Abstract: In order to explore the impact of anthropogenic heat emissions on the urban ecological environment, the central urban area of Guangzhou was taken as the research area, the temporal and spatial evolution of urban anthropogenic heat emissions from 2004 to 2020 were explored using the Landsat remote sensing data and the surface energy balance model, and the spatial-temporal characteristics of anthropogenic heat emissions and the norms of the landscape pattern change were analyzed with the transfer matrix and the landscape pattern index. The following results were obtained. First, in terms of time, rapid expansion of anthropogenic heat emission area was the main feature from 2004 to 2014 and dramatic increase of anthropogenic heat emission intensity was the main feature from 2014 to 2020; in terms of space, the intensity of anthropogenic heat emission in the southern part was higher than that in the northern part. Second, as the transfer matrix showed, the anthropogenic heat emission in the central area of Guangzhou changed significantly during the study period. The area of low-emission zone (decreased by 15.72%) in central area changed the most, the area of medium-emission zone (increased by 14.51%) underwent less change, and the area of high-emission zone (increased by 0.28%) changed the least. Transfer from low-emission zone and transfer into middle-emission zone were the main forms, accounting for 97.36% and 98.43% respectively. Third, the contagion index (CONTAG) increased, while the shannon's diversity index (SHDI) and the modified simpon evenness index (MSIEI) decreased. Consequently, the landscape pattern heterogeneity and the degree of fragmentation of anthropogenic heat emission increased. Moreover, there were obvious differences in the evolution of the landscape pattern in zones of different levels of anthropogenic heat emission. The landscape pattern of low- and low-medium-emission zones tend to be stable and regular while the landscape pattern of medium-, medium-high- and high-emission zones tend to be fragmented and irregular.