ABSTRACT Driven by rapid global warming, the irreversible degradation of permafrost has become a primary engine of hydrological reorganization, carbon mobilization, and geohazard development across the Arctic and the “Third Pole.” Due to the extreme spatiotemporal heterogeneity of permafrost systems, no single detection technique can fully capture its dynamics. This study systematically reviews recent advances in permafrost detection and simulation and proposes a process‐oriented cross‐scale framework for data acquisition and integration. The framework encompasses microstructural CT/NMR imaging, shallow high‐frequency in situ sensors, regional/global‐scales geophysical imaging (ERT/GPR/TEM), and regional remote sensing (InSAR/LiDAR), clarifying the applicability boundaries and dominant sources of uncertainty for each method in monitoring ice–water phase changes, unfrozen water dynamics, pore‐scale hydrology, and surface deformation. Furthermore, the coupling of detection data with numerical models—including physics‐based approaches (e.g., CryoGrid and GIPL) and AI‐enabled data assimilation—enables cross‐scale predictions from pore to regional scales, substantially improving the reliability of active layer thickness, ground‐temperature evolution, and permafrost‐degradation risk forecasts. Finally, the deep integration of detection technologies with environmental models and artificial intelligence highlights the future direction of permafrost research: establishing quantitative, dynamic Earth system monitoring, advancing from “method stacking” toward mechanistically testable and prediction‐ready science, and providing a robust scientific basis for ecological and engineering adaptation strategies in polar and high‐altitude regions.
Yang et al. (Thu,) studied this question.