Increasing control of vapor pressure deficit on the end of the growing season in boreal forest
Abstract
Understanding the dynamics of boreal forest phenology in response to climate change is critical for predicting land-atmosphere interactions at high latitudes. Utilizing multi-source remote sensing time-series data from 1982 to 2022, this study systematically quantifies the response intensity and spatial heterogeneity of the end of the growing season (EOS) to the start of the growing season (SOS), temperature (TMP), precipitation (PRE), solar radiation (SRAD), and vapor pressure deficit (VPD) through trend analysis, ridge regression, and geographical detector method. Our results reveal a significant shift in the long-term trend of EOS around 2001, transitioning from a delayed trend (0.17 days/year, p < 0.05) to an advanced trend (0.16 days/year, p < 0.1). Spatial driver analysis further indicates that within the 45°N–70°N latitudinal belt, VPD has superseded temperature as the dominant factor controlling EOS variations, with its explanatory power increasing significantly from 10.4% to 26.2%. Crucially, employing a geostatistical framework, this study identifies for the first time two ecologically significant VPD thresholds (0.335 kPa and 0.885 kPa), which are substantially lower than the continental-scale average, underscoring the heightened sensitivity of boreal forests to atmospheric aridity. The findings suggest that persistently rising VPD may weaken the regional carbon sink capacity by triggering earlier EOS and shortening the growing season—a key mechanism currently underrepresented in existing global land surface models. This research provides essential spatially explicit parameters for refining high-latitude ecological process models and highlights the urgency of integrating remotely sensed phenological threshold behaviors into climate prediction frameworks.
Fields of Study & Subjects
Remote Sensing in Agriculture
Plant Water Relations and Carbon Dynamics
Climate variability and models
Publication details
- Journal / source
- Figshare
- Publication date
- 1 January 2026