Evaluating CMIP6 models for CO₂ and CH₄ concentrations across Africa: performance, biases, and implications for climate predictions

Understanding and accurately predicting greenhouse gas concentrations, particularly CO₂ and CH₄, are critical for assessing climate change impacts and informing policy decisions. This study addresses the challenge of evaluating the performance of CMIP6 climate models in simulating these key greenhouse gases across Africa. By comparing model outputs with reanalysis data, we assess the accuracy and reliability of these models in capturing CO₂ and CH₄ concentrations, which are essential for understanding regional climate dynamics and informing adaptation strategies. Our evaluation utilized various statistical methods, including linear regression, Pearson's correlation (mean: 0.82 ± 0.05 for CO₂, 0.67 ± 0.08 for CH₄), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Relative Mean Bias (RMB), to determine model performance and variability. Results indicate that most CMIP6 models exhibit a strong correlation with reanalysis data for CO₂ (mean RMSE: ± 5.3 ppm; MAE: ± 4.8 ppm), although models such as CESM2 and CESM2-FV2 show significant overestimation (RMB: + 15%). In contrast, CH₄ simulations display substantial underestimation, particularly in northern Africa, with error margins ranging from ± 13.7 to ± 22.4 ppb, revealing difficulties in accurately representing CH₄ emissions and sinks. These findings underscore the models’ strengths in CO₂ simulations but highlight notable limitations in CH₄ modelling. Future research should focus on improving model parameterizations and addressing regional biases to enhance the accuracy of climate projections and deepen our understanding of greenhouse gas impacts.