Nonlinear and lagged associations between active-layer thickness and vegetation greenness in the source reach of the Yangtze River basin from 1990 to 2024

Permafrost degradation can alter alpine vegetation through changing hydrothermal conditions, but nonlinear relationships between active-layer thickness (ALT) and vegetation and their temporal lags remain poorly quantified on the Qinghai–Tibet Plateau. We integrated a cross-sensor-calibrated 30-m Landsat normalized difference vegetation index (NDVI) time series (1990–2024) with gridded permafrost and environmental data for the Source Reach of the Yangtze River Basin (SRYRB). Segmented regression, cross-correlation analysis, and Geodetector were used to characterize statistical thresholds, lagged associations, and their spatially heterogeneous environmental context. From 1990 to 2024, mean ALT increased by 0.135 m decade −1 , while approximately 65.8% of extremely stable permafrost degraded. NDVI increased by 0.0018 yr −1 overall, with a slightly higher rate in seasonally frozen ground (0.0021 yr −1 ) than in permafrost (0.0017 yr −1 ). The positive NDVI trend weakened when ALT reached approximately 1.45–1.65 m, and pixel-level turning events were most frequently associated with cumulative ALT thickening of 0.3–0.5 m. The strongest ALT–NDVI association occurred at zero lag across 58.0% of the study area and at lags of 1–3 years elsewhere. TEMP had the largest individual q statistic, whereas its pairwise combinations with PRECIP, LULC, ALT, and MAGT showed greater joint explanatory power. Continued permafrost degradation may move more of the SRYRB into ALT ranges associated with weaker greening, providing a basis for monitoring of vegetation–permafrost change in Tibetan Plateau.