The Spatial Variations of Urban Land Surface Temperatures: Pertinent Factors, Zoning Effect, and Seasonal Variability

Abstract

Remote sensing of urban land surface temperatures (LSTs) has been conducted based largely on pixel-by-pixel correlation with land use and land cover (LULC) types. Few studies have examined the spatial variations of LST within land use zoning polygons, in spite of its significance on the knowledge of environmental implications or planning practices. This study aimed to analyze the spatial patterns of LSTs and to explore factors contributing to the LST variations in the city of Indianapolis. Four Terra's ASTER images, representing distinct seasons, were used in conjunction with other types of spatial data for the analysis. The potential factors were grouped into the categories of LULC composition, biophysical conditions, intensity of human activities, and landscape pattern. Statistical analyses were conducted to determine the relative importance of each group of the variables. Moreover, the spatial variations of LST were examined at both the residential and general zoning levels, so that the environmental effect of urban planning on LST may be assessed. By analyzing the mean and standard deviation values of normalized LSTs, the seasonal dynamics of LST were finally studied. Results show that the biophysical variables were most significant in explaining the spatial variations of LST. At both zoning levels, LST possessed a weaker relationship with the LULC compositions than with the biophysical variables. Principal component analysis further indicates that the cumulative variance was always larger in residential zoning, implying that the factors contributing to the LST variations in general zoning might be more complex than those for the residential zoning. An interesting finding of this study was in the relationship between LST and the landscape metrics of zoning polygons. It suggests that smaller residential zoning polygons were associated with larger temperature variations, and that the more complex in shape a residential zoning category was, the more intrapolygon variation of LST tended to be. These correlations, however, did not exist in the nonresidential zoning categories. The spatial pattern of LST in Indianapolis may be characterized as concentric in the central part of the city, a hot ring along the Highway 465, and several hot corridors along the radial highways outward to the countryside. The seasonal fluctuation of LST was weak in the central part, but increased towards the countryside. Due to the amount of anthropogenic heat, land use zones with less human activities were found to have a strong seasonal variability, whereas the zones with intensive human activities fluctuated less in LST.