Aligned 30-m factor rasters and graph datasets for urban land evolution across ten Chinese metropolitan areas, 2000–2020
Description
This dataset contains the processed spatial inputs used to analyse urban land evolution across ten Chinese metropolitan study areas: Beijing, Changchun, Chengdu, Chongqing, Fuzhou, Shanghai, Shenzhen, Urumqi, Wuhan, and Xi’an. It covers two transition periods, 2000–2010 and 2010–2020. The dataset is organised into three components: (1) factors, containing 18 aligned 30-m driving-factor rasters for each metropolitan area and period, together with factor_mapping.csv; (2) landex, containing the corresponding categorical land-expansion destination rasters; and (3) pt, containing the final PyTorch Geometric graph objects used in the analysis. All city–period files are organised consistently by data type, transition period, and metropolitan area. The factor mapping documents the correspondence between variables x1–x18, their descriptive names, raster filenames, and graph-feature columns. The dataset supports reproduction and further examination of the spatial response and interaction analyses reported in the associated study.
Files not available for this dataset
This contains only metadata
Steps to reproduce
1. Extract the factors, landex, and pt archives while preserving their directory structures. 2. Select the same transition period and metropolitan area from each directory. 3. Use factors/factor_mapping.csv to map variables x1–x18 to their raster files and graph-feature columns. 4. Read the GeoTIFF files using GDAL, rasterio, or equivalent geospatial software. Factor and land-expansion rasters are provided on the aligned analytical grid. 5. Load the corresponding .pt file using PyTorch and PyTorch Geometric for graph-based modelling. 6. Use the factor rasters as model inputs and the land-expansion destination raster as the categorical response, following the model specification described in the associated manuscript.