The role of snow cover affecting boreal-arctic soil freeze–thaw and carbon dynamics

Abstract

Abstract. Northern Hemisphere permafrost affected land areas contain about twice as much carbon as the global atmosphere. This vast carbon pool is vulnerable to accelerated losses through mobilization and decomposition under projected global warming. Satellite data records spanning the past 3 decades indicate widespread reductions (~ 0.8–1.3 days decade−1) in the mean annual snow cover extent and frozen-season duration across the pan-Arctic domain, coincident with regional climate warming trends. How the soil carbon pool responds to these changes will have a large impact on regional and global climate. Here, we developed a coupled terrestrial carbon and hydrology model framework with a detailed 1-D soil heat transfer representation to investigate the sensitivity of soil organic carbon stocks and soil decomposition to climate warming and changes in snow cover conditions in the pan-Arctic region over the past 3 decades (1982–2010). Our results indicate widespread soil active layer deepening across the pan-Arctic, with a mean decadal trend of 6.6 ± 12.0 (SD) cm, corresponding to widespread warming. Warming promotes vegetation growth and soil heterotrophic respiration particularly within surface soil layers (≤ 0.2 m). The model simulations also show that seasonal snow cover has a large impact on soil temperatures, whereby increases in snow cover promote deeper (≥ 0.5 m) soil layer warming and soil respiration, while inhibiting soil decomposition from surface (≤ 0.2 m) soil layers, especially in colder climate zones (mean annual T ≤ −10 °C). Our results demonstrate the important control of snow cover on northern soil freeze–thaw and soil carbon decomposition processes and the necessity of considering both warming and a change in precipitation and snow cover regimes in characterizing permafrost soil carbon dynamics.

Characterizing permafrost active layer dynamics and sensitivity to landscape spatial heterogeneity in Alaska

“…The model simulations were conducted using a detailed soil process model (Rawlins et al, 2013;Yi et al, 2015) primarily driven by global satellite observation records including land surface "skin" temperature (LST), SCE, and surface to root zone (≤ 1 m depth) soil moisture (SM). The soil process model defines up to 23 distinct soil layers down to 60 m below surface.…”

“…The model also accounts for the impacts of SOC content on soil thermal properties. The model was successfully applied to the pan-Arctic region for mapping permafrost extent and active layer dynamics, but at a relatively coarse (∼ 25 km) spatial resolution (Yi et al, 2015). In the previous study, global coarse-resolution (∼ 0.5 • ) reanalysis data, including surface air temperature and precipitation, were used as primary model inputs; the model soil thermal properties were regulated by soil moisture content simulated by a water balance model coupled with the soil thermal model (Rawlins et al, 2013).…”

“…MERRA-2 daily snow depth and density data were used to account for the effects of seasonal snow cover evolution on the ground thermal regime, with changes in seasonal snow thermal properties derived from snow density (Yi et al, 2015). The soil thermal regime is particularly sensitive to changes in snow cover conditions during snow onset and offset periods, and large-scale reanalysis snow datasets generally have difficulty capturing snow cover spatial heterogeneity, especially during seasonal transition periods (Westermann et al, 2017).…”

“…2.3.1. The soil physical properties for each soil layer were assumed to be a weighted combination of values of mineral soils and pure organic soils based on the estimated soil carbon fraction following Yi et al (2015).…”

“…However, the accuracy of these methods is limited by the ability of sparse ground measurements representing landscape heterogeneity, and the resulting empirical models provide only limited insight and mechanistic understanding of underlying processes affecting active layer conditions. Detailed process models have been developed to address the above limitations, while the model accuracy is constrained by a lack of information for effective model parameterization, limited process understanding, and coarse spatial scales of regional drivers (Yi et al, 2015;Jafarov and Schaefer, 2016). Particularly, large uncertainties remain in characterizing regional variability of subsurface soil organic carbon (SOC) content due to limited ground observations of this parameter in the Arctic region (Ping et al, 2008;Burnham and Sletten, 2010) and its effect on ground temperature evolution.…”