Snow depth and spring temperature dominate the spring phenological shifts and control growing season dynamics on the Tibetan Plateau

To explore the spatio-temporal variability of vegetation phenology and its drivers under rapid climate change on the Tibetan Plateau (TP) over the past four decades, a monthly normalized vegetation index (NDVI) dataset was constructed for the TP from 1982 to 2020 using pixel-level univariate linear regression models based on GIMMS NDVI and MODIS NDVI. The extended NDVI dataset passed a consistency check (R² = 0.99, P < 0.001). From here, the optimal thresholds for retrieving vegetation phenology were determined based on phenological observation data. Spatial differences among the pathways of influence of how climate change affected vegetation phenology were analyzed using lagging correlation analysis and structural equation modeling. Based on the extended dataset, the optimal thresholds for the start of the growing season (SOS) and the end of growing season (EOS) were 0.30 and 0.80, respectively. The SOS had a three-month lag in response to snow depth and a one-month lag in response to temperature. The variation in SOS was mainly influenced by a negative effect of snow depth in the central-western TP and a negative effect of spring temperatures in the south-eastern TP, while the variation in EOS was mainly influenced by a positive effect of fall temperature in the central-western TP and a positive effect of SOS in the south-eastern TP. Additionally, phenological changes displayed altitude dependence in response to climate change, with the reduction in snow depth delaying the SOS more at higher altitudes than at lower altitudes. This can be attributed to elevation-dependent warming, where snow depth is reduced more quickly at higher altitudes. Thus, alpine ecosystems at higher elevations on the TP may be particularly sensitive to snow cover changes under future warming scenarios.