Effect analysis of land-use pattern with landscape metrics on an urban heat island

In order to explore the effect of land uses on urban heat island (UHI), the correlation between land-use pattern and land surface temperature (LST) is quantitatively analyzed separately in the urban and suburban areas. In the case of the Changsha–Zhuzhou–Xiangtan city group of China, the LST is retrieved by the Landsat-5 TM image acquired on September 2009, and the land-use composition and configuration are described by five landscape metrics: the percentage of landscape, edge density, mean patch area, the mean Euclidean nearest-neighbor distance, and Shannon’s evenness index. Pearson’s correlation coefficients and an ordinary least squares multiple linear regression model are utilized to evaluate the correlation between each landscape metric and LST, respectively, and to select the optimal combination of landscape metrics to describe the LST at an optimal scale in the whole and the urban/suburban parts of the study area. According to the conclusions, it is suggested that, in order to restrain the aggravating UHI and its influence, all land uses should be reasonably distributed, and the proportion of critical land-use types should be optimized.