LAUSR

Abstract The effects of urban morphology on urban climates have been investigated mainly in terms of average urban spatial characteristics represented by the mean values of urban spatial variables, such as average building height of an urban area of interest. This study examines the potential impacts of spatial heterogeneity on surface urban heat island (SUHIs) in comparison with those of average spatial values through a case study of the Greater London and Seoul areas. The standard deviation measure of spatial variables was used to indicate spatial heterogeneity, specifically spatial variation within an urban area. Methods were developed to extract a comprehensive set of spatial characteristics from the GIS data. Land surface temperatures (LSTs) of four different time periods were used to analyze the SUHIs distribution across the two cities. Correlation analyses were conducted to investigate correlations among spatial variables, relative importance of variables, and relationship direction between spatial variables and LSTs. Correlation analyses within spatial variables show high correlations, which may present confounding effects in statistical models. Hence, the Genizi method was selected to reliably quantify the effect of highly correlated spatial measures on the LST. The final statistical results indicate the heterogeneity measures of the green space ratio, building coverage ratio, and canyon H/W ratio had a significant impact on the LST.