CSLD: a long-term sandy land distribution in China from 2000 to 2022

The China Sandy Land Dataset (CSLD) developed in this study provides annual 30-m sandy land maps for the period 2000–2022. The dataset is freely available on the Zenodo platform (https://doi.org/10.5281/zenodo.17799809). All data layers follow the EPSG:4326 (WGS 84) coordinate reference system and are provided in GeoTIFF format. Each annual map is named using the format “CSLD-XXXX”, where XXXX denotes the corresponding year (2000–2022). Each CSLD raster file contains pixel-wise sandy land classification results with the following coding scheme: 1 – Mobile sandy land, 2 – Semi-fixed sandy land, 3 – Fixed sandy land. The spatial resolution of the dataset is 30 m, matching the resolution of the Landsat reflectance data used for classification. Each GeoTIFF file is self-contained and can be directly loaded into GIS and remote-sensing platforms such as ArcGIS, QGIS, ENVI, or Google Earth Engine. Users may visually inspect, analyze, or integrate these maps with other geospatial datasets for further applications in land degradation assessment, ecological restoration monitoring, or desertification studies. The annual CSLD series represents the classification outputs generated by the multi-year Random Forest model described in this paper, followed by spatiotemporal consistency filtering to ensure temporal stability. As a result, the dataset provides a consistent, long-term record of sandy land dynamics across China, enabling robust spatiotemporal analyses over more than two decades.