ABSTRACT High‐resolution mapping of permafrost in ecologically and topographically complex landscapes remains a major challenge. Existing models of permafrost extent often rely on equilibrium assumptions, which can misrepresent conditions in regions where permafrost persists largely due to ecosystem structure. In such disequilibrium settings, climate‐driven approaches alone tend to underpredict permafrost distribution, underscoring the need for methods that integrate both ecological and climatic controls. This study presents a hybrid framework that combines a binary logistic regression model, calibrated with community‐based active layer and permafrost surveys, with the temperature at the top of permafrost (TTOP) model. By leveraging the strengths of each approach, the hybrid model improves prediction of permafrost probability relative to TTOP alone, with the clearest gains in burn scars and some mixed‐wood forest classes, while hydrologically complex wetlands remain areas of weaker performance, and it maintains the ability to be driven by climate scenarios. Overall classification accuracy of the hybrid model is comparable to that of the TTOP implementation and slightly lower than that of the existing logistic regression model, reflecting a trade‐off between maximal present‐day accuracy and enhanced capacity for spatial extrapolation and temporal projection. The model was trained and validated using 139 cryotic assessment sites (CAS), along with high‐resolution meteorological and ground temperature data. Results demonstrate that the hybrid approach better aligns climate‐driven predictions with observed ecosystem‐modified permafrost and provides equilibrium‐based scenarios under future climate forcings. This study highlights the potential of hybridized approaches for advancing permafrost mapping in heterogeneous boreal landscapes and offers a practical tool to support community planning and adaptation in northern regions facing rapid climate change.
Bonnaventure et al. (Fri,) studied this question.