Downscaled global 60-meter resolution estimates of irrigation water sources (2000-2015).

This dataset provides high-resolution (60 m) global irrigation maps to support water resource and agricultural management. It identifies the likely irrigation status (rainfed or irrigated) and water source (groundwater or surface water) of croplands for 2000, 2005, 2010, and 2015. We downscaled a 10-km irrigation dataset derived from national and subnational statistics (GMIA) using (i) spatial patterns between high-resolution (30 m) cropland and nearby surface water, and (ii) irrigation water requirements from a global crop model. Validation used household agriculture surveys in India (N = 8,355) and a U.S. well database (N = 1,505,371). In the U.S., our method achieved 85% accuracy in distinguishing groundwater use within 2 km of wells - substantially higher than GMIA (25%). In India's groundwater-dominated regions, our estimates performed comparably to GMIA (73% vs. 72%). These results suggest our dataset offers a more accurate and spatially detailed representation of irrigation water sources, enabling improved analysis of agricultural water use.