Evapotranspiration occurs via land‐atmosphere coupling. To understand the impacts of irrigation on local potential evapotranspiration, we carried out a couple of simulations with the Weather Research and Forecast (WRF) model, where an irrigation scheme was incorporated, for the irrigated area in the middle reaches of the Heihe River, Northwest China. By comparing the irrigation simulation (hereafter, the IRRG case) with the simulation excluding irrigation (hereafter, the NATU case), we found that irrigation may lead to a decrease in reference evapotranspiration (ET₀) during the growing season (May to September), which is a metric of potential evapotranspiration, by approximately 1.05 mm d⁻¹, accounting for approximately 19% of NATU ET₀. The ET₀ in the IRRG is closer to the ground measurements than the NATU ET₀, with a root‐mean‐square error of 0.75 mm d⁻¹ in the IRRG and 1.86 mm d⁻¹ in the NATU cases. Meanwhile, irrigation leads to an actual evapotranspiration (ET_a) increase of approximately 2.1 mm d⁻¹ for croplands. Such an asymmetric relationship between ET₀ decrease and ET_a increase was found over the irrigated croplands. The ET₀ decrease was mainly induced by the moistening and cooling effects of irrigation through intensified evapotranspiration and reduced sensible heat. This study highlights the importance of considering land‐atmosphere coupling when assessing the impacts of climatic change on irrigation water requirements.