This dataset is under embargo and will be made available by 31 December 2026 at the latest Find out more »
Pal, I.; Mukhopadhyay, A.
Land use and land cover data of the Indian Ganges Brahmaputra Meghna Delta, 1990-2021
https://doi.org/10.5285/6f56ad04-c622-49f9-9326-386f17dcd44a
Cite this dataset as:
Pal, I.; Mukhopadhyay, A. (2026). Land use and land cover data of the Indian Ganges Brahmaputra Meghna Delta, 1990-2021. NERC EDS Environmental Information Data Centre. https://doi.org/10.5285/6f56ad04-c622-49f9-9326-386f17dcd44a
Download/Access
This dataset is under embargo and will be made available by 31 December 2026 at the latest Find out more »
The Indian Sundarban delta, part of the Sundarbans in West Bengal, exhibits diverse land use and land cover including dense mangroves, tidal creeks, mudflats, agricultural fields, aquaculture ponds, and settlements. LULC data supports monitoring of coastal erosion, salinity intrusion, cyclone impacts, and ecosystem-based management planning.
Land Use-Land Cover (LULC) data for the years 1990, 1998, 2000, 2008, 2010, 2019, and 2021 were prepared using satellite imagery downloaded from the United States Geological Survey (USGS) EarthExplorer portal. Appropriate Landsat sensors corresponding to each time period were selected to ensure data consistency and availability. Landsat 5 Thematic Mapper (TM) images were used for 1990, 1998, and 2000, while Landsat 7 Enhanced Thematic Mapper Plus (ETM+) data were used for 2008 and 2010. For recent years, 2019 and 2021 were a derived product of ESRI land use land cover open source data
All images were selected with minimal cloud cover and downloaded as Level-1 products. Pre-processing steps included radiometric correction, atmospheric correction, layer stacking, and subsetting to the study area. Subsequently, supervised classification using training samples was conducted to generate LULC maps. The classified outputs were then validated and reclassified into consistent land-cover categories to enable temporal comparison and change detection across the selected years.
Land Use-Land Cover (LULC) data for the years 1990, 1998, 2000, 2008, 2010, 2019, and 2021 were prepared using satellite imagery downloaded from the United States Geological Survey (USGS) EarthExplorer portal. Appropriate Landsat sensors corresponding to each time period were selected to ensure data consistency and availability. Landsat 5 Thematic Mapper (TM) images were used for 1990, 1998, and 2000, while Landsat 7 Enhanced Thematic Mapper Plus (ETM+) data were used for 2008 and 2010. For recent years, 2019 and 2021 were a derived product of ESRI land use land cover open source data
All images were selected with minimal cloud cover and downloaded as Level-1 products. Pre-processing steps included radiometric correction, atmospheric correction, layer stacking, and subsetting to the study area. Subsequently, supervised classification using training samples was conducted to generate LULC maps. The classified outputs were then validated and reclassified into consistent land-cover categories to enable temporal comparison and change detection across the selected years.
Publication date: 2026-05-19
126 views
Formats
Shapefile, TIFF
Spatial information
Study area
Spatial representation types
Vector
Raster
Raster
Spatial reference system
WGS 84
Spatial resolution
1000 metres
Temporal information
Temporal extent
1990-01-01 to 2021-12-31
Temporal resolution
decadal
Provenance & quality
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.
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.
Licensing and constraints
This dataset is under embargo and will be made available by 31 December 2026 at the latest Find out more »
This dataset will be available under the terms of the Open Government Licence
Cite this dataset as:
Pal, I.; Mukhopadhyay, A. (2026). Land use and land cover data of the Indian Ganges Brahmaputra Meghna Delta, 1990-2021. NERC EDS Environmental Information Data Centre. https://doi.org/10.5285/6f56ad04-c622-49f9-9326-386f17dcd44a
Correspondence/contact details
Authors
Other contacts
Publisher
NERC EDS Environmental Information Data Centre
info@eidc.ac.uk
Rights holder
Newcastle University
Additional metadata
Topic categories
Keywords
Funding
Natural Environment Research Council Award: NE/S008926/1
