3.1
Estimating and Forecasting Floods Globally Using Satellite Rainfall Information and Hydrological Models

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 4 February 2014: 12:00 AM
Room C210 (The Georgia World Congress Center )
Robert F. Adler, University of Maryland, College Park, MD; and H. Wu and Y. Tian

Floods occurring around the globe are probably the number one disaster in terms of lives lost and damage done. Detection and forecasting of such events is critical to minimize loss of life, and information on flood extent, intensity, and time history is critical for disaster management and mitigation activities. This is true anywhere, but becomes especially important in less developed parts of the world where in-situ disaster information may be lacking, hampering emergency responses, especially in the critical early stages of relief planning. Many national agencies (e.g., DoD, DoS) and international agencies and humanitarian organizations (World Food Program, International Red Cross) require rapid, accurate information on the when, where and how many people are affected, which in turn requires useful real-time information on the time-space characteristics of the flood events.

To help meet these needs, a real-time experimental system to estimate and forecast floods over the globe, the Global Flood Monitoring System (GFMS), has been developed and tested to provide flood detection, streamflow and inundation mapping information at high resolution (as fine as 1 km) and nowcasts and forecasts (out to five days). Images and output data are available for use by the community with updates available every three hours (http://flood.umd.edu). The system uses satellite-based rainfall information, currently the TRMM Multi-satellite Precipitation Analysis [TMPA]), other satellite and conventional information and a hydrological and routing combination model, the Dominant river Routing Integrated with VIC Environment (DRIVE) system. The surface hydrological calculations are carried out at 0.125 latitude-longitude resolution with routing, streamflow and other calculations done at that resolution and at 1km resolution. Flood detection and intensity estimates are based on water depth and streamflow thresholds calculated from a 15-year retrospective run using the satellite rainfall and model. The satellite rainfall data are also integrated with global model NWP rainfall predictions (adjusted to the satellite data) to extend the flood calculations out to five days. Examples of results for recent flood events are presented along with validation statistics and comparison with other flood observations (e.g., inundation calculations vs. MODIS and/or SAR flood maps).

The outlook for further development in this area in terms of increased utility for national and international disaster management/response agencies is described in relation to new data sources (e.g., improved rainfall and soil moisture information), improved modeling, and further integration and dissemination of remote sensing flood products (e.g., hydrological calculations, MODIS/SAR flood mapping, etc.).

Supplementary URL: flood.umd.edu