Mapping Social Vulnerability and Exposure Parameters to Extreme Heat Events in Missouri
MetadataShow full item record
Extreme heat events are considered to be the natural disasters with most deaths in the United States. Due to the global warming phenomenon, extreme heat events are expected to increase in frequency, duration, and intensity in the coming years. Multiple heat-specific social vulnerability studies have been published in the literature recently but many focused on metropolitan areas with very few studies including rural areas. In this study we aimed to apply two common methods of social vulnerability assessment to study county-level statewide extreme heat vulnerability and propose a new methodology for vulnerability prediction. The state of Missouri was chosen for the extreme heat vulnerability assessment because Missouri is expected to have more extreme heat events in the future, and because social vulnerability has not been studied recently for the entire state. Fifteen heat vulnerability indicator variables were included in the analysis, namely: population density; age over 65; age under 5; disabled over 65; poor; living alone; race other than White; low education; low English skills; no vehicles; unemployed; outdoor workers; land cover; crime index; and heat exposure. The two methods were; the Extreme Heat Risk Index (EHRI) using the percentile ranking methodology, and the Combined Exposure Vulnerability Index (CEVI) using the principal component analysis statistical method. The study also proposed a new method, Predictive Heat Vulnerability (PHV), using the decision tree analysis. The spatial distribution of heat vulnerability was different between the three different methods, but in general, the southeast counties of Missouri had the highest heat vulnerability scores for all methods. Urban counties had low vulnerability scores. The coincidence matrix analysis was carried out to evaluate the performance and compare the different methodologies by comparing the results of each method to the hyperthermia morbidity rates per county. The overall coincidence percentage was 32.6% for the EHRI, 37.4%, for the CEVI, and 83.2% for the PHV method. The PHV method using decision tree analysis outperformed EHRI and CEVI in extreme- heat social vulnerability assessment. The decision tree-based approach provided a reliable, easy to implement and interpret approach for identifying heat vulnerable counties.
Principal Component Analysis