Permafrost thicknesses range from very thin layers only a few centimeters thick to about 1500 m in unglaciated areas of Siberia (11).

Because permafrost is highly susceptible to long-term warming, it has been designated a “geoindicator,” to be used as a primary tool for monitoring and assessing environmental change (12). Empirical evidence strongly indicates that impacts related to climate warming are well underway in the polar regions (13). Permafrost at many Arctic locations has experienced temperature increases in recent decades (10).

A significant reduction in the area of near-surface permafrost could occur during the next century. The hypothesis and existing evidence that warming will increase the thickness of the active layer, resulting in thawing of ice-rich permafrost, ground instability, and surface subsidence, require further investigation under a variety of contemporary environmental settings (10).

Permafrost is defined as the ground where soil temperature remains at or below 0°C for at least two consecutive years. Approximately 1/4 of the Northern Hemisphere land area is permafrost. Melting of permafrost under global warming will affect hydrology and water resources, because of the water that flows out of the melting soil (27). It will affect ecosystems, because the heating of the soil and its changing hydrology changes the biogeochemical cycles in the soil (28). It will affect human infrastructures, because the soil gets less stable and buildings, roads, oil and gas pipelines, etc., settle differently from one point to another (29). In fact, melting permafrost affects climate change itself, because of the release of carbon from the degrading soil (30).

According to these climate models, global warming of 2°C above preindustrial levels will be reached in the first half of the 22nd century (2037  / 2045 under the RCP4.5 / RCP8.5 scenario). Global warming at northern latitudes will exceed the global mean (31). When the global mean temperature rise reaches 2°C, air temperature in the permafrost region increases by at least 2.9-4.4°C and 3.0-4.1°C under the RCP4.5 and RCP8.5 scenario, respectively.  As a result, the Northern Hemisphere’s permafrost soil temperature will increase by 2.34-2.67°C at 6 m depth relative to the period 1990-2000 (26).

Under 2°C global warming the permafrost extent will obviously retreat north and decrease by about 25%. The thickness of the so-called active soil layer, the layer that thaws and freezes in turn, will increase by 0.42-0.45 m on average. Locally the increase may be much higher though, up to 5 metres. Ground settlement owing to permafrost thaw is estimated at 3.8-15 cm on average for the Northern Hemisphere permafrost land area, but may reach several metres locally (26). 

In high-latitude regions of the Earth, temperatures have risen 0.6 °C per decade, twice as fast as the global average (24). The resulting thaw of frozen ground exposes substantial quantities of organic carbon to decomposition by soil microbes (23). The permafrost region contains twice as much carbon as there is currently in the atmosphere (28). A substantial fraction of this material can be mineralized by microbes and converted to CO2 and CH4 on timescales of years to decades. At the proposed rates, the observed and projected emissions of CH4 and CO2 from thawing permafrost are unlikely to cause abrupt climate change over a period of a few years to a decade. Instead, permafrost carbon emissions are likely to be felt over decades to centuries as northern regions warm, making climate change happen faster than we would expect on the basis of projected emissions from human activities alone (25).

Abrupt permafrost thaw occurs when warming melts ground ice, causing the land surface to collapse into the volume previously occupied by ice. This process, called thermokarst, alters surface hydrology. Water is attracted towards collapse areas, and pooling or flowing water in turn causes more localized thawing and even mass erosion. Owing to these localized feedbacks that can thaw through tens of metres of permafrost across a hillslope within only a few years, permafrost thaw occurs much more rapidly than would be predicted from changes in air temperature alone. Abrupt thaw is an important mechanism of rapid permafrost degradation, yet abrupt thaw is not included in large-scale models, suggesting that important landscape transformations are not currently being considered in forecasts of permafrost carbon–climate feedbacks. This is in part due to the fact that we do not know at this stage what the relative importance of abrupt to gradual thaw across the landscape is likely to be (25). 

In these model projections mid-century is defined as the period 2055–2064, and the damage is compared to the reference period 2015–2024. The results show that under the moderate scenario of climate change 29% of roads, 23% of railroads, and 11% of buildings will be affected by permafrost degradation. Under the high-end scenario, these numbers are 44% of roads, 34% of railroads, and 17% of buildings.

In Scandinavia, the current risk of slumps due to the thawing of permafrost is generally low (35).

Thawing of ground permafrost will disrupt access through shorter ice road seasons and cause damage to existing infrastructure (4). On the other hand, reduced sea ice and thawing ground in the Arctic will increase marine access and navigable periods for the Northern Sea Route (5).

The irregular surface created by thawing of ice-rich permafrost is known as thermokarst terrain. Global warming is likely to trigger a new episode of widespread thermokarst development, with serious consequences for a large proportion of the engineered works constructed in the permafrost regions during the twentieth century. Thawing of ice-rich permafrost is presently creating thermokarst terrain in the Alaskan interior and is having significant effects on subarctic ecosystems and infrastructure (15). Recent geographic overviews indicate that the hazard potential associated with ice-rich permafrost is high in many parts of the Arctic (16).

The accelerated thawing of permafrost not only disrupts vegetation and essential infrastructure, such as roads and pipelines, but could eventually release into the atmosphere vast amounts of methane. The maximum permafrost area has already shrunk by 7% since 1900 (6).

Coastal erosion

Warming will also accelerate the erosion of shorelines and riverbanks, threatening the infrastructure located on eroding shorelines. Increased storminess and higher waves are eroding arctic coasts at greater rates than in the past (18). The combination of increased wave action and warming permafrost especially threatens lowlying coastal villages (19). Several villages in Alaska have lost buildings to the sea (20).

Systematic measurements have been carried out of European mountain permafrost temperatures from a latitudinal transect of six boreholes extending from the Alps, through Scandinavia to Svalbard. Boreholes were drilled in bedrock to depths of at least 100 m between May 1998 and September 2000. Geothermal profiles provide evidence for regional-scale secular warming, since all are nonlinear, with near-surface warm-side temperature deviations from the deeper thermal gradient (7).

Topographic effects lead to variability between Alpine sites. First approximation estimates, based on curvature within the borehole thermal profiles, indicate a maximum ground surface warming of +1°C in Svalbard, considered to relate to thermal changes in the last 100 years. In addition, a 15-year time series of thermal data from the 58 meter deep Murtèl–Corvatsch permafrost borehole in Switzerland, drilled in creeping frozen ice-rich rock debris, shows an overall warming trend, but with high-amplitude interannual fluctuations that reflect early winter snow cover more strongly than air temperatures (7).

For the spatial distribution of palsa mires in Fennoscandia it was estimated as very likely (>90% probability) that a loss of area suitable for palsa mires to less than half of the baseline distribution will occur by the 2030s and very likely or likely (>66%) that all suitable areas will disappear by the end of the 21st century under the A1B and A2 emissions scenarios. For the B1 scenario, it was more likely than not (>50%) that a small proportion of the current palsa mire distribution would remain until the end of the 21st century (8).