Intra-urban spatial patterns of societal risk and vulnerability to heat waves

A spatial analysis of societal vulnerability and risk to excessive heat were conducted for two urban areas in the United States: Phoenix, AZ and Philadelphia, PA. To identify populations at risk for adverse health outcomes during extreme heat events we analyzed two cities in terms of their climatological, environmental and societal characteristics. Geographic Information Systems (GIS) and spatial statistics were used to identify spatial patterns of biophysical and social factors contributing to the heat-related morbidity and mortality. Analysis presented in this work, expands upon previous univariate ecologic studies and takes a multivariate approach to investigate the relative importance of neighborhood level heat exposure, socio-economic vulnerability and neighborhood stability to heat morbidity or mortality. We also compare the sensitivity and specificity of two multivariate statistical methods; a traditional Generalized Linear Model and a machine learning Classification and Regression Tree algorithm. Philadelphia neighborhoods with heat mortality cases are characterized by low housing values and a higher fraction of black residents. Such characteristics interestingly also describe neighborhoods with a disproportionate burden of lead poisoning cases. Phoenix neighborhoods with high heat distress calls tend to have a higher fraction of socially isolated, black, hispanic, disabled, and linguistically isolated residents. Enhanced urban heat island effect temperatures measured from remotely sensed satellite imagery also explains heat distress call variability across the city. A similar societal risk and vulnerability analysis may be particularly beneficial for municipalities without a heat wave preparedness plan or a comprehensive heat watch/warning system.