WSB-DA: Water Surface Boundary Detection Algorithm Using Landsat 8 OLI Data

Abstract

The satellite remote sensing technique provides a comprehensive output, which helps to assess the wetland condition and regional-to-global hydrological cycle. Several algorithms facilitate the delineation of water surface boundary, but these are computationally difficult and vulnerable to unrealistic representation of the inland wetlands. The objective of this study is to develop a semiautomated, generic, parameter-free, and multispectral water surface boundary-detection algorithm (WSB-DA) to delineate the free water surface boundary of wetlands. The uniqueness of WSB-DA is that, it is a combination of five different methods: 1) standard water indices (e.g., ${\text {AWE}}{\text{I}_{{\rm nsh}}}$ , MNDWI, and ${\text {NDW}}{\text{I}_{{6} - {3}}}$ ); 2) the tasseled cap transformation; 3) hue saturation value (HSV) color transformation; 4) high-pass filtering; and 5) vector analysis. Furthermore, this algorithm uses “ultra-blue” rather than “normal blue” band of the Landsat 8 OLI image. The final output of the WSB-DA is verified using detailed GPS survey points and real time photographic evidences of Sambhar Lake. Additionally, the proposed algorithm provides appropriate results for other wetland test sites, i.e., Great Salt Lake, Chilka Lake, and parts of Sundarban Delta. The WSB-DA allows extraction of water surface boundary in a simple way, which is currently nonexistent for extremely shallow wetlands. Moreover, the WSB-DA is a reasonable proxy for the establishment of free water surface boundary detection, which allows for avoiding extensive ground survey.