Enhancement of Water Index Feature of Satellite Image in Mountainous Areas with Slope Information

Abstract

A novel slope adjusted water index (SAWI) is proposed to enhance the performance of the modified normalized difference water index (MNDWI) in mountainous areas, where water body and terrain shadow are prone to be confused, due to their similar spectral reflectance in the green and medium-wave infrared (MIR) bands. To overcome this problem, this method introduces a slope information derived from readily available ASTER GDEM-V2 data to the MNDWI, which provides a different scaling parameter to the MIR band for each pixel, such that the difference between terrain shadow and open water can be easily separated in mountainous areas. To validate the effectiveness of the proposed method, four typical sub-scenes clipped from the Operational Land Imager image with different terrain conditions were analyzed, and the results demonstrate that our method can not only possess the ability to relieve the effect of terrain shadow in rugged regions, but also enhance the subtle perception of narrow water body in plain areas, when compared with the MNDWI method. Comparisons with the normalized difference water index and the decision tree classification method were also implemented, and experimental results consistently show that the SAWI outperforms them for all the cases in terms of the used recall, precision and area under curve measurements.