Irrigation impacts on grassland hydrological regimes in an arid endorheic river basin

Irrigation is an important management measure for maintaining grassland health and improving productivity in arid and semi-arid areas; however, it can also impact hydrological regimes. Owing to terrain and water source constraints, suitable grassland areas for irrigation are limited and widely dispersed, making it difficult to accurately extract the area and spatial distribution of irrigated grasslands. In this study, we used the Google Earth Engine platform and optical remote sensing (Sentinel-2) data to classify irrigated grassland and other land-use types in the Bayin River Basin, a typical arid endorheic river basin. The classification results were then combined with measured irrigation information from the Delingha Irrigation Management Bureau and input into the SWAT–MODFLOW coupled model to simulate differences in plant growth status, groundwater recharge, and evapotranspiration between irrigated and non-irrigated (rain-fed) grasslands. The accuracy of grassland extraction and secondary classification of irrigated grasslands based on optical remote sensing images was high. Irrigation effectively increased the leaf area index of grasslands during later stages of the growing season and significantly increased the groundwater recharge depth, evapotranspiration, and total water yield during the irrigation season, as well as the annual depth of groundwater replenishment. These changes stabilised annual evapotranspiration and substantially reduced the impact of annual precipitation.