GlobalRice20: A 20 m resolution global paddy rice dataset for 2015 and 2024 derived from multi-source remote sensing

GlobalRice20 is a global 20 m paddy rice dataset for 2015 and 2024, designed to support global food security assessment and tracking progress toward SDG 2 (Zero Hunger). The maps were generated using a dedicated Time-Series-to-Vision framework (T2VRCM) that converts heterogeneous optical and SAR time series into standardized 2D visual representations, enabling robust mapping despite cloud contamination, irregular temporal sampling, and missing sensor modalities, particularly in cloud-prone tropical regions. The dataset was produced from Sentinel-1/2 and Landsat-8 imagery and validated against reference samples, achieving an overall accuracy of 92.33%; comparisons with national agricultural statistics further demonstrate strong consistency. The data are released as binary uint8 GeoTIFFs in EPSG:4326, where 1 denotes paddy rice, and are organized into global 45° × 45° tiles with one GeoTIFF per tile. Files follow the naming convention 2015_Lat{minLat}_Lon{minLon}.tif and 2024_Lat{minLat}_Lon{minLon}.tif (e.g., 2024_Lat0_Lon135.tif).