The Global Flood Database answers a question that regulatory flood maps like FEMA's don't: where has flooding actually occurred, and for how long? Built by Cloud to Street and the Dartmouth Flood Observatory from MODIS satellite observations, it provides mapped inundation extents for 913 major flood events across nearly two decades.
Unlike modeled flood zones that estimate probability, this dataset shows observed flood footprints — where water actually was — along with duration and population exposure for each event. That distinction makes it a powerful complement to probabilistic flood maps for validating risk models, calibrating hydrological simulations, and identifying areas that flood repeatedly.
The database's global coverage is particularly valuable in regions where detailed flood mapping infrastructure doesn't exist. In much of the developing world, there are no equivalents to FEMA flood maps or national hydrological models, so satellite-derived historical flood records may be the only spatial evidence of flood risk available. Disaster management agencies, development banks, and insurance analysts use the data to screen investments and infrastructure plans against historical flood exposure.
The 250-meter resolution is suited to regional and national-scale analysis rather than parcel-level assessment, but for identifying flood-prone corridors, comparing event severity across basins, or building training datasets for machine learning flood models, the Global Flood Database is one of the few consistent global sources available.
