Across high-latitude and high-elevation landscapes, permafrost has been warming and thawing in response to rising temperatures and increasing disturbance such as wildfire. Where the permafrost is rich in ice, permafrost thaw induces instability through subsidence and slope failures. These geohazards impact infrastructure, habitat, and the carbon cycle. There is a need to monitor terrain stability and map permafrost ground ice, but high-resolution data products are lacking across much of the permafrost domain. Although remote sensing instruments do not directly observe permafrost, there has been rapid progress in inferring permafrost conditions through indirect remote sensing observations such as surface deformation.

This talk will briefly review permafrost and remote sensing of permafrost landscapes, then focus on remote sensing of ground ice. Late-season subsidence from satellite radar can identify ice-rich permafrost, enabling automated mapping of top-of-permafrost excess ice independent of surface features. A Bayesian inversion can quantify near-surface ice concentrations by matching predicted subsidence to observations. Not only can these remote sensing products inform planning, but they also provide insight into processes, such as the response of permafrost to climate extremes. Multi-satellite analyses over Northwestern Alaska show widespread permafrost degradation during an extreme year, followed by rapid recovery on subdecadal timescales. Integrating subsidence observations with surface features through machine learning can help anticipate terrain instability and sustainably steward permafrost ecosystems.