Researchers reworked the snow physics in CLASSIC, the Canadian land surface model used in climate assessments, to better represent Arctic snowpacks. The updated single-layer scheme uses wind-dependent compaction, a newer thermal conductivity relationship, and a baseline turbulent exchange to fix large biases. At Arctic test sites, average snow depth error dropped from 12 centimeters to 0.1 centimeters, and winter soil temperature errors fell from 5.3 to 3.0 degrees Celsius on average. Since Arctic soils hold roughly twice as much carbon as the atmosphere, those corrections change simulated permafrost thaw and winter carbon respiration. Because the model is part of Canada's earth system model and feeds IPCC-style assessments, the fix improves confidence in circumpolar climate projections. The study is published in Geoscientific Model Development and validates the changes against independent sites, including Trail Valley Creek. This matters for anyone tracking permafrost carbon feedbacks, carbon cycle science, and climate policy built on land surface model outputs.
