Seagrass Meadow Mapping in the Bay of Bengal Using Machine Learning and Remote Sensing Lingeswaran A(1), Melvin Fredrick J.S(1), Vimalathitthan Shanmugam(1*) and Lathaselvi G(1)
(1) Department of Information Technology, St. Josephs College of Engineering, Chennai, India
*email:vimalathitthan[at]stjosephs.ac.in
Abstract
Seagrasses are vital for the health of coastal ecosystems, maintaining biodiversity and acting as buffers against climate change through carbon storage. Seagrasses remain threatened by coastal development, pollution, and climate change, especially in sensitive areas such as the Bay of Bengal. Reliable and timely seagrass distribution mapping is crucial in ensuring successful conservation and management. This research is interested in the automated mapping of Bay of Bengal seagrass beds based on sophisticated machine learning algorithms and satellite remote sensing. In this research, we use machine learning algorithms to process satellite remote sensing data to map seagrass meadows throughout the Bay of Bengal. Multispectral images acquired from Sentinel-2 and Landsat 8 satellites are processed with supervised classification techniques, such as Random Forest and Support Vector Machines, to differentiate seagrass from other benthic substrates (Maxwell et al., 2017- Lyons et al., 2012). Spectral indices like Normalized Difference Vegetation Index (NDVI) and Bathymetric Index are utilized as input features to enhance model performance. Existing ecological surveys^ ground truth data and marine biodiversity repositories (e.g., UNEP-WCMC, 2020) are utilized for training and validation. The initial results indicate excellent classification accuracy and large-scale, automated monitoring capability. The resulting machine learning-driven seagrass distribution maps provide valuable information to marine resource managers, policy-makers, and conservationists operating in the Bay of Bengal.
