High-resolution flood mapping from low-resolution passive microwave data

We have developed a new method for making high-resolution flood extent maps (e.g. at the 30-100 m scale of digital elevation models) in real-time from low-resolution (20-70 km) passive microwave observations. Microwave radiometric measurements are useful for flood monitoring because they sense surface water in clear-or-cloudy conditions and can provide more timely data (e.g. compared to radars) from relatively wide swath widths and an increasing number of available platforms (DMSP, ADEOS-II, Terra, NPOESS). The chief disadvantages for flood mapping are the radiometers\´ low resolution and the need for local calibration of the relationship between radiances and open-water fraction. We present our method for transforming microwave sensor-scale open water fraction estimates into high-resolution flood extent maps and show several animated examples of the method in action during a retrospective study of the 1993 Great Midwest Flood. The method builds a "flood-potential" database from elevations and historic flood imagery and uses it to create a flood-extent map consistent with the observed open water fraction. We discuss the method\´s potential improvement through as yet unimplemented algorithm enhancements and expected advancements in microwave radiometry (e.g. improved resolution and atmospheric correction).