Modelling air, ground surface, and permafrost temperature variability across four dissimilar valleys, Yukon, Canada

Spatial maps of the air and ground thermal regime were generated for four Yukon valleys. The aim was to model air, ground surface, and ground temperature (at fine spatial resolution) using locally measured inverted surface lapse rates (SLR) to better predict temperature along an elevation gradient. These local models were then compared to a regional permafrost probability model, which utilized differing inversion assumptions, as well as circumpolar and national models generated without considering inversions. Overall, permafrost probability in the regional model matched well with the local models where assumptions of treeline and inverted SLRs held true. When normal SLRs were assumed, permafrost presence was overestimated in each valley. This discrepancy was greatest at high elevations where permafrost was predicted to be the coldest and most widespread. However, the difference between valleys was dependent on surface and subsurface characteristics such as higher snow cover, mature forest, or thick organic layers which show a greater disassociation from the air temperature overall. Appropriate characterization of the SLR is essential for accurate predictions of the ground thermal regime’s spatial distribution and permafrost presence. These models also provide a starting point for better predictions of warming in these valleys and other areas subject to inversions of similar magnitudes.