Algorithms for Water Body Extraction from Remote Sensing Data

In several disciplines, including remote sensing, environmental monitoring, and land use planning, the identification of water bodies is a critical undertaking. Due to changes in water quality and quantity as well as cloud cover and shadows, finding water bodies can be difficult. Accurate water body recognition has now become practical and widely employed in a variety of applications thanks to advancements in machine learning algorithms and remote sensing technology. Various techniques have been proposed for water body detection, including threshold-based methods, machine learning-based methods, and index-based methods. This article discusses various methods for extracting water bodies and details how well these algorithms can identify the water bodies. When attempting to extract information about water bodies, different methods will yield varying results, and one crucial aspect of the accuracy of this process is how each method's optimal selection of its parameters. Currently, normalized difference water body index (NDWI) method, SVM, CART, Object-oriented detection, Agglomerative and K-Means clustering, U-NET and CNN are the algorithms experimented and the corresponding observations of accuracies of water body detection methods are 0.99, 0.97, 0.98, 0.92, 0.96, 0.90, 0.99, and 0.98 respectively for remotely sensed satellite images.