Automatic detection and delineation of surface water bodies in airborne hyperspectral data

In this investigation we focused on the development of a new water detection algorithm based on VIS-NIR imaging spectroscopy data. By analyzing different images containing inland and ocean water types we found the slopes of the reflectance spectrum of water at specific spectral wavelengths within the VIS-NIR spectral region to be diagnostic features for surface water identification. However, the presence of these features depends on the spectral superimposition of water constituents and bottom coverage. This aspect has been considered in the development of a knowledge-based classifier. The results (probability of detection generally lies above 90%) indicate the great potential of the developed algorithm for surface water body detection and delineation within urban, rural and coastal scenes.