Permafrost is a ubiquitous phenomenon in the Arctic. Its future evolution is likely to control changes in northern high-latitude hydrology and biogeochemistry. Here we evaluate the permafrost dynamics in the global models participating in the Coupled Model Intercomparison Project (present generation – CMIP6; previous generation – CMIP5) along with the sensitivity of permafrost to climate change. Whilst the northern high-latitude air temperatures are relatively well simulated by the climate models, they do introduce a bias into any subsequent model estimate of permafrost. Therefore evaluation metrics are defined in relation to the air temperature. This paper shows that the climate, snow and permafrost physics of the CMIP6 multi-model ensemble is very similar to that of the CMIP5 multi-model ensemble. The main differences are that a small number of models have demonstrably better snow insulation in CMIP6 than in CMIP5 and a small number have a deeper soil profile. These changes lead to a small overall improvement in the representation of the permafrost extent. There is little improvement in the simulation of maximum summer thaw depth between CMIP5 and CMIP6. We suggest that more models should include a better-resolved and deeper soil profile as a first step towards addressing this. We use the annual mean thawed volume of the top 2 m of the soil defined from the model soil profiles for the permafrost region to quantify changes in permafrost dynamics. The CMIP6 models project that the annual mean frozen volume in the top 2 m of the soil could decrease by 10 %–40 of global mean surface air temperature increase.
29 July 2023 in Journal of Geophysical Research: Earth Surface
Long‐term (1991–2018) thaw tube measurements highlight widespread permafrost thaw and ground surface (GS) subsidence over a large portion of northwest Canada. Statistically significant positive trends in thaw penetration (TP), measured with respect to a fixed datum, were observed at 18 of 28 sites with data that span three decades at a median rate of 0.8 cm a−1. This rate implies thawing of about 22 cm of permafrost over the study period. Similarly significant trends in GS subsidence occurred at 21 of 28 sites, at a median rate of 0.4 cm a−1. In contrast with TP and GS elevation, long‐term trends in active layer thickness (ALT) were less consistent, with 10 of 28 sites having statistically significant trends indicating increasing ALT (median rate: 0.6 cm a−1), and 7 sites having trends indicating decrease in ALT (median: 0.3 cm a−1). The ALT measurements underestimated the thawing of ice‐rich permafrost due to GS subsidence. Sites with high rates of thaw had relatively warm permafrost, deep snow cover, and poor drainage. Masking of ice‐rich permafrost degradation by ALT measurements has important implications for modeling permafrost thaw and climate change reporting. Accurate simulation of ice‐rich permafrost thaw is conceivable for sites where conditions are well‐defined. However, the prediction of subsidence and TP at regional or broader scales is only possible in general terms due to variation in surface conditions and ground ice content. Trends in TP and GS subsidence more accurately characterize ice‐rich permafrost degradation than trends in ALT.
Field observations show significant impacts of wildfires on active layer thickness and ground temperatures. However, the importance of fires to permafrost conditions at regional scales remains unclear, especially with climate warming. This study evaluated the regional impacts of fire on permafrost with climate change from 1942 to 2100 using a process‐based model in a large subarctic region in the Northwest Territories, Canada. Climate warming is shown to be the dominant factor for permafrost reduction. The warming trend of climate reduces permafrost extent in this region from 67% at present to 2% by 2100. For burned areas, fire increases the reduction of permafrost extent by up to 9% on average, with up to 16% for forest, 10% for tundra and bogs, and 4% for fens. Fire accelerates permafrost disappearance by 5 years on average. The effects of fire on active layer thickness and permafrost extent are much larger in forest areas than in tundra, bogs, and fens. Since active layer is thicker after a fire and cannot recover in most of the areas, the fire effects on active layer are widespread. On average, fires thickens active layer by about 0.5 m. The fire effects on active layer increased significantly after 1990 due to climate warming.
Most spatial modelling of climate change impacts on permafrost has been conducted at half-degree latitude/longitude or coarser spatial resolution. At such coarse resolution, topographic effects on insolation cannot be considered accurately and the results are not suitable for land-use planning and ecological assessment. Here we mapped climate change impacts on permafrost from 1968 to 2100 at 10 m resolution using a process-based model for Ivvavik National Park, an Arctic region with complex terrain in northern Yukon, Canada. Soil and drainage conditions were defined based on ecosystem types, which were mapped using SPOT imagery. Leaf area indices were mapped using Landsat imagery and the ecosystem map. Climate distribution was estimated based on elevation and station observations, and the effects of topography on insolation were calculated based on slope, aspect and viewshed. To reduce computation time, we clustered climate distribution and topographic effects on insolation into discrete types. The modelled active-layer thickness and permafrost distribution were comparable with field observations and other studies. The map portrayed large variations in active-layer thickness, with ecosystem types being the most important controlling variable, followed by climate, including topographic effects on insolation. The results show deepening in active-layer thickness and progressive degradation of permafrost, although permafrost will persist in most of the park during the 21st century. This study also shows that ground conditions and climate scenarios are the major sources of uncertainty for high-resolution permafrost mapping.
Bodies of peatland permafrost were examined at five sites along a 300 km transect spanning the isolated patches permafrost zone in the coastal barrens of southeastern Labrador. Mean annual air temperatures ranged from +1 ∘ C in the south (latitude 51.4 ∘ N) to −1.1 ∘ C in the north (53.7 ∘ N) while mean ground temperatures at the top of the permafrost varied respectively from −0.7 to −2.3 ∘ C with shallow active layers (40–60 cm) throughout. Small surface offsets due to wind scouring of snow from the crests of palsas and peat plateaux, and large thermal offsets due to thick peat are critical to permafrost, which is absent in wetland and forested and forest–tundra areas inland, notwithstanding average air temperatures much lower than near the coast. Most permafrost peatland bodies are less than 5 m thick, with a maximum of 10 m, with steep geothermal gradients. One-dimensional thermal modelling for two sites showed that they are in equilibrium with the current climate, but the permafrost mounds are generally relict and could not form today without the low snow depths that result from a heaved peat surface. Despite the warm permafrost, model predictions using downscaled global warming scenarios (RCP2.6, RCP4.5, and RCP8.5) indicate that perennially frozen ground will thaw from the base up and may persist at the southern site until the middle of the 21st century. At the northern site, permafrost is more resilient, persisting to the 2060s under RCP8.5, the 2090s under RCP4.5, or beyond the 21st century under RCP2.6. Despite evidence of peatland permafrost degradation in the study region, the local-scale modelling suggests that the southern boundary of permafrost may not move north as quickly as previously hypothesized.