A 1-km annual dataset of cropland net profit for wheat, maize and rice across China, 2000–2019

Overview This dataset provides annual 1-km gridded net profit estimates (CNY mu⁻¹) for wheat, maize, and rice cultivated on cropland across mainland China over a 20-year period (2000–2019). Net profit is defined as gross revenue minus total production costs (including material inputs, hired and imputed family labour, and land costs), strictly following the accounting framework of the National Agricultural Product Cost-Benefit Survey Yearbook. Methodology & Validation Pixel values were generated by spatially downscaling provincial cost-benefit statistics to 1-km resolution using a detrended XGBoost model driven by 15 biophysical and socioeconomic predictors, followed by additive bias correction to enforce exact consistency with yearbook-reported provincial totals. The reliability of the dataset is supported by rigorous validation, including leave-one-year-out cross-validation (residual R² = 0.76–0.88), independent comparison with the National Rural Fixed-Point Survey (~20,000 households, 2009–2015), and sensitivity analysis across sample-definition thresholds. Significance & Application Crucially, these spatially explicit economic metrics provide the essential foundation for identifying and analyzing uneconomic cropland areas. By accurately mapping areas where total production costs exceed gross revenues, this dataset supports empirical analysis of spatial heterogeneity in agricultural profitability and cropland economic viability. Note: For comprehensive methodological details, data structure (60 GeoTIFF files), and usage instructions (including spatial projection handling), please refer to the accompanying README.md file and the associated manuscript currently under review. Code is available upon reasonable request or will be released upon publication.