Coupled Model Intercomparison Project Phase 6 simulations of the spatial structure of rainfall variability over East Africa: Evaluation and projection

• Published in: International journal of climatology Vol. 42; no. 16; pp. 9865 - 9885
• Main Authors: Mbigi, Dickson, Onyango, Augustine Omondi, Mtewele, Zacharia Florence, Kiprotich, Paul, Xiao, Ziniu
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 30.12.2022; Wiley Subscription Services, Inc
• Subjects: 21st century; Analysis; Climate models; CMIP6; East Africa; Empirical analysis; empirical orthogonal function; Interannual variability; Intercomparison; Modelling; Orthogonal functions; Precipitation; Rainfall; Rainfall patterns; Rainfall regime; Rainfall simulators; Rainfall variability; Simulation; Variability; East Africa
• ISSN: 0899-8418; 1097-0088
• DOI: 10.1002/joc.7868

Modifications in rainfall patterns may have significant effects on a variety of natural and human systems. This study evaluates the ability of 20 Coupled Model Intercomparison Project Phase 6 (CMIP6) to simulate the interannual variability of rainfall over East Africa (EA) using a method based on the empirical orthogonal function (EOF) analysis. The future changes in rainfall variability during the near (2021–2040), middle (2041–2060) and late (2080–2099) future are analysed under two different shared socioeconomic pathways (SSP), SSP2‐4.5 and SSP5‐8.5. Results reveal that most models captured better spatial climatological rainfall pattern than simulated amplitude in the EA region receiving bimodal rainfall pattern (EABM) compared to that with unimodal rainfall regime (EAUM) in the historical period. An ensemble mean of all models (AMME) and a set of 13 models that best simulated the rainfall variability in the base period were selected using a robust method based on the EOF analysis for further analysis. Most of the selected models and their ensemble mean (BMME) displayed good capability in representing the annual standard deviation (SD) in recent decades, whereas BMME corroborates AMME, particularly over the EABM and EAUM regions. Based on these findings, the AMME and BMME were used to evaluate the future changes in rainfall variability. The models project a significant increase in rainfall variability during March by the mid and late 21st century over the EAUM region under SSP5‐8.5, whereas the increase appears much earlier in the near‐future over the EABM region. In all future periods and SSPs, SD demonstrates a considerable increase over most of the EABM region, and the magnitude gradually increases from the AMME to BMME projections. Moreover, a relatively stronger increase is anticipated to actualize by the mid of 21st century. A method based on the EOF analysis generally shows smaller RMSE values in most models and all models' ensemble (AMME) indicating better simulation of the historical rainfall variability. The corresponding simulated monthly standard deviation (SD) values also reveal a consistent pattern relative to the observations. As such, the ensemble mean of the best models (BMME) and AMME shows varied future SD values subject to the EA subregions and seasons under different SSP scenarios, indicating overall projected change in rainfall variability.