Mangrove forest health assessment using hyperspectral remote sensing for Gulf of Kutch, Gujarat: a case study

Mangrove forests provide crucial ecological and socio-economic benefits but are increasingly threatened by anthropogenic activities and climate change. Traditional monitoring methods face challenges due to the dense, remote, and tidal nature of these ecosystems. This study explores the potential of hyperspectral remote sensing for large-scale and periodic assessments of mangrove forests, particularly where field data collection is limited. Using hyperspectral vegetation indices (HVIs) derived from Airborne Visible/Infrared Imaging Spectrometer – Next Generation (AVIRIS-NG) data, mangrove areas in Marine National Park (MNP), Jamnagar, India, were classified into “most,” “moderate,” and “less” healthy categories based on spectral indicators. The study proactively addresses potential limitations by carefully selecting the Decision Tree classifier and integrating HVIs sensitive to mangrove biophysical and chemical parameters, thereby mitigating issues such as overfitting. This classification serves as a preliminary measure of forest condition rather than a substitute for detailed ecological assessments. The selected HVIs were correlated with Leaf Area Index (LAI) from Sentinel-2 MSI Level-2A data, confirming their effectiveness in capturing key biophysical parameters. With an overall classification accuracy of 88.93% and a kappa coefficient of 0.82, this study demonstrates the feasibility of hyperspectral imaging for remote mangrove monitoring. The findings highlight its value in identifying areas of concern for targeted field investigations, providing a scalable approach for conservation planning and ecosystem management.