Bayesian spatio-temporal modeling of African CO2 emissions (1990–2020): A hierarchical approach to identify determinants, regional trends, and local dynamics

Africa's unique position in global CO₂ emissions demands rigorous analysis for effective climate policy development. Despite contributing only 4% to global emissions, the continent faces disproportionate climate impacts while undergoing rapid development and complex economic transitions. Current research lacks extensive continental analysis of the spatial dependencies, temporal evolution of emission patterns, and their key drivers, which is fundamental for evidence-based climate policy and sustainable development strategies. We applied Bayesian hierarchical spatio-temporal modeling to analyze CO₂ emissions across African countries (1990-2020), integrating rotated empirical orthogonal function (REOF) analysis with spatial autocorrelation techniques (Local Moran's I and LISA) to capture complex emission patterns. Our hierarchical framework incorporated demographic and environmental predictors, revealing urbanization as the dominant emission driver. Surface temperature and relative humidity showed significant positive associations, while forest cover demonstrated a strong negative relationship. Spatial analysis identified distinct emission clusters, with the first three REOF modes explaining 78% of total variance. Strong positive spatial autocorrelation in North Africa contrasts with negative patterns in Southern regions, suggesting regional development networks could influence emission trajectories. These findings highlight opportunities for low-carbon development during Africa's urbanization phase through integrated urban planning and forest preservation. The spatial dependencies highlight the importance of coordinated regional approaches to emission reduction, providing evidence for targeted climate policies that balance local contexts with regional interdependencies.