The increasing persistence of high daytime and nighttime temperatures poses a growing threat to urban public health, with approximately 55% of the population in European cities exposed to sustained heat stress. However, the cumulative effects of prolonged heat exposure remain insufficiently quantified. This study develops a data-driven framework integrating satellite-based land surface temperature (LST) and reanalysis data with machine learning to assess the association between sustained heat exposure and heat-related mortality across European cities. Using 7,567,756 death records from 122 cities, we quantified the duration and contribution of multi-week cumulative heat effects. Results show that prolonged heat exposure substantially elevates mortality risk over time. Female mortality (205 deaths per million, 95% CI: 160–250) exceeds that of males (145 deaths per million, 95% CI: 95–195), with risk increasing markedly with age. On average, cumulative heat effects persist for approximately four weeks, with notable regional variability. Temperature from the preceding week contributes up to 70% of the current week’s mortality risk, highlighting a strong temporal carryover effect. Incorporating cumulative exposure substantially improves model performance (R2 = 0.679, RMSE = 26.823 deaths per million), with up to a 75% increase in explanatory power compared to models using only concurrent temperature. Independent validation across 90 additional cities confirms moderate generalizability (R2 = 0.584, RMSE = 35.276 deaths per million). These findings provide robust evidence of the multi-week cumulative impact of heat exposure and highlight the value of satellite-based temperature data for large-scale heat-health risk assessment.
Qian et al. (Tue,) studied this question.