Spatio-temporal; Artificial Intelligence; Multi-Hazard Risk Mapping; Renewable Energy Sitting; Multi-objective Optimization

The siting of renewable energy systems (RES) in regions vulnerable to multiple climate hazards presents a critical challenge for sustainable infrastructure planning. Traditional approaches, primarily driven by static assessments of solar and wind potential, often neglect the compounded risks posed by floods, droughts, and windstorms, resulting in investments that are operationally vulnerable and economically unsustainable. This study proposes a novel spatio-temporal artificial intelligence (AI) framework for multi-objective RES deployment that integrates satellite-derived resource maps, high-resolution hazard data, and dynamic climate time series into a unified optimization pipeline. The methodology employs a gated recurrent unit (GRU)-based encoder to capture temporal hazard dynamics, combined with a multi-objective evolutionary algorithm (NSGA-II) to balance energy yield and resilience. A case study in South Africa’s Vhembe District demonstrates the framework’s effectiveness: the proposed model reduces average hazard exposure by 31.6% while preserving 96.4% of baseline energy output. Attention-based saliency analysis reveals that flood and windstorm hazards are the dominant drivers of site exclusion. Compared to conventional siting methods, the proposed framework achieves superior trade-offs between performance and risk, ensuring alignment with South Africa’s Just Energy Transition and Climate Adaptation strategies. The results confirm the value of spatio-temporal embeddings and hazard-aware multi-objective optimization in guiding resilient, data-driven energy infrastructure development. This model offers direct benefits to energy planners, climate adaptation agencies, and policy-makers seeking to implement resilient, data-driven renewable energy strategies in hazard-prone regions.