Data and trained coefficients for: "Projecting global changes in land use and ecosystem services using SEALS"

This dataset contains the primary spatial outputs of the SEALS (Spatial Economic Allocation Landscape Simulator) model reported in Johnson, Colesanti Senni, Thakrar & von Jeetze, "Projecting global changes in land use and ecosystem services using SEALS" (manuscript in preparation). Contents: Downscaled land-use maps (32 GeoTIFFs). Global land-use/land-cover at 300 m resolution, downscaled from LUH2 coarse projections by SEALS, for 8 SSP-RCP scenarios across 4 time steps (2030, 2050, 2070, 2100). Filenames follow the pattern lulc_esa_seals7_<ssp>_<rcp>_luh2-<iam>_<ssp>_<rcp>_<year>.tif (for example, lulc_esa_seals7_ssp1_rcp19_luh2-image_ssp1_rcp19_2030.tif). Scenarios: SSP1-RCP1.9 (IMAGE), SSP1-RCP2.6 (IMAGE), SSP2-RCP4.5 (MESSAGE), SSP3-RCP7.0 (AIM), SSP4-RCP3.4 (GCAM), SSP4-RCP6.0 (GCAM), SSP5-RCP3.4 (MAgPIE), SSP5-RCP8.5 (MAgPIE). Trained model coefficients (1 CSV: trained_coefficients_seals_manuscript_global_2026_3600sec.csv). The calibrated suitability coefficients, by calibration zone and land-use class, used to generate the maps above (global manuscript calibration). Raster specification: GeoTIFF, 300 m resolution, coordinate reference system WGS 84 (EPSG:4326), global extent. Pixel values are the categorical SEALS-7 land-use class: 1 = urban, 2 = cropland, 3 = grassland, 4 = forest, 5 = other natural, 6 = water, 7 = barren and other. Classes 1 to 5 are actively reallocated by SEALS. Water (6) and barren and other (7) are held fixed. Input data are not redistributed here. They are publicly available from their original providers. The spatial-predictor sources are listed in Table S1 of the paper (ESA-CCI land cover, SoilGrids250m, WorldClim 2, SRTM elevation, Baccini et al. above-ground carbon, Accessibility to Cities travel time, GPW population, WDPA protected areas, and wetlands presence from the Copernicus Global Land Service (Buchhorn et al., 2020)). The coarse LUH2 land-use projections that SEALS downscales are described in the main text (Hurtt et al., 2020).