全球变化知识资源中心

A statistical framework for evaluating changes in extreme events is proposed and applied to evaluate a 20-member, regional climate model ensemble simulation with 30-km resolution. The model is found to represent the statistics and distributions of extreme events, including observed wet day characteristics, wet/dry days, and wet/dry spell characteristics, reasonably across Africa. Simulations of the mid-twenty-first and late-twenty-first century project statistically-significant changes in these societally-relevant climate characteristics in three regions. Intensification of rainfall is projected for the Sahel rainy season, including large increases in wet spell frequency, wet spell duration, and wet spell intensity. These changes are statistically significant at mid-twenty-first century and become more spatially robust by the end of the century. A weaker intensified rainfall trend is also projected over East Africa, for northern Ethiopia in boreal summer, Tanzania in boreal winter, and southern Ethiopia, Somalia, and the Lake Victoria region in boreal fall. The changes are significant in scattered regions at mid-twenty-first century, but widespread and highly significant by the end of the century. In contrast, increased dry periods is projected for parts of southern Africa (Angola, Zambia, Malawi), including 50-100% reductions in wet spell frequency annually and increases in dry spell duration in austral spring. The spatial coherence and rigorous statistical analysis of the projected changes combined with their physical consistency with the findings of previous studies support confidence in these results.