Short Wave Infrared Band Based Flood Classification Near the Garonne and Ohio River Basins

Remote sensing images provides rich information over different wavelengths creating multispectral images capable of flood related applications. Visible bands in the multispectral images, namely, red, blue and green bands are certain times not sufficient enough to detect floods. Hence it is important to utilize Short-Wave Infrared (SWIR) bands due to their sensitivity to various surface materials and climatic factors. The results of flood classification using random forest in the bench-marking flood dataset demonstrates the importance of SWIR bands in addition to visible bands.