The Peel Plateau, NT, Canada, is an area underlain by warm continuous permafrost where changes in soil moisture, snow conditions, and shrub density have increased ground temperatures next to the Dempster Highway. In this study, ground temperatures, snow, and thaw depth were monitored before and after tall shrub removal (2014). A snow survey after tall shrub removal indicated that snow depth decreased by a third and lowered winter ground temperatures when compared with control tall shrub sites. The response of ground temperatures to shrub removal depended on soil type. The site with organic soils had cooler winter temperatures and no apparent change in summer temperatures following shrub removal. At sites with mineral soil, moderate winter ground cooling insufficiently counteracted increases in summer ground heat flux caused by canopy removal. Given the predominance of mineral soil along the Dempster, these observations suggest tall shrub removal is not a viable short-term permafrost management strategy. Additionally, the perpendicular orientation of the Highway to prevailing winter winds stimulates snow drift formation and predisposes the site to warmer permafrost temperatures, altered hydrology, and tall shrub proliferation. Subsequent research should explore the effectiveness of tall shrub removal at sites with colder winter conditions or different snow accumulation patterns.

Seasonal snow cover has an important impact on the difference between soil- and air temperature because of the insulation effect, and is therefore a key parameter in ecosystem models. However, it is still uncertain how specific variations in soil moisture, vegetation composition, and surface air warming, combined with snow dynamics such as compaction affect the difference between soil- and air temperature. Here, we present an analysis of 8 years (2012-2020) of snow dynamics in an Arctic ecosystem manipulation experiment (using snow fences) on Disko Island, West Greenland. We explore the snow insulation effect under different treatments (mesic tundra heath as a dry site and fen area as a wet site, snow addition from snow fences, warming using open top chambers, and shrub removal) on a plot-level scale. The snow fences significantly changed the inter-annual variation in snow depths and -phenology. The maximum annual mean snow depths were 90 cm on the control side and 122 cm on the snow addition side during all study years. Annual mean snow cover duration across 8 years was 234 days on the control side and 239 days on the snow addition side. The difference between soil- and air temperature was significantly higher on the snow addition side than on the control side of the snow fences. Based on a linear mixed-effects model, we conclude that the snow depth was the decisive factor affecting the difference between soil- and air temperature in the snow cover season (p < 0.0001). The change rate of the difference between soil- and air temperature, as a function of snow depth, was slower during the period before maximum snow depth than during the period between the day with maximum snow depth until snow ending day. During the snow-free season, the effects of the open top chambers were stronger than the effects of the shrub removal, and the combination of both contributed to the highest soil temperature in the dry site, but the warming effect of open top chambers was limited and shrub removal warmed soil temperature in the wet site. The warming effects of open top chambers and shrub removal were weakened on the snow addition side, which indicates a lagged effect of snow on soil temperature. This study quantifies important dynamics in soil-air temperature offsets linked to both snow and ecosystem changes mimicking climate change and provides a reference for future surface process simulations.