Wei-Heng Ni, Guoan Yin, Fu-Jun NIU, Zhan-Ju LIN, Jing Luo, Hong-Ye YAN, Ze-Yong GAO, Xin JU, Qian Liu
Permafrost degradation along the vital Qinghai‒Tibet Engineering Corridor (QTEC) poses a severe, escalating threat to critical infrastructure, yet high-resolution assessments of future risk and economic costs are urgently required to inform resilient infrastructure planning and cost-benefit optimization. Here, we integrate 90-m resolution multi-source data using machine learning and a multi-indicator framework to project the evolution of permafrost stability, infrastructure risk, and associated economic losses through the year 2090 under different climate scenarios. Our projections reveal a severe degradation trajectory. By 2090 under a high-emission scenario (SSP5-8.5), permafrost coverage will shrink by over 80%, with mean ground temperatures rising by 5.2 °C and active layer thickening by 1.34 m. Consequently, high-risk zones will expand to cover 84% of the corridor. Our risk zonation is robustly validated by InSAR observations, which show significantly ( p < 0.01) higher deformation rates in the predicted high-risk zones, and by thermokarst inventories, with over 70% of hazards occurring in these zones. Economically, this translates to additional infrastructure replacement costs reaching 2.55 billion USD by 2090 (SSP5-8.5), with road and rail systems accounting for 91.3% of costs. This study provides the first 90-m resolution quantification of coupled permafrost‒risk‒cost dynamics, offering a critical scientific foundation for climate-adaptive engineering and proactive infrastructure investment.