Search

Climate change is causing rapid changes of Arctic ecosystems. Yet, data needed to unravel complex subsurface processes are very rare. Using geophysical and in situ sensing, this study closes an observational gap associated with thermohydrological dynamics in discontinuous permafrost systems. It highlights the impact of vegetation and snow thickness distribution on subsurface thermohydrological properties and processes. Large snow accumulation near tall shrubs insulates the ground and allows for rapid and downward heat flow. Thinner snow pack above graminoid results in surficial freezing and prevents water from infiltrating into the subsurface. Analyzing short-term disturbances, we found that lateral flow could be a driving factor in talik formation. Interannual measurements show that deep permafrost temperatures increased by about 0.2°C over 2 years. The results, which suggest that snow-vegetation-subsurface processes are tightly coupled, will be useful for improving predictions of Arctic feedback to climate change, including how subsurface thermohydrology influences CO2 and CH4 fluxes.

Quantitative understanding of controls on thaw layer thickness (TLT) dynamics in the Arctic peninsula is essential for predictive understanding of permafrost degradation feedbacks to global warming and hydrobiochemical processes. This study jointly interprets electrical resistivity tomography (ERT) measurements and hydro-thermal numerical simulation results to assess spatiotemporal variations of TLT and to determine its controlling factors in Barrow, Alaska. Time-lapse ERT measurements along a 35-m transect were autonomously collected from 2013 to 2015 and inverted to obtain soil electrical resistivity. Based on several probe-based TLT measurements and co-located soil electrical resistivity, we estimated the electrical resistivity thresholds associated with the boundary between the thaw layer and permafrost using a grid search optimization algorithm. Then, we used the obtained thresholds to derive the TLT from all soil electrical resistivity images. The spatiotemporal analysis of the ERT-derived TLT shows that the TLT at high-centered polygons (HCPs) is smaller than that at low-centered polygons (LCPs), and that both thawing and freezing occur earlier at the HCPs compared to the LCPs. In order to provide a physical explanation for dynamics in the thaw layer, we performed 1-D hydro-thermal simulations using the community land model (CLM). Simulation results showed that air temperature and precipitation jointly govern the temporal variations of TLT, while the topsoil organic content (SOC) and polygon morphology are responsible for its spatial variations. When the topsoil SOC and its thickness increase, TLT decreases. Meanwhile, at LCPs, a thicker snow layer and saturated soil contribute to a thicker TLT and extend the time needed for TLT to freeze and thaw. This research highlights the importance of combination of measurements and numerical modeling to improve our understanding spatiotemporal variations and key controls of TLT in cold regions.

Permafrost soils are a critical component of the global carbon cycle and are locally important because they regulate the hydrologic flux from uplands to rivers. Furthermore, degradation of permafrost soils causes land surface subsidence, damaging infrastructure that is crucial for local communities. Regional and hemispherical maps of permafrost are too coarse to resolve distributions at a scale relevant to assessments of infrastructure stability or to illuminate geomorphic impacts of permafrost thaw. Here we train machine learning models to generate meter‐scale maps of near‐surface permafrost for three watersheds in the discontinuous permafrost region. The models were trained using ground truth determinations of near‐surface permafrost presence from measurements of soil temperature and electrical resistivity. We trained three classifiers: extremely randomized trees (ERTr), support vector machines (SVM), and an artificial neural network (ANN). Model uncertainty was determined using k‐fold cross validation, and the modeled extents of near‐surface permafrost were compared to the observed extents at each site. At‐a‐site near‐surface permafrost distributions predicted by the ERTr produced the highest accuracy (70%–90%). However, the transferability of the ERTr to the sites outside of the training data set was poor, with accuracies ranging from 50% to 77%. The SVM and ANN models had lower accuracies for at‐a‐site prediction (70%–83%), yet they had greater accuracy when transferred to the non‐training site (62%–78%). These models demonstrate the potential for integrating high‐resolution spatial data and machine learning models to develop maps of near‐surface permafrost extent at resolutions fine enough to assess infrastructure vulnerability and landscape morphology influenced by permafrost thaw. Plain Language Summary: Accurate spatial assessments of the extent of near‐surface permafrost in regions of discontinuous permafrost are essential for evaluating the potential impacts of permafrost thaw. Most current estimates of permafrost extent are provided at the scale of 10s of meters to kilometers, which is often too coarse to understand the impacts thaw might have on infrastructure or geomorphic processes. We trained machine learning models to generate maps of near‐surface permafrost at 3 m spatial resolution in three sites in western Alaska. These maps provide an advancement in evaluating the extent of permafrost in the region as well as highlighting the landscape and ecological factors most important in driving predictions. Key Points: High‐resolution maps of near‐surface permafrost on hillslopes were generated from machine learning modelsThe normalized difference vegetation index was the most important feature in prediction of near‐surface permafrostThe support vector machines generally generated the highest balanced accuracy for predictions at sites where the model was not trained [ABSTRACT FROM AUTHOR]

Assessing the lateral and vertical extent of permafrost is critical to understanding the fate of Arctic ecosystems under climate change. Yet, direct measurements of permafrost distribution and temperature are often limited to a small number of borehole locations. Here, we assess the use of co‐located shallow temperature and electrical resistivity tomography (ERT) measurements to estimate at high‐resolution the distribution of permafrost in three watersheds underlain by discontinuous permafrost. Synthetic modeling shows that co‐located temperature and ERT measurements allow for supervised classification schemes that provide 60% higher accuracy compared to unsupervised methods. Linking resistivity and size of the identified permafrost bodies to surface observations, we show that tall vegetation (>0.5 m) and gentle slopes (<15°) are related to warmer and smaller permafrost bodies, and a more frequent occurrence of taliks. Plain Language Summary: To better understand how the Arctic may be changing due to climate warming, we need to understand the distribution of permafrost in the subsurface. Although the temperatures of the ground can be measured in boreholes, only a small number of boreholes exist to do these measurements. Here, we use machine learning to link measurements of temperature and of the electrical resistivity of the ground to obtain detailed distributions of permafrost in the subsurface. Connecting the properties of permafrost with observations above ground, we show that south facing slopes, tall vegetation, and gentle slopes relate to warmer and smaller permafrost bodies. Key Points: High‐resolution estimation of permafrost distribution and depth over large spatial extent is obtained using geophysical and temperature dataMachine learning using co‐located temperature and resistivity measurements improve estimates of permafrost thicknessPermafrost depth and temperature varies with surface cover and slope angle [ABSTRACT FROM AUTHOR]