Extreme events such as hurricanes pose challenges to clean power system planning. Existing studies rarely jointly manage the interactive relationships among uncertain climate, socioeconomic, and environmental risks. To address this gap, this study develops a Bayesian Network-based non-deterministic optimization model that integrates multidimensional risk management with system optimization to support long-term climate-resilient and low-carbon power system planning. The model enables robust decision-making under multiple risks while balancing trade-offs among system resilience, economic costs, and emission mitigation. A case study of Nova Scotia, Canada (2026–2045) demonstrates its applicability in hurricane-prone regions. To evaluate hurricane impacts on system reliability, a system performance loss index (SPL) is developed. Results show that high-intensity hurricanes could increase SPL to 10.69–22.70%, causing 25,847.54–54,891.40 GWh of unavailable supply and C$ 208.22–776.78 million in economic losses. SPL may also rise five-fold to 17.81–37.81% with increased hurricane frequencies. Low-fragility natural gas generation will supply ∼30% of electricity, with its capacity losses roughly one-third of wind generation. Energy storage-based planning would strengthen system resilience and reduce additional emissions of ∼3.42 Mt but comes with high electricity costs, reaching 0.21–6.87 C$/kWh and increasing by 20–50% under high hurricane frequency. In comparison, diversified generation and grid hardening strategies would maintain stable costs within 0.064–0.081 C$/kWh. Therefore, transitioning to a diversified mix of natural gas and renewables with underground lines, supported by cost-feasible storage measures, would offer a balanced climate-resilient strategy. This study provides quantitative insights into climate adaptation-mitigation linkages and identifies cost-effective management plans.
