Abstract: Understanding vegetation phenology of temperate regions and its responses to climate change is crucial for diagnosing the responses of ecosystems to regional climate change and projecting regional ecosystem carbon cycle. Based on MODIS EVI data from 2000 to 2018, we used asymmetric Gaussian (A-G) function and dynamic threshold method to extract the phenological parameters of forest and grassland in Hedong, Gansu Province. Combining meteorological data, we explored the response of vegetation phenology to climate change. A significant difference in phenological parameters was found between forest and grassland, both of which with a trend of advancing SOS, but advancing EOS for forest and delaying EOS for grassland. SOS displayed an advanced trend in 61% of the total area, a delayed trend of EOS in 41%, and a lengthened trend of LOS in 53%. With increasing elevation and latitude, SOS, EOS, and LOS presented a delayed, advanced, and shortened trend, respectively, but such a trend was weakening. The degrees and directions of the effects of preseason climate on SOS and EOS were different. Warming autumn and winter delayed SOS, while warming spring advanced SOS. Increasing precipitation in spring and summer advanced EOS, but warming autumn delayed EOS. The effect of minimum temperature was more critical in Hedong area. Forest and grassland differed in the degree of their response to climate change. The correlation between SOS of forest and the minimum temperature changed from positive to negative in March, and the grassland was only positive in February and negative in other months.  Precipitation only had a significant impact in some individual months. The relationship between EOS of forest (grassland) and precipitation and minimum temperature changed from positive to negative in September (July).

Key words: vegetation phenology, climate change, partial least squares regression, climate sensitivity, Hedong area.