What question did this study set out to answer?

This research aims to understand the relationship between heat exposure and social vulnerability in urban areas, particularly during heatwaves.

April 21, 2026Open Access

Identifying Vulnerable Population under Heat Stress: A Data-Driven Urban Risk Assessment Using Remote Sensing, Citizen Science and Census Data

Key Points

This research aims to understand the relationship between heat exposure and social vulnerability in urban areas, particularly during heatwaves.
Integrated 30 parameters into a 2.5D urban model using remotely sensed and citizen science data.
Employed machine-learning techniques, including Random Forest and Partial Least Squares regression, for heatwave exposure predictions.
Assessed social vulnerability through 2022 census data, focusing on age structure as a core indicator.
Developed a Heat Vulnerability Index (HVI) highlighting urban areas at highest risk during heat events.
Identified residents over 65 as particularly sensitive to heat, with younger cohorts also at risk.
Revealed spatial patterns of heat risk, providing insight for municipal adaptation strategies.

Abstract

Cities are increasingly confronted with rising temperatures and more frequent heatwaves, posing significant challenges for urban resilience and spatial planning. In this study, we investigate heat exposure and its interaction with social vulnerability in the cities of Augsburg, Zwickau and Hamm, Germany, based on remotely sensed, geo- andcensus data as well as citizen science air temperature data. A comprehensive 2.5D urban model was developed, integrating 30 different parameters such as vegetation shares, impervious surfaces, and building parameters like density or floor area ratio, as well as spectral characteristics like Normalised Difference Vegetation Index (NDVI) or satellite-based Land Surface Temperature (LST). In combination with citizen science (‘gathered by the public’) air temperature data measurements, it enabled 100m resolution modelling of heatwave peaks. Machine-learning techniques (Random Forest, RF) and multivariate techniques, particularly Partial Least Squares (PLS) regression, performed convincingly in handling the uncertainties inherent in citizen-science air temperature loggers and ensured robust, area-wide heat wave exposure predictions. In parallel to modelling heatwave exposure, social vulnerability was assessed using the 2022 census data in Germany. Age structure serves as the core vulnerability indicator, with residents above 65 identified as particularly sensitive, and younger cohorts considered moderately vulnerable compared to the remainder of the population. Merging the information on social vulnerability with the modelled heat exposure, an exposure weigheted Heat Vulnerability Index (HVI) was created at the grid level, revealing spatial patterns of heat risk across the three cities.TheHVI highlights the most heat-endangered urban areas where residents are most likely to suffer from heat. The findings provide a robust evidence base for municipal adaptation strategies and risk-sensitive planning. The study underscores that meaningful heat-risk assessment is only possible throughintegration of social vulnerability, and that its spatial granularity is crucial for designing effective, locally adapted measures to enhance urban resilience.

Bookmark

View Full Paper