Do urban green and urban blue mitigate heat-related excess mortality? Evidence from Lisbon, Portugal

It has been shown that elevated temperature can exert adverse effects on human health and well-being. Especially increased levels of mortality during periods of extreme heat have been highlighted in the literature. Several studies have shown that different cities and population groups exhibit different responses to heat. The underlying reasons for these differences are only partially understood. The demographic composition as well as the spatial structure, the degree of urbanisation, population density or the urban design and morphology might be crucial in shaping the atmospheric effect. Given the superimposed urban heat island effect, urban areas seem to be particularly vulnerable toward heat stress. The shape and magnitude of the urban heat island is rather heterogeneously developed throughout the urban landscape. While bigger and more densely build and populated urban areas generally show higher excess temperatures compared to their rural surroundings, there are also intra-city specifications of the UHI with warmer and less warm micro- and meso-environments. The primary objective of this study was to assess the influence of urban characteristics on heat-related excess mortality on a small-scale intra-urban level. Special attention was given to the mitigation potential emanating from urban vegetation and urban water bodies. In a first step we investigated the association between day- and night-time land surface temperature (LST), urban green, urban density and coastal proximity. Following, the data was stratified by spatial criteria, i.e. amount of urban green, density and proximity to the coast. Additionally, we calculated the Universal Thermal Heat Index (UTCI) in order to assess the combined effects of different meteorological parameters, i.e. temperature, humidity, wind-speed, and radiation on the human heat balance. In order to assess the mortality-atmosphere relationship, the modelled atmospheric information was combined with the spatially stratified mortality data using different kinds of Poisson regression models adjusting for various confounders. For determining the lag structure of heat effects, we applied Distributed Lag Non-Linear Models (DLNM). Subsequently, we used non-parametric generalized additive models (GAMs) including interaction terms in order to allow for interaction by spatial category. For quantifying heat effects we modelled GAMs having segmented relationships were modelled. Land surface temperature shows to be associated with heat mortality and findings furthermore demonstrate a mitigating effect of urban green. Similarly, distance to the coast seems to have an effect on heat-related mortality with decreased heat-related mortality in areas located within closer proximity to the coast line.