Climate hazards and power-system resilience: why heat, hydropower and storms require different indicators

Abstract Climate hazards affect electricity systems through different physical channels and at different temporal and spatial scales. This study tests whether internationally comparable monthly data can identify those distinct responses. Two linked panels for 2019–2025 are analysed: a 12-country climate panel with 1,008 country-month observations and a 74-country storm panel with 6,216 observations. Storm tracks from the International Best Track Archive for Climate Stewardship are spatially linked to 34,936 geocoded power plants. Models include country-by-calendar-month and global year-month effects, small-cluster inference, and extensive sensitivity tests. Each additional degree of monthly cooling-threshold exposure above 18°C is associated with 2.62% higher electricity demand. By contrast, operational hydropower shortfalls do not robustly predict absolute fossil generation or net imports, and tropical-storm exposure does not produce a statistically distinguishable national monthly generation response. These null aggregate estimates should not be interpreted as evidence that droughts or storms cause no disruption. They show that national monthly data capture persistent heat-related pressure more effectively than short, localized outages or restoration processes. The study reframes this asymmetry as a hazard-measurement result and demonstrates that resilience indicators must match each hazard’s physical pathway, geographic footprint, and duration. The findings support heat-responsive adequacy planning and higher-resolution monitoring of hydrological and storm risks.

Hali tarjima qilinmagan