Economic valuation of temperature-related mortality attributed to urban heat islands in European cities

The study investigates how urban heat islands (UHIs) affect temperature-related mortality across European cities, considering both adverse heat impacts and potential protective effects against cold. While UHIs increase exposure to extreme heat due to reduced evaporative cooling, increased radiative absorption, and anthropogenic heat emissions, they may also reduce cold-related risks. Prior research often focused on thermal comfort or a limited set of cities and lacked intra-urban granularity. Economic impacts of UHIs on health are substantial but under-quantified. This work aims to quantify, at high spatial resolution, the health risks attributable to UHIs throughout the year across many cities, disentangle seasonal trade-offs between heat and cold, identify controlling factors (exposure and vulnerability), and express impacts in monetary terms to inform urban planning and policy.

Existing literature documents adverse health impacts of heat and cold, with notable historical events (e.g., 2003 European heatwave). UHIs are known to exacerbate heat-related risks, but evidence also suggests protection against cold extremes. Most UHI mitigation studies emphasize thermal comfort rather than quantified health outcomes, and many temperature–mortality studies cover few cities without accounting for intra-urban variability. Prior work indicates UHIs can significantly increase the economic impacts of climate change, and city-level costs (e.g., Melbourne) can be substantial. Economic valuation of mortality (VSL/VOLY) has influenced estimates like the social cost of carbon. These gaps motivate a multi-city, high-resolution assessment combining epidemiology, urban climate modeling, and economic valuation to capture spatial heterogeneity and seasonal contrasts.

• Study domain and period: 85 European cities, 2015–2017.
• Urban climate and temperature data: Daily mean 2 m air temperatures simulated by UrbClim (original 100 m, regridded to 500 m), driven by ERA5, with land surface scheme (urban, vegetated, bare soil), anthropogenic heat fluxes, and city/terrain properties. Urban/rural classification based on CORINE land cover; water bodies and grids with elevation differing by >100 m from population-weighted mean excluded.
• Population and covariates: Population density (SEDAC), land imperviousness (Copernicus), elevation (MERIT DEM), Köppen–Geiger climate class, Eurostat mortality and population by age/sex (NUTS 3), life expectancy (NUTS 2).
• Epidemiological exposure–response functions (ERFs): City- and age-group-specific temperature–mortality ERFs from Masselot et al. using a two-stage framework with quasi-Poisson DLNMs including lagged effects up to three weeks and meta-regression with spatial kriging. ERFs expressed in absolute temperature using UrbClim city mean temperatures.
• Attribution of temperature-related mortality: For each city, age group, and day, attributable fraction AF = (RR − 1)/RR from ERFs; attributable numbers obtained by multiplying AF by age- and region-specific annual all-cause mortality rates. Warm vs cold days defined relative to the age-specific minimum mortality temperature (MMT). Heat/cold extremes defined as warmest/coldest 2% of days.
• UHI impact definition: Difference between urban-average and rural-average attributable mortality within each city domain, computed daily. Aggregated as seasonal/annual averages and for heat/cold extremes. Analyses across imperviousness and population density gradients to capture intra-urban variability. Additional age-standardized analyses use the 2013 European standard population.
• Economic valuation: Primary valuation via Value of a Statistical Life (VSL) adjusted to 2021-EUR (3.91 million EUR) using OECD PPP/CPI. Mortality impacts (counts) converted to economic impacts per adult city inhabitant per year. Comparative benchmarks include PM2.5- and ozone-related mortality costs (valued with the same VSL) and urban costs (rent, public transport). Supplementary sensitivity via Value of a Life Year (VOLY) with YLL (base 46,000 2021-EUR per life year), acknowledging wide uncertainty.
• Statistical analyses: Correlations (Spearman rank) between UHI impacts and city characteristics (ERF RRTmin/RRTmax, number of warm days, UHI magnitude, age structure, latitude/longitude, average temperature). Uncertainty propagated via 1,000 Monte Carlo simulations of ERFs to derive CIs for city-specific impacts and medians across cities.

• Acute impacts: Across 85 cities, during heat extreme days (warmest 2%), UHI increases mortality risk by a median 0.25 (CI: 0.21–0.27; IQR: 0.18–0.29) additional deaths per 100,000 population per day, a 45% median increase relative to rural areas. During cold extreme days (coldest 2%), UHI decreases mortality risk by a median 0.05 (CI: 0.04–0.07; IQR: 0.02–0.11) deaths per 100,000 per day, a 7% decrease.
• Annual net effect: UHI exhibits a weak protective net annual impact in most cities (70 of 85; 90% CI: 60–75). Median 2.8 fewer deaths per 100,000 people per year (CI of median 1.7–3.5; IQR across cities 0.7–4.4) in urban vs rural temperatures due to the high frequency of cold-to-mild days. Most protective annual effects: Glasgow (−16; CI: −18 to −14), Porto (−12; CI: −14 to −11), London (−11; CI: −13 to −9) deaths per 100,000/year. Most adverse: Trieste (+5; CI: 2–7), Genoa (+4; CI: 1–8), Bologna (+3; CI: 1–5).
• Intra-urban gradients: Mortality risk change with imperviousness is steepest near the urban–rural boundary; median risk differential of 67% (IQR: 46–91%; 0.35, IQR: 0.24–0.44 additional deaths per 100,000 per day) between most (≥90%) and least (≤10%) built-up areas during heat extremes. Population-density bias would raise heat-day urban-average risk by median 6.4% (IQR: 2.9–9.4%) and reduce cold-day risk by 1.3% (IQR: 0.7–2.3%) if accounted for.
• Climate grouping: Temperate no-dry-season hot-summer cities tend toward adverse annual net UHI impacts; other groups generally show protective annual nets, with variability and overlaps across groups.
• Controlling factors: UHI impacts during heat extremes correlate with RRTmax (heat vulnerability) and during cold extremes with RRTmin (cold vulnerability). Annual net impact correlates with cold vulnerability (RRTmin), indicating dominance of cold-season effects. The number of warm days strongly influences summer impacts; UHI magnitude correlates with heat-extreme and seasonal impacts. Older age structure matters during extremes and winter. Geographically, northern/eastern and colder cities show less protective effects in winter/cold days; warmer southern/eastern cities show stronger adverse summer impacts.
• Age standardization: Standardizing to the 2013 European population increases median risk estimates by ~9%.
• Economic valuation (VSL, 2021-EUR): Median heat-related UHI impact €192 (IQR: €142–296) per adult resident per year; cold-related €−314 (IQR: €−429 to −235). Net adverse UHI-related mortality costs in 15 (CI: 10–25) of 85 cities, with greatest adverse nets in Turin and Bologna. Heat-related UHI costs are about one-fifth of PM2.5 mortality costs and roughly −1.2 times ozone-related mortality (noting differing baselines and seasonality). Variability in UHI heat costs is smaller than for PM2.5.
• YLL/VOLY sensitivity: Strong correlation with VSL-based results; VOLY-based valuations are much lower in magnitude (median ~14% of VSL-based, rising to ~35% with higher VOLY).

The analysis shows UHIs substantially elevate acute mortality risks during heat extremes, while providing modest protection during cold extremes. Because European cities experience many cold-to-mild days annually, the aggregate annual balance skews slightly protective in most cities during the 2015–2017 baseline; however, heat-day impacts are more intense, and future warming will likely increase the number of warm days, shifting the balance toward adverse annual impacts. The findings clarify the trade-off between summer risks and winter benefits and emphasize that mitigation should be seasonally targeted: measures like evapotranspirative cooling, reflective/cool roofs, and seasonally adaptive materials can reduce summer heat risks without sacrificing winter benefits. The economic results contextualize UHI-related mortality alongside other urban costs (air pollution, transport), supporting the inclusion of health and social costs in urban planning. Controlling factors—population vulnerability (RRTmin/RRTmax), frequency of warm days, UHI magnitude, and age structure—explain inter-city variability and point to tailored, city-specific strategies. Geographic patterns suggest differing adaptation levels to cold and heat across Europe. Comparisons with air pollution highlight both shared and distinct risk patterns and the importance of considering seasonality and exposure definitions.

This study provides a high-resolution, multi-city quantification of the health and economic impacts of UHIs across Europe, capturing both acute heat risks and cold-season protections. Most cities exhibit a weakly protective annual net effect during 2015–2017, yet heat extremes produce markedly higher daily risks, and projected warming is likely to tip annual nets toward adverse outcomes. Economic valuations indicate UHI-related mortality costs of the same order as other urban burdens, underscoring the need to integrate health costs into urban design and policy. Future research should incorporate finer-grained social vulnerability metrics, indoor vs outdoor exposures, population mobility, nighttime heat effects, and compound events (prolonged heatwaves/cold spells). Scenario analyses under climate change and evaluations of seasonally optimized mitigation strategies can guide healthier, more equitable, and climate-resilient urban development.

• Epidemiological ERFs are city-level averages by age group and may not capture differences in healthcare access, housing characteristics, or socioeconomic disparities between urban and rural communities or within cities.
• Indoor exposures, building thermal properties, air conditioning/ventilation, and daily mobility patterns (including commuting and tourism) are not explicitly modeled, potentially biasing exposure estimates.
• Nighttime temperature effects, known to be important for heat-related mortality, are not treated separately; analyses are based on daily mean temperatures.
• Compound and prolonged events (long heatwaves or cold spells) and timing within the season (early vs late) are not explicitly considered beyond lag structures in DLNMs.
• UHI risk is defined as urban–rural differences, whereas air pollution comparisons use thresholds that may also be exceeded outside cities; this affects comparability.
• Economic valuation via VSL assumes equal life expectancy at death across causes and does not explicitly weight by remaining life expectancy; VOLY/YLL estimates carry substantial uncertainty and ethical considerations.
• Population exposure biases (e.g., higher UHIs in more deprived neighborhoods) and correlations with vulnerability may compound risks beyond what city-average ERFs reflect.