Accurate electricity demand forecasting is a critical strategic imperative for Tunisia, a country characterized by high fossil fuel dependency and increasing vulnerability to Mediterranean climate extremes. While conventional forecasting models are typically calibrated on stationary historical dynamics, their predictive reliability is increasingly compromised by intensifying heatwaves.,This study investigates the impact of a significant structural climate shift—identified around 2016—on the performance of a SARIMAX modeling framework. To address the resulting loss of robustness, an innovative corrective metric is introduced: the Thermal Anomaly Index (∆T). Its integration markedly restores forecast accuracy and provides a resilient framework for energy planning under climate uncertainty.