A vast number of weather forecast and climate models have a common warm-and-dry bias, accompanied by the underestimation of evapotranspiration and overestimation of surface net radiation, over the central United States during boreal summer. Various theories have been proposed to explain these biases, but no studies have linked the biases with the missing representation of human perturbations, such as irrigation. Here we argue that neglecting the impact of irrigation contributes to the longstanding warm surface temperature and lack of precipitation biases over this region. By using convection-permitting multi-season simulations over the contiguous United States coupled with an operational-like irrigation scheme, we show that irrigation increases surface evapotranspiration and decreases surface temperature by increasing evaporative fraction. By increasing the frequency of mesoscale convective systems, irrigation reduces the summertime model precipitation deficit and improves the simulated precipitation diurnal cycle over the Great Plains. The increased precipitation also alleviates the warm bias in our simulation setup, likely by damping the positive feedback between soil moisture and temperature.

在北方夏季，美国中部的许多天气预报和气候模型都存在普遍的暖干偏差，伴随着蒸散量的低估和表面净辐射的高估。已经提出了各种理论来解释这些偏差，但是没有研究将偏差与诸如灌溉之类的人类摄动的缺失表示联系起来。在这里，我们认为忽略灌溉的影响会导致该地区长期的温暖地表温度和缺乏降水偏差。通过在连续的美国使用允许对流的多季节模拟，再加上类似操作的灌溉方案，我们表明灌溉可通过增加蒸发量来增加地表蒸散量并降低地表温度。通过增加中尺度对流系统的频率，灌溉减少了夏季模型降水不足，并改善了大平原地区模拟的降水日循环。降水的增加也减轻了我们模拟设置中的温暖偏差，这可能是通过抑制土壤水分和温度之间的正反馈来实现的。