Change Detection and Patch Analysis of Sundarban Forest During 1975–2018 Using Remote Sensing and GIS Data

Krishan Kundu,Prasun Halder,Jyotsna Kumar Mandal

DOI: https://doi.org/10.1007/s42979-021-00749-8

2021-06-30

SN Computer Science

Abstract:In the present study, change detection of Sundarban forest has been analyzed over the last 43 years (1975–2018) using Normalized Difference Vegetation Index (NDVI). Spectral indices like NDVI method are more superior as compared with the other technique because NDVI directly derived from the satellite image. The present study deals with five Landsat multispectral satellite data (1975, 1990, 2000, 2010, and 2018) and the prime objective is to extract the meaningful information from the satellite images. The images are classified into four classes on the basis of the NDVI values and that are dense forest, sparse forest, water bodies, and wet land. The classification results revealed that net forest areas were gradually declined by 2.1% (47.48 km²), although it was not homogeneous over the total period and likewise other features also change. The study examined that 94–99%, 39–82%, 59–73%, and 62–91% areas were unchanged for the dense forest, sparse forest, water bodies, and wet land, respectively. The most significant change was that sparse forest areas were translated into wet land and water bodies, and water bodies to wet land. Fragmentation of forest was detected using the estimation of number of patches which shows that numbers of patches were increased over the time, whereas sizes of each patches were gradually decreased. The classification accuracy was quantified through the overall accuracy and Kappa statistic. The present study highlights the changes of different features may be useful for the policy- or decision-makers to take appropriate measures for protection of Sundarban forest.