A new reconstruction of cropland spatial distribution in China since 1850

Accurately reconstructing historical cropland distribution is essential for quantifying the environmental impacts of agricultural land-use change and understanding human–environment interactions. Existing statistical inference methods provide long temporal coverage but suffer from limited spatial accuracy, while remote sensing–based approaches offer high spatial resolution but are constrained in historical depth and often exhibit area biases. To overcome these limitations, this study developed an integrated reconstruction framework and produced a yearly cropland spatial distribution dataset for China from 1850 to 2020 at a 5 km resolution. Multi-source data, including historical archives, statistical yearbooks, and national land surveys, were synthesized to construct a temporally continuous and internally consistent provincial cropland area time series through phased identification, correction, and interpolation. A dynamically weighted joint probability index was then established by combining remote sensing–derived spatial probabilities with physical suitability–based potential probabilities, enabling consistent spatial allocation of cropland before and after the remote sensing era. Results indicate that China’s cropland area experienced a “fluctuating increase–rapid expansion–continuous decline” trajectory, peaking around 1980. Cropland distribution shifted northward toward Northeast China since the early 20th century and toward Northwest China after the 1950s, while several southern and eastern provinces exhibited net cropland loss by 2020. Compared with existing datasets, the reconstructed product shows higher consistency with national land surveys and more accurately captures key spatial evolution processes. This dataset provides robust support for long-term cropland change assessment and environmental impact studies in China.