Development, Demonstration, and Evaluation of Flood Product due to Snow/Ice Melting from Suomi-NPP/VIIRS Data

Flood caused by snow melting and ice jam occurs almost every year in the U.S., especially in high latitude area such as Alaska and Minnesota. The flood happening along Yukon River near Galena, AK during late May 2013 resulted in dramatic loss to the local residents. Along Red River Basin in Minnesota, snow melting and ice jam frequently cause severe floods. How to obtain dynamic flooding information during snow/ice melting season is crucial for river forecasters to monitor and analyze river status. The Suomi-NPP/VIIRS data, with moderate spatial resolution 375m in Imager bands, large swath coverage and steady resolution at the swath edges, provides a good data source for natural disaster analysis. Over high latitude area, satellite observations are particularly useful to detect and monitor floods caused by snow melting and ice jam. Thus, a flood detection package is developed by using the Suomi-NPP/VIIRS Imager data to dynamically monitor and analyze flood status during snow/ice melting season. Based on spectral characteristics, the package automatically identifies cloud and snow/ice cover in a first, and then decision-tree approach is applied to extract water cover information. A geometric method is also developed to remove cloud shadow from water pixels. Based on water detection result, a dynamic nearest neighboring searching method is utilized to retrieve water fractions. Finally, the retrieved water fractions are compared with MODIS 250m water mask (MOD44W) to determine flooding area. The package is tested and validated with thousands of granules of VIIRS data and shows promising results in various ground and weather conditions. Currently, the package is in the process to run operationally under the CSPP for river forecast centers in Alaska and Minnesota to detect and monitor floods during snow/ice melting season.