Phenological Divergences in Vegetation with Land Surface Temperature Changes in Different Geographical Zones

Exploring the phenological divergences in vegetation caused by global climate change is of great significance for gaining a deeper understanding of the carbon cycling process in natural ecosystems. However, in many existing studies, the response of the start of the growing season (SOS) and the end of the growing season (EOS) to temperature exhibited multi-scale inconsistencies. In view of this, we took 259 Chinese urban agglomerations and their rural regions as the study areas, using MODIS phenological products (MCD12Q2), land surface temperature (LST) datasets, altitude, and latitude as data, and explored the phenological divergences in vegetation with LST changes in different geographical zones through box plots, linear regression models, and Spearman’s correlation analysis. The mean SOS and EOS in urban areas were both the earliest on approximately the 100.06th day and 307.39th day, respectively, and were then gradually delayed and advanced separately along an urban–rural gradient of 0–25 km. The divergences in vegetation phenology were no longer significant in rural areas 10 km away from urban boundaries, with change amplitudes of less than 0.4 days. In high latitude (40–50° N) regions, the correlation coefficients between the SOS and EOS of various urban agglomerations and LST were −0.627 and 0.588, respectively, whereas in low latitude (18–25° N) regions, the correlation coefficients appeared to be the opposite, being 0.424 and −0.426, respectively. In mid- to high-altitude (150–400 m) areas, LST had a strong advanced effect on SOS, while in high-altitude (above 1200 m) areas, LST had a strong delayed effect on EOS, with the R 2 values all being above 0.7. In summary, our study has revealed that within the context of varying geographical zones, the effects of LST on phenology exhibited significant spatial heterogeneity. This may provide strong evidence for the inconsistencies in the trends of phenology observed across previous studies and more relevant constraints for improving vegetation phenology prediction models.