Planning-Data-enabled Land Cover of China (PDLC) 2021

This dataset is the output of the manuscript: An Accurate Land Cover Classification Map in China Integrating High-resolution Satellite Imagery and Planning Data with Deep Learning

This dataset encompasses the land cover of the entire land area of China, including nearshore islands but excluding offshore islands. The land cover product is generated using 2-meter resolution satellite imagery from Google Earth acquired in 2021. Rigorous validation was carried out through pixel-level accuracy assessment, point-based validation using an open-access validation set, and large-scale areal consistency evaluation. The results indicate that our product performs comparably to widely used remote sensing datasets and exhibits the strongest agreement with aggregated official statistics from China’s Third National Land Survey, especially in the categories of built-up areas, cropland, and roads. This product provides a robust foundation for fine-grained urban and regional analysis, enabling applications in spatial planning, land use policy development, and sustainable development monitoring. Its high spatial resolution, accuracy, and clear boundaries offer significant practical value for research and decision-making.

The dataset is organized by administrative units (districts/counties) and stored in GeoTIFF format using the geographic coordinate system WGS1984. These administrative units are ordered according to their unit IDs in the accompanying county-districts.shp shapefile. To facilitate efficient file management and reduce the overall number of files, all districts within each prefecture-level city have been merged into a single district polygon.

Due to storage limits and repeated upload failures, the landcover of some counties/districts (ID>=1700) could not be hosted in this repository. Please find the external download links for the full dataset in the file download_links.txt.

This land cover product comprises six categories: built-up areas, cropland, roads, green spaces, water bodies, and others. The relationship between raster values and land cover categories is as follows:

• 0: Null
• 1: Others
• 2: Green spaces
• 3: Cropland
• 4: Roads
• 5: Built-up areas
• 6: Water bodies