QUANTIFYING RIVERBANK EROSION AND ASSOCIATED LAND COVER CHANGES ACROSS THE COASTAL UPAZILA OF HIZLA, BANGLADESH: AN INTEGRATED APPROACH USING GIS AND MACHINE LEARNING

Abstract

Riverbank migration is a common phenomenon in the floodplains of Bangladesh. The continuous changes in river morphology affect its surrounding land-use patterns which pose threats to the life and property of people living near the rivers. The present study utilized thirty-one (1989-2020) years of satellite data to track the erosion-accretion and its influence on land-use and land-cover (LULC) change of Hizla Upazila using GIS-based Modified Normalized Difference Water Index (MNDWI) and Google Earth Engine (GEE) respectively. Statistical analysis revealed the average erosion rate (5.03 km2) is lower than accretion (5.72 km2), resulting in a net land gain of 24.93 Km2. The spatial distribution of erosional activity suggests that the central and western parts of Hizla Upazila are mostly affected, compared to the eastern part, where new deltas are forming. This phenomenon is attributed to the westward movement of the Lower Meghna River (LMR), making the central and western parts of Hizla more vulnerable. The pattern of land-use change manifests that nearby settlements and vegetation are primarily at risk due to channel migration. A significant decrease in total water area (2.1%) and an increase in bare land area (5.1%) between 1997-2010 indicates substantial deposition. Concurrently, there was a decrease in the total settlement (1.53 km2) and vegetation area (9.8 Km2), indicating natural hazards like floods and high-intensity rainfall. The overall kappa accuracy for LULC is over 85% demonstrating its suitability for forecasting. The outcomes of this study will aid the local community, policymakers, and researchers in mitigating risk and ensuring sustainability.