Land use and land cover data of the Indian Ganges Brahmaputra Meghna Delta, 1990-2021

The LULC datasets were subject to multiple verification and quality control steps to ensure consistency and reliability. Historical LULC information for 1990, 1998, 2000, 2008, 2010, and 2019 obtained from government reports and previously demarcated as particular class (Such as agriculture , Huge waterbody etc).Spatial consistency was checked by overlaying the datasets with high-resolution imagery available from Google Maps, and visually-verified selected reference points.

For recent datasets (2019 and 2021), land-cover products derived from ESRI were used. These datasets were further calibrated and validated by ground truthing through field-based GPS observations collected during the study period (2022 - 2023). Ground control points were compared with classified land-cover categories to assess positional accuracy. Additional cross-checking, attribute correction, and consistency checks across years were performed to ensure temporal comparability and overall data quality.

Advanced AI-based land classification models were used to generate high-resolution land use - land cover maps by training algorithms with billions of human-labeled image pixels. These models were applied to the complete annual archive of imagery from the Sentinel-2(10 Meter Resolution ) Earth observation mission, processing more than two million scenes using six spectral bands. The approach produced a consistent nine-class land cover map, including vegetation types, cropland, water, bare land, and built-up areas. The datasets are available through the ArcGIS Living Atlas of the World and provide accurate, high-resolution information that supports land-use planning, resource management, and decision-making by governments and planners worldwide.