AI-powered insights for coral ecosystems: an innovative approach to seafloor habitat mapping and classification

Coral reefs represent one of the most significant marine ecosystems. They also offer numerous ecological, economic, and protective benefits. Their biological importance includes serving as habitats for marine organisms, safeguarding coastal areas from erosion, and contributing to economic growth through tourism and fisheries. However, coral reefs are increasingly threatened by climate change, pollution, and anthropogenic activities. Therefore, efficient and effective monitoring strategies required. Conventional field-based mapping techniques are often costly and time-consuming. Consequently, this study explores the viability of remote sensing as an alternative by utilizing Sentinel-2 satellite imagery and artificial intelligence (AI) algorithms, which Random Forest and Support Vector Machine to enhance coral habitat mapping in Redang Island, Terengganu, Malaysia. The study integrates satellite imagery, ground truth data, and spectral indices to improve classification accuracy. In the data preparation phase, image preprocessing is conducted on Sentinel-2 imagery, followed by data enhancement through performing spectral indices such as Normalized Difference Vegetation Index (NDVI) and Normalized Difference Water Index (NDWI). Subsequently, model training and accuracy assessment will be carried out using Overall Accuracy and Kappa statistics. The expected findings will demonstrate a significant improvement in mapping efficiency compared to conventional method and underscore the potential application of artificial intelligence in marine conservation, environmental management, and sustainable tourism development.