Evaluation of Extreme Drought and Flood Events in East Africa Based on CMIP Models

Abstract

This study evaluates 19 CMIP6 historical precipitation model datasets spanning 1979–2014, using the GPCP monthly and GPCC daily precipitation datasets as benchmarks to ensure comparability. The analysis compares individual model results against observational data to dissect simulation differences across various models. The evaluation covers several critical aspects of precipitation modeling, including climatological mean changes, which are examined through spatial distribution, range, and standard deviation. Seasonal changes are also scrutinized, focusing on the long rainy season (March–May) and the short rainy season (October–December) to capture seasonal precipitation dynamics. Furthermore, the study investigates interannual changes by analyzing precipitation trends and eigenvectors to understand long-term variability. Key findings include significant variability in model performance—INM-CM5-0 shows strong correlation with observations, while CNRM-CM6-1 performs poorly—and most models can basically simulate precipitation spatial patterns but with overestimation/underestimation in specific regions. Additionally, it assesses model performance in capturing extreme drought/flood events, providing insights for regional climate projection and disaster early warning.