Objective Under the impact of urbanization, the spatial unevenness of thermal environment within urbanized area leads to more prominent spatial heterogeneity of urban plant phenology. The spatial unevenness of leaf coloring date (LCD), the end of deciduous trees’ growing season has a far-reaching influence on the annual carbon sequestration of urban vegetation and the primary productivity of urban ecosystem, also causes spatial variation in the autumn seasonal aspect. Therefore, the spatial heterogeneity of leaf senescence is one critical cut-in point for monitoring urban ecology and landscape dynamics, and deserves profound exploration. Regarding this point, this research aimed to explore the spatial heterogeneity characteristics of leaf coloring date (LCD) and the phenological response to thermal environment of underlying surface in the highly urbanized Beijing City.

Method This research selected five autumn-color tree species as the research objects that distributed in 9 green spaces along the urban-suburb gradient in Beijing’s northwestern urban area. We applied ground phenological observation to collect the LCD data during 2017−2019, and collected thermal environment data by retrieving MODIS land surface temperature (LST). Then, we analyzed the spatial difference of LCD and its correlation with the autumn thermal environment.

Result (1) the LCD of various tree species in Beijing’s urban area occurred during mid-October to early December and clustered in early and mid-November, with the species order of LCD ranked as follows: Fraxinus pennsylvanica, Acer truncatum, Ginkgo biloba, Metasequoia glyptostroboides, Salix matsudana. The LCD for female ginkgo trees occurred significantly earlier than male ones. (2) The LCD of various species gradually advanced from the central urban area in 2nd Ring to the outskirts beyond 5th Ring, with the spatial range of LCD reaching (10.1 ± 0.3) d in average. A significant phenological difference existed among different sample plots and the LCD outside of 4th Ring occurred significantly earlier than that within 3rd Ring. (3) The LCD of various species showed a significantly positive correlation with the mean LST in autumn-LST_a (P < 0.01), which means the high accumulation of surface heat in autumn can drive leaf senescence to delay in Beijing’s urban area. The response sensitivity of LCD to the spatial variation of LST_a was (4.11 ± 0.83) d/℃ with F. pennsylvanica and M. glyptostroboides as the most sensitive species.

Conclusion The leaf coloring date shows a significant response to the spatial variation of surface thermal environment in Beijing’s urban area. UHI effect represents the microcosms of long-term climate change, and the phenological response to thermal environment variation in urban area can reflect the potential impact of future climate change on plant phenology, i.e., a space-time substitution.