Arctic landscapes are in a period of transition, where increasing temperatures and changing precipitation regimes are intensifying disturbances associated with permafrost degradation. A framework for determining the susceptibility of the landscape to thermokarst would allow us to understand the future of circumpolar landscapes better. One approach to examine thermokarst susceptibility is with geospatial analyses and statistical models. These models are developed to represent the susceptibility of an area given the current distribution of climate, topographic and material conditions, with an understanding that these conditions likely influenced past thermokarst. Through the model, spatial patterns of thermokarst can be investigated in relation to specific geophysical variables (i.e., glacial history, slope, permafrost characteristics, surficial geology, etc.) and, in return, relevant information on relationships between terrain variables and thermokarst can be explored. Modelling results are presented as probabilities, with areas on the landscape identified as being more or less susceptible to thermokarst. While many of these models are trained with point-based data, the ability to move from point-based measurements to spatially distributed assessments of factors contributing to thermokarst is essential for forming a generalizable understanding of both earth-system processes and the interaction between natural systems. This presentation will highlight: 1) thermokarst susceptibility modelling trained with thermokarst inventories at different scales (i.e. field-based mapping and inventories developed using fine and course resolution imagery); 2) explore the physical factors driving the alteration of thermokarst systems and how they are represented at different scales; 3) examine how results of these models can inform risk assessments and decision making to improve public safety and environmental management.