A statistical model for the urban heat island and its application to a climate change scenario

Abstract

AbstractA linear statistical model relating the nocturnal urban heat island (UHI) intensity of Hamburg with meteorological conditions is constructed from observations taken by the German Meteorological Service (DWD). To find the appropriate predictors the relationship between different meteorological variables and the UHI of Hamburg is analyzed. Results and physical plausibility suggest that cloud cover, wind speed and relative humidity are the relevant variables and can be used to construct a statistical model. The parameters for the statistical model are determined with the generalized least square method. With the help of the statistical model up to 42% of the UHI variance can be explained. The statistical model is then used to statistically downscale results from climate runs of the regional climate models (RCM) REMO and CLM. Both RCMs were driven with the A1B SRES emission scenario runs of the global climate model ECHAM5/MPI‐OM. The resulting values for the future UHI are analyzed with respect to monthly averages and the frequency distribution. Results show that changes in the UHI are different for the different months. Significant change (decrease of UHI) in the results of both RCMs and for both realizations of the A1B scenario can be found for April in at the middle and the end of the century and in December at the end of the century. For the summer months which are most relevant to the development of adaption strategies the results differ between the RCMs. REMO results show no significant changes for the summer, while analyses of CLM suggest significant increase in July and August. The frequency distribution of the summer UHI shows no significant changes for REMO and only in one realization of CLM a significant increase in moderate and strong UHI days can be found for the end of the century. Copyright © 2011 Royal Meteorological Society