Water Body Detection of Chembarambakkam Lake Using the Combination of NDWI and Random Forest Classification

This paper examines the combination of the Normalized Difference Water Index (NDWI) with the Random Forest Classification (RFC) algorithm to improve the detection of water bodies in satellite data. The NDWI is utilized to emphasize water bodies according to their spectral reflectance characteristics, whilst the RFC enhances the classification procedure by utilizing its ensemble learning features. The NDWI+RFC approach attains an accuracy of 52.92%, a recall of 68.93 %, a specificity of 69.25 %, and an F1-score of 66.73 %. The results markedly surpass the independent NDWI approach, which achieves an accuracy of 45.12%, a recall of 66.82%, a specificity of 67.25%, and an F1-score of 62.41 %. The improved performance illustrates the efficacy of the integrated method for water body recognition, providing a significant resource for hydrological research and environmental assessment.