Permafrost regions store an estimated half of the global belowground organic carbon pool and twice the global atmospheric carbon level. A warming climate results in increased carbon gas emission, therefore knowing more about the amount and composition of organic carbon stored in permafrost regions is crucial for understanding feedbacks on global climate change. Using the Yukon-Kukskowim (YK) Delta, Alaska as a study site, we quantified belowground carbon pools and their potential vulnerability to release into the atmosphere as greenhouse gasses. We identified relevant landcover classes (burned and unburned upland peat plateaus, wetlands, ponds/lakes) in the YK Delta, from which we quantified total belowground carbon pools (30cm) and assessed the composition of the organic matter using Fourier-transform infrared spectroscopy. To characterize the size and distribution of soil carbon pools in the YK Delta, we built a Random Forest Machine Learning model that mapped the spatial distribution of soil carbon to a depth of 30 cm over a 1910 km2 watershed. The map product was produced in Google Earth Engine and used covariates that include, but are not limited to, Worldview2 high-resolution optical imagery (2m), ArcticDEM (5m), and Sentinel-2 level 1C multispectral imagery (10 m), including NDVI. We found substantial variation across landcover classes in soil characteristics that affect organic matter vulnerability, including gravimetric water content, thaw depth, bulk density, and percent carbon. Compared to upland areas, thaw depths were significantly deeper in wetlands and lakes, where we detected no surface permafrost (to 1m). Soil carbon content (%) was greatest in moss-dominated wetlands; however, these areas also had the lowest bulk density. Carbon pools and organic matter characteristics also varied between burned and unburned areas. Therefore, we expect that carbon vulnerability varies by landcover class and that future carbon emissions are driven by total carbon pools, thaw depths, and composition of the carbon stored in organic matter pools. These carbon pool and vulnerability maps will contribute to better understanding the impacts of subarctic warming and are critical for developing a more accurate assessment of carbon cycling feedbacks from permafrost regions on global climate change.

Retrogressive Thaw Slumps (RTSs), a highly dynamic form of mass wasting, are accelerating geomorphic change across ice-cored permafrost terrain, yet the main controls on their activity are poorly constrained. Questions over the spatial variability of environmentally sensitive buried massive ice (MI) bodies and a paucity of high-spatial and temporal resolution topographic data have limited our ability to project their development and wider impacts. This research addresses these key problems by investigating RTS processes on Peninsula Point — the type site for intra-sedimental MI in the Western Canadian Arctic. Utilizing high-resolution topographic data from drone surveys in 2016, 2017 and 2018 we (1) measure the temporal and spatial variations in headwall properties and retreat rates, (2) determine the spatial pattern of subsurface layering using passive seismic monitoring and (3) combine these to analyse and contextualise the factors controlling headwall retreat rates. We find that headwall properties, namely MI thickness and overburden thickness, are significant controls over rates of headwall retreat. Where persistent ice exposures are present and overburden thickness remains < 4 m, headwall retreat is typically more than double that of other headwalls. Furthermore, a 3D site model was created by combining photogrammetric and passive seismic data, highlighting the variability in internal layering, demonstrating the limitations of extrapolations based on headwall exposures, and improving predictions of headwall retreat rates compared to long term averages and extrapolations from the previous year. These results provide fresh insights into the controls on headwall retreat rates and new approaches to improve their predictability.