Understanding the causes for discrepancies between modelled and observed regional climate trends is important for improving present-day climate simulation and reducing uncertainties in future climate projections. Here, we analyse the performance of coupled climate models in reproducing regional precipitation trends during the satellite era. We find statistically significant observed drying in southwestern North America and wetting in the Amazon during the period 1979–2014. Historical climate model simulations do not capture these observed precipitation trends. We trace this discrepancy to the inability of coupled simulations to capture the observed Pacific trade wind intensification over this period. A linear adjustment of free running historical simulations, based on the observed strengthening of the Pacific trade winds and modeled ENSO teleconnections, explains the discrepancy in precipitation trends. Furthermore, both the Pacific trade wind trends and regional precipitation trends are reproduced in climate simulations with prescribed observed sea surface temperatures (SST), underscoring the role of the tropical Pacific SST patterns.

了解模拟和观测的区域气候趋势之间差异的原因对于改进当前的气候模拟和减少未来气候预测的不确定性非常重要。在这里，我们分析了耦合气候模型在再现卫星时代区域降水趋势方面的表现。我们发现 1979 年至 2014 年期间北美西南部观测到的干燥和亚马逊地区湿润具有统计显着性。历史气候模型模拟并未捕捉到这些观测到的降水趋势。我们将这种差异归因于耦合模拟无法捕捉这一时期观察到的太平洋信风强度。基于观察到的太平洋信风增强和模拟 ENSO 遥相关，对自由运行历史模拟进行线性调整，解释了降水趋势的差异。此外，太平洋信风趋势和区域降水趋势均在气候模拟中通过规定的观测海面温度（SST）重现，强调了热带太平洋海表温度模式的作用。