Optimizing cropland allocation enhances climate resilience more than time management under extreme rainfall

The gridded dataset used in this repository consists of 0.5° × 0.5° random-forest emulator outputs. The emulators were trained on site-year yield data generated with an improved APSIM-Barley configuration that explicitly represents waterlogging effects. For each grid cell and year over 1982–2100, separate models were developed for VT (conventional/normal genotype) and VS (waterlogging-tolerant genotype), using as predictors background climate, soil properties, CO₂, growing degree days, and a suite of extreme temperature and precipitation indices over the growing season. Daily climate forcing was taken from five bias-corrected and statistically downscaled CMIP6 GCMs (CanESM5, CNRM-CM6-1, CNRM-ESM2-1, EC-Earth3, MIROC6) under the SSP5-8.5 scenario, and the dataset provided here is the median across these five GCMs. Together, these emulator outputs provide spatially explicit projections of crop yields and their responses to future extreme climate conditions, with explicit sensitivity to waterlogging processes. The example code used to develop these emulator-based datasets is provided in the GitHub - llinchao/emulator.