Seagrass Meadow Mapping in the Bay of Bengal Using Machine Learning and Remote Sensing Lingeswaran A(1), Melvin Fredrick J S(1), Lathaselvi G(2), Vimalathitthan Shanmugam(3*)
(1) Student, Department of Information Technology, St. Josephs College of Engineering, Chennai, India
(2) Associate Professor, Department of Information Technology, St. Josephs College of Engineering, Chennai, India
(3*) Assistant Professor, Department of Information Technology, St. Josephs College of Engineering, Chennai, India
*email:vimalathitthan[at]stjosephs.ac.in
Abstract
Seagrass meadows play a crucial role in maintaining coastal biodiversity, stabilizing sediments, and mitigating climate change through carbon storage, yet they are declining rapidly due to human-induced pressures. Effective large-scale monitoring remains difficult, particularly in turbid coastal waters where field-based surveys are limited. This study presents a satellite-based framework for mapping seagrass distribution using multispectral imagery from Sentinel-2. The input data were corrected, normalized using Min-Max scaling, and resampled to a uniform 10-m spatial resolution. Spectral reflectance patterns and vegetation indices were employed to distinguish seagrass from surrounding benthic features, and the classification process was validated against reference observations. The mapping workflow was applied using a chunk-based approach to handle large raster datasets efficiently, generating probability surfaces and binary presence-absence maps of seagrass cover. The results demonstrate high thematic accuracy and clear spatial delineation of seagrass habitats, even under challenging water conditions. This framework offers a reproducible and cost-effective tool for long-term ecological monitoring, providing critical information for conservation planning, restoration initiatives, and sustainable management of coastal ecosystems.
