Land–water transition zone mapping challenges in dynamically changing environments

Land water transition zones (LWTZ) exhibit physicochemical features, which in relation with the spectral sensitivity of the spaceborne sensor, signal to noise ratio of the satellite bands and orbit specifications impose limiting factors for the accuracy of derived inundation maps. Various approaches are implemented based on the degree of specialization of the applied mapping techniques. This study attempts a comparison across the various techniques used by water utilities for monitoring this dynamically changing land area. Visual interpretation of Sentinel-2 images over an isolines map, generated across years at the reservoir of Polyfytos in northern Greece, is compared with the computer-aided derivation of the Normalized Difference Wetness Index to showcase the benefits of multispectral information in fine tuning of the product. Automatic thresholding techniques, originating from computer vision, further enhance the discrimination capacity.