This study derived twelve Extreme Rainfall Indices (ERIs) such as the Maximum Dry Spell (MDS) and Maximum Wet Spell (MWS) from daily rainfall observed over the period 1961–1990 at nine locations across East Africa. Capacity of six CO-ordinated Regional Climate Downscaling EXperiment (CORDEX) Africa Regional Climate Models (RCMs) driven by twenty six Climate Model Intercomparison Project phase 5 (CMIP5) General Circulation Models (GCMs) to reproduce the observed ERIs with respect to long-term mean and trends was evaluated. Four RCMs and their five driving GCMs were further analyzed with respect to ERIs. Ensemble means of the RCMs' biases in simulating trends in several ERIs were of magnitudes above 50%. On average, biases in reproducing long-term mean were smaller than those for trends in ERIs. The difference between the performances of RCMs and GCMs depended on the selected RCM–GCM pair. The ensemble means of the RCMs reproduced observed ERIs better than the individual RCMs corroborating that the use of multi-model ensembles can boost credibility of climate change simulations and projections. The RCMs performed better than their driving GCMs in reproducing MDS. The biases of both the RCMs and GCMs were smaller in reproducing the MWS than MDS. Nonetheless, in reproducing observed MWS, the ensemble mean of RCMs' biases was slightly larger than that of the driving GCMs indicating possible adding up of the uncertainties from the GCMs and RCMs. Suggested RCMs' improvements regarding aerosol impacts on rainfall include adding missing constituents (like nitrate), and refining the crudely represented components. RCMs also require high resolution description (in both space and time) of land use types, land surface covers and characteristics as well as landscape heterogeneity. The GCMs to be used as the initial and lateral boundary conditions for the RCMs require improvement in their representation of key dynamical and thermodynamical feedbacks in the Tropical Indian Ocean.

这项研究从1961–1990年期间在东非的9个地点观察到的每日降雨量得出了12个极端降雨指数（ERI），例如最大干法术（MDS）和最大湿法术（MWS）。由二十六个气候模型比对项目第5阶段（CMIP5）的一般环流模型（GCM）驱动的六个协调一致的区域气候缩减试验（CORDEX）非洲区域气候模型（RCM）的能力，可再现长期观察到的ERI均值和趋势进行了评估。关于ERI，进一步分析了四个RCM及其五个驱动GCM。RCM在模拟几个ERI趋势中的偏差的综合手段，其幅度在50％以上。平均而言，重现长期均值的偏差小于ERI趋势中的偏差。RCM和GCM的性能之间的差异取决于所选择的RCM–GCM对。RCM的整体方法比单个RCM更好地再现了观察到的ERI，这证实了使用多模型合奏可以提高气候变化模拟和预测的可信度。在复制MDS方面，RCM的性能优于其驱动GCM。在复制MWS时，RCM和GCM的偏差都小于MDS。但是，在重现观测到的MWS时，RCM偏差的总体平均值稍大于驱动GCM的偏差的平均值，这表明可能会合并来自GCM和RCM的不确定性。建议的RCM在气溶胶对降雨的影响方面的改进包括添加缺失的成分（如硝酸盐），以及提炼粗略表示的成分。区域协调机制还要求对土地利用类型，土地表面覆盖和特征以及景观异质性进行高分辨率描述（在空间和时间上）。用作RCM初始和横向边界条件的GCM需要改进它们在热带印度洋中关键动力和热力反馈的表示形式。