40-Year Volume Changes of West African Lakes Derived from SWOT and Optical Imagery

The datasets described below are published along with the article "40-Year Volume Changes of West African Lakes Derived from SWOT and Optical Imagery" (submitted to Water Resources Research) F. Girard, L. Kergoat, J. M. Silva Paiva, R. Yonaba, and M. Grippa For more information, you can contact: - Félix Girard ([e-mail]) - Manuela Grippa ([e-mail]) - Laurent Kergoat ([e-mail]) ########## Volume change time series (volume_change_time_series.csv): ########## Dataset (CSV) containing the 40-year volume change time series for 1363 different West African lakes included in the SWOT Prior Lake Database (PLD). For each different PLD ID, volume changes are referenced to the first acquisition of the time series. The columns correspond to: - pld_id: ID of the lake's PLD polygon - time: acquisition time of Landsat image - vol_change (km3): lake volume change between the present time and the previous one - sat_ls: Landsat satellite associated with the present acquisition (e.g. LT05 for Landsat-5) - tile_ls: ID of the Landsat tile associated with the present acquisition (e.g. 195_49) ########## Hypsometry dataset (lake_hypsometry.csv): ########## Dataset (CSV) containing hypsometry data (pairs of time-matched water surface area and elevation measurements) for 1363 different West African lakes included in the SWOT Prior Lake Database (PLD). For each different PLD ID, hypsometry data are sorted by ascending water surface_area. The columns correspond to: - pld_id: ID of the lake's PLD polygon - time_swot: acquisition time of SWOT measurement - time_ls: acquisition time of Landsat image - wse: SWOT-based lake water surface elevation - water_area: CNN/Landsat-based lake water surface area - tile_SWOT: ID of the SWOT tile (e.g. 432_126L) - tile_ls: ID of the Landsat tile (e.g. 195_49) ########## CNN/Landsat train and test polygons (cnn_landsat_train_polys.gpkg and cnn_landsat_test_polys.gpkg): ########## Two datasets (Geopackage) containing the polygons (EPSG:4326) used to train and test the CNN/Landsat model, respectively. Training dataset (763 polygons) spans different climates from southern Burkina Faso to Central Mali. Testing dataset (355 polygons) covers surface contexts in Niger, Mali, Burkina Faso, and Mauritania which have not been observed during the training of the model. The columns correspond to: - id: unique integer id - ls_image: name of the Landsat image associated with the polygon - geometry: coordinates of the polygon