The urban thermal environment is an important element for the urban ecological environment, urban climate and urban disasters. This paper selected MOD11A2, the MODIS LST night data to study the thermal environment evolution in Beijing, Shanghai and Guangzhou, which are the three major cities of China in the past decade. Three methods have been applied in the paper, Landscape centroid evolution, Landscape pattern index and spatial autocorrelation. Three main conclusions have been drawn as follows. Firstly the thermal landscape distributions in the three cities have moved from the suburb to the downtown. And the evolution trend of the thermal landscape is changed from the low temperature region, sub-middle temperature region to middle temperature region, sub-high temperature region and high temperature region. Secondly, among these five types of thermal landscape, the middle temperature region is the most prevalent. The urban thermal landscape fragmentation was highest in Shanghai among the three cities, and sub-middle and high temperature region has the highest fragmentation. The urban thermal landscape dispersion was highest in Beijing, and the dispersion of low and high temperature region was higher than the other types of thermal landscapes. Thirdly, thermal environment spatial autocorrelation analysis showed that the high-high temperature zones were adjacent, low-low temperature areas were adjacent, which are the main types in the temperature spatial agglomeration. And for Beijing and Guangzhou city, the high-high temperature zone located in the south of the city, the low-low temperature region located in the north. While, the spatial autocorrelation distribution of LST in Shanghai is more complicated. The distribution areas of high-high temperature varied among the three cities in the past decade. In Beijing, the distribution area increased shortly after decreasing, and in Guangzhou, the distribution area continued to decline, which preliminary reflects the heat island effect problem aggravated in Beijing, while weakened in Shanghai and Guangzhou.
Through comparisons and analysis, the paper has provided a reference for urban planning and urban living environment improvements, but there are still some inadequacies to be further studied. Firstly, this study only selected the January night LST data in the three cities. Because the time factors, such as season, daytime and nocturne, will affect the urban heat environment pattern, the comprehensiveness of the thermal environment pattern changes need to be improved. In addition, the paper only selected the data in the period of three years, the evolution regulation of the urban thermal environment pattern is not precise. Secondly, the landscape of urban heat environment were impacted by many factors, including the pattern of landuse, urban surface construction, weather conditions, terrain, anthropogenic heat emissions factors and so on. The analysis between the urban heat environment and impact factors will help reveal the mechanism of urban heat environment and which will be studied further.