Spatial-scale dependent risk factors of heat-related mortality: A multiscale geographically weighted regression analysis

Extreme heat is a leading cause of weather-related human mortality throughout much of the world, posing a significantly heavy burden on the development of healthy and sustainable cities. To effectively reduce heat health risk, a better understanding of where and what risk factors should be targeted for intervention is necessary. However, little research has examined how different risk factors for heat-related mortality operate at varying spatial scales. Here, we present a novel application of the multiscale geographically weighted regression (GWR) approach to explore the scale of effect of each underlying risk factor using Hong Kong as a case study. We find that a hybrid of global and local processes via multiscale GWR yields a better fit of heat-related mortality risk than models using GWR and ordinary least squares (OLS) approaches. Predictor variables are categorized by the scale of effect into global variables (i.e., age and education attainment, socioeconomic status), intermediate variables (i.e., work place, birth place and language), and local variables (i.e., thermal environment, low in-come). These findings enrich our understanding of the spatial scale-dependent risk factors for heat-related mortality and shed light on the importance of hierarchical policy-making and site-specific planning processes in effective heat hazard mitigation and climate adaptation strategies.

Automated summary

A study using a multiscale geographically weighted regression approach examined the scale of effect of heat-related mortality risk factors, finding that a combination of global and local processes provides a more accurate prediction of heat-related mortality risk.