data

Abstract

Flood disasters occur frequently in China, resulting in property damage and fatalities. To reduce these risks, a high-resolution flood forecasting and monitoring system was developed based on a simple, distributed hydrological model: CREST. There are mainly four parts in this flood forecasting system, which are the data pre-processing and input module, the distributed hydrological model, the results output and post-processing module, as well as the results verification module. The data inputted in this system are high-resolution CMORPH satellite remote sensing precipitation and surface condition data, including ASTER DEM, HWSD soil types and MODIS vegetation and land cover types, or so. Those data were interpolated to 1 km spatial resolution and then used to drive the hydrological model to run. The CREST model was developed to provide rapid online prediction of large area which may be continental scale or even global scale, while it is also applicable at small scales, such as small basins. It can simulate the spatiotemporal variation of water and energy fluxes as long as storages on a regular grid with the grid cell resolution being user-defined, which ranges from meters to kilometers. The results outputted from the forecasting system are hydrological variables like soil moisture, potential and actual evapotranspiration, surface and river runoffs, etc. Results were then applied to GIS system and then translated into flood disaster warning information in the data output and post-processing module of the forecasting system. Both hydrological variables and flood disaster warning information will be tested and verified in the results verification module. With the 1 km horizontal spatial resolution, the forecasting system successfully simulated basic hydrological variables and processes such as actual evapotranspiration, soil moisture, and surface runoff according to the analysis and comparison with observation data. System hindcasted flood disaster information of several flood events were carefully analyzed. The results showed that the hydrographs of several stations hindcasted by the forecasting system demonstrated that the forecasting system can approximately hindcast river runoffs and streamflows. The performance of the forecasting system was tested by analyzing several real cases of flooding in different places in China, and results indicated that the system hindcasted flood information, including flood timing and spatial extent, were useful for disaster preventions and mitigations.