Journal of Applied Remote SensingAnalysis of human factors on urban heat island and simulation of urban thermal environment in Lanzhou city, China
|Format||Member Price||Non-Member Price|
Urban heat island (UHI) effect is a global phenomenon caused by urbanization. Because of the number and complexity of factors contributing to the urban thermal environment, traditional statistical methods are insufficient for acquiring data and analyzing the impact of human activities on the thermal environment, especially for identifying which factors are dominant. The UHI elements were extracted using thermal infrared remote sensing data to retrieve the land surface temperatures of Lanzhou city, and then adopting an object-oriented fractal net evolution approach to create an image segmentation of the land surface temperature (LST). The effects of urban expansion on the urban thermal environment were quantitatively analyzed. A comprehensive evaluation system of the urban thermal environment was constructed, the spatial pattern of the urban thermal environment in Lanzhou was assessed, and principal influencing factors were identified using spatial principal component analysis (SPCA) and multisource spatial data. We found that in the last 20 years, the UHI effect in Lanzhou city has been strengthened, as the UHI ratio index has increased from 0.385 in 1993 to 0.579 in 2001 and to 0.653 in 2011. The UHI expansion had a spatiotemporal consistency with the urban expansion. The four major factors that affect the spatial pattern of the urban thermal environment in Lanzhou can be ranked in the following order: landscape configuration, anthropogenic heat release, urban construction, and gradient from man-made to natural land cover. These four together accounted for 91.27% of the variance. A linear model was thus successfully constructed, implying that SPCA is helpful in identifying major contributors to UHI. Regression analysis indicated that the instantaneous LST and the simulated thermal environment have a good linear relationship, the correlation coefficient between the two reached 0.8011, highly significant at a confidence level of 0.001.