Prevailing thermal models underestimate permafrost thermal state in the Tibetan Plateau: Implications for cryosphere adaptation

Understanding thermal dynamics of permafrost is essential for maintaining the cryosphere eco-hydrological system and infrastructure. The mean annual ground temperature (MAGT) serves as a key indicator calculated by thermal models to assess the extent of permafrost degradation. However, the ability of these models to accurately represent real thermal changes, along with the uncertainties and sensitivities involved, remains unclear limiting global permafrost projection and adaptations. This study focuses on a representative permafrost basin in the Tibetan Plateau to quantify the biases, sensitivities, and uncertainties of two widely used thermal models—empirical equilibrium models and transient numerical models using the Generalized Likelihood Uncertainty Estimation (GLUE) framework. Our results reveal that thermal models originally developed for Arctic permafrost may underestimate the MAGT of permafrost on the Tibetan Plateau, particularly in mountainous areas, due to the underestimation of soil water and ice content in thick unsaturated zones. Uncertainty analysis indicates that variations in thermal parameters can induce a MAGT variation range exceeding 6 ℃, driven primarily by the parameter associated ground surface temperature to air temperature during freezing days. The models show different sensitivities to climate warming, under 1 ℃ warming, the empirical model’s MAGT response is 0.6 ℃, double that of the transient model’s 0.3 ℃ response, which amplifies uncertainty in future projections. These findings highlight biases in widely used models for Tibetan Plateau permafrost, affecting tipping point projections and downstream ecohydrology, and call for the development of thermal models tailored to the region’s climate and hydrogeological conditions, supporting the global cryosphere modeling community.