CR 49:45-58 (2011)  -  DOI: https://doi.org/10.3354/cr01009

ABSTRACT: Assessment of spatial variability in the long-term urban heat island (UHI) is severely restricted by coverage and availability of measurements. This also limits the opportunities to analyse its relation to surface characteristics. In this study we therefore introduce a new proxy dataset derived from floristic mapping. The basic assumption is that the species composition in an area reflects climatologic conditions of a certain time period. Ellenberg indicator values for temperature (EITs) were processed to summarise the overall temperature preferences of the occurring plant species. The EITs showed a clear heat island pattern, were highly correlated with existing measurements, and showed increased values in densely built urban classes. Hence, they are considered suitable as UHI proxies. Further, they were related to a large number of typical UHI predictors. The normalised difference vegetation index (NDVI) showed the strongest correlation with the derived pattern and was comparatively robust towards cloud contamination. Urban morphology also explained a noticeable proportion of the overall variance. All predictors explained more than 2/3 of the overall spatial variability, while the redundancy was high. Therefore, the predictors at least allow qualitative statements about the differential exposure to heat-related risks in cities.

KEY WORDS: Floristic mapping · Urban heat island · Ellenberg indicator values · UHI predictors · Local climate zones · Infrared thermography