Multi-decadal climate change impacts on African solar hydrogen: Weighted ensemble forecasting and techno-economic vulnerability

Africa's immense solar endowment positions the continent as a pivotal hub for green hydrogen production, yet infrastructure investment decisions routinely assume climatic stationarity, ignoring how multi-decadal climate change will progressively degrade solar resources over a project's operational lifetime. This study addresses this critical gap by developing an integrated analytical framework combining weighted ensemble machine learning, climate projections, and climate-adjusted techno-economic analysis to forecast solar-based hydrogen production across fifteen geographically distributed African locations through 2045. Five complementary machine learning and deep learning algorithms were trained on 19 years of meteorological data (2005–2023), with a dynamic inverse-root mean square error (RMSE) weighting scheme selecting the top three performers per location. The ensemble achieves high-fidelity agreement with the physics-based simulation framework and delivers a 92.2% RMSE reduction relative to a deterministic daily-average physics baseline. Under combined warming scenarios, continental mean production declines reach −7.6±3.6% by 2035, escalating to −9.8±3.7% by 2045 and amplifying to −12.77±4.09% under the high-warming pathway. Regional vulnerability is strongly differentiated: East Africa emerges as the most exposed region (−12.9% by 2045 combined), while West Africa exhibits the greatest resilience (−7.4%). Climate-adjusted Levelized Cost of Hydrogen analysis reveals baseline costs of $3.35–$7.10/kg escalating to $3.90–$10.29/kg under the 2045 high-warming scenario, representing increases of +10.7% to +44.9%, with capacity factor confirmed as the dominant economic driver. These findings establish that geographic site selection based on long-term climate stability and cost trajectory is equally critical as technology selection for African green hydrogen infrastructure investment.