Objective  Using the method of extracting vegetation phenology by remote sensing,taking Xiaoxing’an Mountains of northeastern China as the study area, this paper analyzes the temporal and spatial changing characteristics of forest vegetation phenology and the response of forest vegetation phenology change to climate change.

  Method  Based on GIMMS NDVI 3g image, the forest vegetation growth curve was fitted by univariate sixth order polynomial, and the phenological parameters of forest vegetation SOS, vegetation EOS and vegetation LOS in Xiaoxing’an Mountains from 1982 to 2015 were extracted by pixel dynamic threshold method. Using ArcGIS software, the temperature, precipitation and sunshine duration data were analyzed pixel by pixel with vegetation phenological parameters to obtain partial correlation coefficients between phenological parameters and meteorological factors.

  Result  (1) The characters in spatial distribution of multi-year mean value of vegetation phenology showed a kind of gradual change from northwest to southeast, including the advance of vegetation SOS, the delay of vegetation EOS and the extension of vegetation LOS. (2) The SOS of forest vegetation in Xiaoxing’an Mountains was concentrated among 112.1th−128.3th day of the year, with an abrupt change around 1998. Before 1998, it showed a significant advancing trend (R² = 0.284, P = 0.028), and after 1998, it showed a non-significant delaying trend (R² = 0.002, P = 0.86), resulting in a non-significant change during the whole analysis period (1982−2015), with a rate of change at 0.12 d per ten years (R² = 0.001, P = 0.872). The forest vegetation EOS was concentrated among 277.3th−294.8th day of the year, and showed a significant delaying trend during the whole analysis period, with a rate range of 2.33 d per ten years (R² = 0.294, P < 0.01). The LOS of forest vegetation was concentrated among 149.5−167.5 days, and showed a significant extension trend during the whole analysis period, with a rate of change at 2.45 d per ten years (R² = 0.231, P < 0.01). (3) The forest vegetation SOS was correlated with the April temperature of current year, followed by the February temperature of current year. The vegetation EOS was correlated with the precipitation in August of current year, followed by the June temperature in current year.

  Conclusion  (1) The multi-year average value of forest vegetation phenology and the multi-year average value of hydrothermal conditions in Xiaoxing’an Mountains show relatively consistent spatial characteristics. (2) The change of vegetation EOS during the study period is mainly driven by the change of precipitation in August. The decline of precipitation in August is the main reason for the significant delay of vegetation EOS. (3) The stagnation of global warming at the end of the 20th century causes the temperature in February to change from a significant increase to a non-significant decrease around 1998, causing a sudden change in the trend of vegetation SOS around 1998, leading to a non-significant change in vegetation SOS during the whole period.