Directions in global flood nowcasts and forecasts using satellite precipitation observations

A global flood detection/prediction system is running in real-time using satellite multi-satellite rainfall analysis in combination with hydrological models and algorithms to estimate key flood parameters (http://trmm.gsfc.nasa.gov). The system also uses satellite-based land surface information such as digital elevation information from the NASA SRTM (Shuttle Radar Terrain Mission) and vegetation information from MODIS in the model and algorithm calculations. The flood determination algorithm consists of three major components: 1) multi-satellite precipitation estimation; 2) characterization of land surface including digital elevation information and other surface information, topography-derived hydrologic parameters such as flow direction, flow accumulation, basin, and river network etc.; 3) a hydrological model to infiltrate rainfall and route overland runoff. Results of calculated water depth over a threshold are then displayed about six hours after real-time. Time-history of inundations are also calculated and displayed. Validation analysis indicates generally good results for flood detection and evolution, but with limitations in the current routing calculations. Occasional flood events are missed due to limitations in the satellite rain estimations.

Recent results with an improved global hydrological model running at 1/8th degree resolution will be discussed that are shown to produce more realistic evolution of flooding events and more detailed information. Validation results over the U.S. will be emphasized. Initial tests of using global numerical weather prediction rainfall forecasts to extend the period of utility of the flood information will also be presented.