Using data and findings from natural and social sciences to assess urban heat vulnerability: a comparison of different methodologies.

Climate change is causing temperatures around the globe to rise, leading to an increase in the frequency and intensity of hot days and heatwaves. In urban areas, this trend is further exacerbated by urban characteristics, such as the high building density and degree of sealing, the high concentration of anthropogenic heat sources or the reduced outgoing radiation. Extreme heat puts a strain on health, especially for elders and people with pre-existing illnesses. For effective and targeted prevention of heat-related morbidity and mortality, information on the spatial variance of peoples exposure and sensitivity, but also their adaptability towards heat can be of great importance. A common practice for determining the distribution of vulnerable population groups within a city or an area is to construct a spatial Heat Vulnerability Index (HVI) based on findings and data from natural and social sciences, including e.g. socio-economic data, health data, remote sensing data, and climate data. However, there is no standardized workflow but a variety of approaches for the construction of an HVI, which may lead to significant differences in the calculated index ranks. In order to assess the impact of changes in the method design on the resulting index, we test different input data sets, weighting methods and spatial scales for the construction of a spatial HVI for the city of Graz (Austria). The input parameters for the HVI include temperature data, derived from satellite data and weather stations, as well as spatial socio-economic data that describe the populations sensitivity towards heat and the capability to adapt to high temperatures. By conducting an uncertainty analysis and a global variance-based sensitivity analysis, the partial contribution of changing input variables, chosen weighting methods and different spatial scales to the outputs variance is determined. In addition, a local sensitivity analysis compares the application of land surface temperature derived from thermal satellite imagery to the use of station temperature data for the construction of an HVI.