The accelerating degradation of permafrost on the Qinghai–Tibet Plateau (QTP) is a critical driver of regional and global climate change. However, conventional models often limit our understanding by overlooking thermal memory and failing to deconstruct complex spatial dynamics. This study introduces a novel diagnostic framework that pairs a thermal-memory-aware machine learning model with a multi-scale spatiotemporal analysis system to overcome these limitations. Our reconstruction from 1960 to 2020 reveals that the total permafrost area shrank by approximately 16% from its peak, while the mean active layer thickness (ALT) deepened, with degradation accelerating sharply after the 1980s. Vertically, we identify a systematic misalignment between the elevation of maximum permafrost stability (peak area) and maximum thermal sensitivity (peak ALT), the magnitude of which serves as a robust indicator of basin-scale vulnerability. Horizontally, we reveal a critical spatiotemporal mismatch: the geometric centroid of the permafrost area remains relatively stable, while its thermal center of mass exhibits large, volatile oscillations. This decoupling is driven by the contrast between rapid degradation at the warm, wet margins and the anchoring effect of the vast thermal inertia in the cold, arid core. Ultimately, our study reveals that permafrost degradation is a complex, multi-scale process rather than a uniform retreat. The diagnostic framework and the identified spatiotemporal decoupling provide a new perspective for assessing the stability and vulnerability of cryospheric systems in a warming world.
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号