Combining Water Indices to Detect Water Bodies using Landsat 8 OLI

The delineation of water bodies plays a vital role in the study of surface rivers. The objective of this paper is to propose a new method for accurately extracting water bodies from Landsat 8 OLI. The proposed method is implemented by generating two water indices which are the Modified Normalized Difference Water Index (MNDWI) and the Automated Water Extraction Index (AWEI) from Landsat 8 Operational Land Imager (OLI). The two water index was employed because of its proven capacity to address challenges in discriminating water bodies with high levels of turbidity and those with high correlation with dark pixels (e.g., from shadows). We then classified the combined input images using the Minimum Distance classification method. The method was tested in three test sites and was compared to other water extraction methods using various water index combination. The experimental results showed that the average overall accuracy of the proposed method is 97%, higher than other methods using a two-band water index combination or a single water index.