Land cover change across 45 years in the world’s largest mangrove forest (Sundarbans): the contribution of remote sensing in forest monitoring

Abstract

This study explored the land use land cover (LULC) change over 45 years (1975 -2020) in the world’s largest mangrove forest, Sundarbansusing Landsat imagery. LULC maps were created with same-season imagery with the lowest cloud cover at four intervals: 1975, 1990, 2005, and 2020. Maximum likelihood classification (MLC) was applied to assign five classes: dense forest, moderate forest, sparse forest, barren land, and water body. Accuracy assessment was carried out with 250 control points for each year resulting in overall accuracy and kappa coefficient ranging from 84.8% to 90.0% and 0.81 to 0.87, respectively. Results show dense forest at its highest cover in 1975 and then decreasing by an estimated annual rate of 1.3% from 1975 to 2020, but not consistently. Dense forest class mostly turned moderate and sparse; most of the sparse forest class turned to barren land. Most of the barren lands were located near the boundary between forest and human settlement, and these two classes were more frequent in the Indian part of Sundarbans than in the Bangladesh part. The conclusion is that the time-series of remote sensing data can validly support effective forest management by identifying space and time changes in the biodiversity of Sundarbans.