Thermal comfort indices improve characterization of weather-driven electricity demand and associated CO₂ emissions in Kuwait

Electricity demand in hot arid regions is strongly influenced by thermal conditions, but it remains unclear whether integrated thermal comfort indices explain demand more consistently than basic meteorological variables. This study evaluated weather metrics associated with daily electricity demand in Kuwait and quantified energy use and carbon dioxide increments above climatic thresholds. A 26-year daily electricity load record from 1999 to 2024 was normalized by population and detrended to reduce interannual non-climatic variation. The adjusted load series was matched with hourly reanalysis data for air temperature, relative humidity, Universal Thermal Climate Index, Heat Index, and Humidex. Weather–load relationships were evaluated using Spearman correlation, simple and multivariable regression, and extreme gradient boosting with Shapley additive explanations, while thresholds were estimated using piecewise regression. Temperature and thermal comfort indices were strongly and positively associated with maximum, minimum, and average electricity load, with Spearman coefficients of 0.92–0.95. Relative humidity showed moderate negative associations, with coefficients of −0.66 to −0.71. Average Heat Index showed the strongest correlations across all load measures and the highest feature importance. Piecewise regression identified average Heat Index and air–temperature thresholds of 18.7 °C and 19.8 °C. Above these thresholds, each 1 °C increase was associated with an additional 2.03 and 2.08 kWh per capita per day, corresponding to 1.40 and 1.43 kg of CO₂ per capita per day. These findings identify average Heat Index as the most consistent weather indicator and provide threshold benchmarks for electricity system planning and climate adaptation in cooling dominated regions.