Bhitarkanika Mangrove Species Change Detection Using Hyperspectral Remote Sensing and Field Survey

Abstract

This study aims to find the temporal changes in mangrove species present in Bhitarkanika Wildlife Sanctuary by utilizing Hyperion satellite images of 2008 and 2015 classified using spectral library of mangrove species and ground truth locations of the species from the study area. Hyperion images require preprocessing such as radiometric calibration, atmospheric correction, and bad bands removal before further analysis. Since spectral library was used for classification, we used a feature selection algorithm, Recursive Feature Elimination (RFE) to reduce the dimension instead of conventional spectral transformation methods and Spectral Angle Mapper (SAM) and Spectral Information Divergence (SID) classifiers were used to map the mangrove species.