Permafrost degradation in the Western Russian Arctic

Temperature monitoring is being performed at all sites (table 1). All but one borehole (R2 at site 3) are 10–12 m deep and were installed more than 35 years ago (1972–1983). The depth of these boreholes corresponds with the general depth of zero annual amplitude in this region upon instillation. Sites 1 through 5 represent undisturbed environmental conditions while sites 6 and 7 are in areas previously used for clear-cut logging that have since revegetated. Clear cutting occurred before these boreholes were installed at sites 6 and 7, and measurements here are illustrative of an initial pulse of permafrost degradation despite subsequent revegetation over the last 40 years. All sites provide valuable long-term thermal records through, for both undisturbed and disturbed landscapes.

Temperature measurements in these boreholes were originally conducted using mercury thermometers inserted in metal casings filled with brass. The accuracy of these early thermometers was ± 0.1 °C. Measurements were recorded manually at 0.5, 1.0 m, and subsequent 1 m intervals to the bottom of a particular borehole, every 10 days until 1990, after which they were taken manually at least annually at the end of the warm season. In 2003 all boreholes were equipped with 4-channel Hobo U12 loggers with ±0.1 °C measurement accuracy in accordance with monitoring guidelines laid out by GTN-P (Biskaborn et al 2015). Automated measurements using these Hobo loggers were then collected at 2.0, 3.0, 5.0, 10.0 and, if total borehole depth allowed, 12.0 m every 6 hours.

After 1995, gridded active layer measurements were conducted annually at all seven sites according to CALM protocols (Brown et al 2000). Either a 100 × 100 m (1 ha) or 1 km² grid with grid nodes every 10 or 100 m, respectively, was established to cover the variety of land covers characteristic of a given site. Thaw depth is measured by manually inserting a metal probe 2 m long until the point of resistance, interpreted as the permafrost table. Thaw depth was measured annually at the end of the thaw period which differs between sites. On the measurement date the average of at least three probe measurements is recorded for each grid node yielding 121 spatially distributed thaw depth values for a given landscape. In locations where thaw depth is deeper than 2 m (total length of a probe), every two or three years, a hand drill is used to determine the permafrost table depth. If the hand drill could not reach the permafrost table, Ground Penetrating Radar (GPR) was employed, though the accuracy is ±0.2 m in sand, a common substrate at the study areas (Melnikov et al 2010, Sadurtdinov et al 2018).

In 1997 CALM instillations at sites 1 through 5 were also equipped with 2-channel Hobo Pro, and later, Hobo Pro V2 temperature data loggers to monitor near-surface soil temperatures at depths less than 2 m. Similar to the data loggers installed in the nearby boreholes, these loggers were programmed to record temperatures every 6 hours and have a measurement accuracy of ±0.1 °C.

Site MAAT decreases toward the northeast of the study area from −3.9 at Cape Bolvansky to −7.6 °C at Marre-Sale. All six weather stations showed increasing MAAT trends over the 1970–2016 observation period. The highest rates of temperature warming (0.07 °C yr⁻¹) were reported at the northernmost Cape Bolvansky and Marre-Sale weather stations on sea coast. Inland weather stations also exhibited increasing trends in MAAT, but at significantly lower rates. For instance, the lowest rate documented was 0.05 °C yr⁻¹ in Nadym (figure 2(A)).

Zoom In Zoom Out Reset image size

Total annual precipitation reported at these five weather stations varies from 320 to 470 mm yr⁻¹. All stations displayed increasing precipitation trends from 1 to 3 mm yr⁻¹ since 1970. Inland sites, such as Nadym and Novy Urengoy, experienced slightly higher rates of precipitation increases compared to stations on the coast.

Snow depths exhibit deepening gradients north to south and from east to west. Snow depths are considerably deeper in mixed forest-tundra and taiga to the south compared to tundra landscapes to the north. For example, in the mixed forest near Novy Urengoy average annual snow depth was 114 cm as opposed to the tundra near Nadym where it was 85 cm. Also, sites in the east part of the study area, such as Vorkuta and Marre-Sale had shallower mean annual snow depths, 44 and 33 cm, respectively, while Cape Bolvansky to the west had 58 cm on average. Total annual snow accumulation is increasing at all sites too. Over the 1998–2018 period snow accumulation increased by 1.8 cm yr⁻¹ on average in tundra landscapes and 0.6 cm yr⁻¹ in northern taiga and mixed tundra-taiga (figure 2(B)).

Active layer soil temperatures from 1997 to 2018 show progressive warming at sites 1 through 5. Initially during the observation period, active layer temperatures from tundra landscapes were from –8.0 °C to –6.0°C, and from –3.8 °C to –1.9 °C at southern sites. By 2007 active layer temperatures increased to –4.8 °C in the typical tundra subzone, and are approaching 0 °C at southern sites (figure 3).

Zoom In Zoom Out Reset image size

The average rate of warming during the 1997–2007 period was 0.25 °C yr⁻¹. After passing the melting point of water, the rate of temperature change decreased to 0.06 °C yr⁻¹. All but site 4 experienced near-surface permafrost thaw, highlighting the fact that permafrost degradation may occur quickly even in locations with relatively cold permafrost.

Increasing thaw and therefore ALT also entails permafrost table lowering. Sites located in typical tundra (bioclimatic subzone D) on continuous permafrost show increasing ALT from 104 to 117 cm during the period from 1997 to 2018. Sites located in southern tundra (bioclimatic subzone E) also show evidence of permafrost table lowering (figure 4). At site 2, the permafrost table has lowered from 1.2 m below the ground surface in 2000 to 1.8 m in 2016, a depth below the average site ALT. At site 1 the permafrost table lowered 0.4 m in the year from 2015 to 2016. However, seemingly in contradiction, throughout the observation period the permafrost table does not appear to have lowered, but previous work by Kaverin et al (2017) suggests that this could be attributed to regional subsidence of the ground surface. Corrected for subsidence, the permafrost table at site 3 has lowered from 0.6 m below the ground surface in 1999 to 1.2 m in 2015.

Zoom In Zoom Out Reset image size

Despite similar climatic conditions at sites 5 and 7, the greatest changes in permafrost table position were in well-drained mixed forest-tundra at site 7. The near-surface peat at site 5 affords better thermoinsulation than the well-drained sands at site 7. Permafrost table lowering was noticeable at site 7 over the first 20 years of observation, but degradation accelerated markedly in the 2000s until it was 10 m below the ground surface by 2014. Similarly, the permafrost table at borehole 5–08 at the same site lowered 3 m in the 2000s until it was 7 m below the ground surface by 2010, after which monitoring at this borehole was discontinued. The permafrost table also lowered by 2 m in northern taiga, characterized by bogs and polygonal peatlands (site 5). Meanwhile, peatland (site 2) appeared relatively stable.

Figure 5 highlights sites in a latitudinal gradient (from 66° to 69° N) also representative of major bioclimatic subzones, all of which show permafrost warming since the beginning of observation (figure 5(A)). Since the 1970s, the highest rate of permafrost temperature increase in the study area and globally was at site 4 (Marre-Sale) with temperatures at the depth of zero annual amplitude increasing 0.06 °C year⁻¹ (this and subsequent rates reported are significant at p < 0.01). Although, permafrost temperatures here are relatively cold, ranging from −3.5 °C in a sand blowout (borehole No. 43a) and −5 °C in wet tundra (borehole No. 3). Mean annual permafrost temperature on the flood plain at site 4 is now −2 °C. In the 1970s site 4, representative of bioclimatic subzone D, or typical tundra, had five out of six boreholes with average annual temperatures from −7.4 °C (borehole 6 in well-drained tundra) to −5.5 °C (borehole 1 in moist tundra). The outlier at this site was borehole 36, located on the river flood plain with an average temperature of −4.5 °C. This spatial variability in permafrost temperature is attributed to differences in snow accumulation. The flood plain where borehole 36 is located accumulates twice the snow height relative to nearby zonal tundra landscapes. While spatial variability in warming are attributable to different soil and vegetation conditions at the site, various landscapes here had little influence on the response of permafrost temperature to warming climate as they all displayed similar warming rates (figure 5(B)).

Zoom In Zoom Out Reset image size

Permafrost temperatures in southern tundra, or bioclimatic subzone E, are represented by sites 1, 2 (figures 5(C)) and 6 (figure 5(D)). At site 2, permafrost temperatures in the 1980s range from −2.4 °C to −1.5 °C with the exception of the borehole in a polygonal peat bog with a temperature of −0.8 °C (borehole No. 56). The latest mean annual temperatures show that moist tundra (borehole 65) is now −0.7 °C and well-drained tundra (borehole 83) is −1.2 °C, while the polygonal peatland warmed only slightly to −0.6 °C. Permafrost temperatures at site 1 (only borehole 3) are greatly affected by snow accumulation. Tall shrubs accumulate snow redistributed by wind resulting in 1.5 times the snow accumulation at nearby shrub-free areas. The proximity of the Pechora River and the talik beneath also influence local thermal conditions. Considering borehole 3 has a short record (2016–2018), mean annual permafrost temperature has not changed significantly (−0.2 °C to −0.15 °C) (figure 5(C)), but the results of GPR profiling indicate the permafrost table is lowering (Sadurtdinov et al 2018).

Zoom In Zoom Out Reset image size

Boreholes at both sites 2 and 6 (figures 5(C) and (D)) have warmed at average rates of 0.04 °C and 0.05 °C yr⁻¹ respectively. Both sites are within bioclimatic subzone E in relatively cold permafrost (e.g. site 6 boreholes 15–03, 15–08, and 15–20) where mean annual permafrost temperatures have warmed from cooler than −5 °C in 1975 to −3.5 °C in 2018. Meanwhile, in well-drained shrub landscapes (borehole 15–06) permafrost temperature was higher and experienced less pronounced warming (from −1.8 °C in 1978 to −0.8 °C in 2018) (figure 5(D)).

Mean annual permafrost temperatures in mixed forest-tundra at site 7 ranged from −3.7 °C to −2.6 °C in the 1970s (boreholes 5–03, 5–07, 5–09, and 5–10) and currently range from −1.0 °C to −0.2 °C, increasing at an average rate of 0.045 °C year⁻¹. Boreholes 5–01 and 5–08 were installed in disturbed landscapes previously logged and burned respectively. Both borehole temperature records warm 2 °C–0.6 °C over the same period. This warming brought borehole 5–01 to the 0 °C thawing threshold and both disturbed landscapes have experienced substantial permafrost table lowering (figure 5(E)).

In northern taiga at site 5 mean annual permafrost temperatures are all warming, but following different trends depending on landscape. The three boreholes in bogs at site 5 are now above −0.2 °C, while the one borehole in polygonal peat remains at −0.3 °C (figure 5(F)). In the bog the permafrost table has lowered to 4 m below the ground surface and the average rate of permafrost warming in northern taiga was 0.035 °C yr⁻¹. This underscores the moderating role of peat under warming climatic conditions.

The results of long-term permafrost and active layer monitoring show three distinguishable stages of permafrost degradation along the latitudinal gradient of bioclimatic subzones present in the western Russian Arctic. This classification of degradation stages serves as an application and elaboration of the 'climate-ecosystem-permafrost' patterns of formation and degradation described by Shur and Jorgenson (2007) based on Alaskan observations (figure 6). Permafrost in the western Russian Arctic was formed through climate-driven processes, however the Pleistocene-Holocene transition and now rapid anthropogenic-driven warming are altering permafrost distribution here as the landscape returns to a state of equilibrium, where disturbance and recovery are balanced (Melnikov 1984).

The initial stage of degradation is a progressive increasing in ALT while permafrost temperatures remain relatively cold. Similarly, the transient layer, affected by climate over decades to centuries, can play a protective or buffering role for underlying permafrost, though extreme hot summers may partially thaw the transient layer (Shur 1988, Shur et al 2005). This initial stage is characteristic of observed trends from site 4 in bioclimatic subzone D Shur and Jorgenson (2007) refer to this as 'climate-driven, ecosystem-modified permafrost' that is thermally stable, but more susceptible to thaw.

The second stage, referred to as 'climate-driven, ecosystem-protected permafrost,' is where enhanced permafrost degradation is initiated after the complete thaw of ground ice in the transient layer. During this stage surface covers, and particularly peat, can moderate warming climatic signals. Accelerated thaw results in increased near-surface temperatures followed by the permafrost table lowering. Cold winters with little snow accumulation may provide conditions for short-term permafrost recovery, or aggregation, but the absence of developed segregation ice typically found in the transient layer may only temporarily slow rather than reverse long-term degradation. Sites 1, 2, 3 and 6 are examples of this metastable stage of permafrost degradation in tall shrub ecosystems of southern tundra, or bioclimatic subzone E.

The third and final stage is when progressive lowering of the permafrost table reduces the active layer to a seasonally frozen layer followed by temperatures at the depth of zero annual amplitude warming and exceeding 0 °C. This third stage occurs with progressive warming, like the situation in the forest-tundra at site 5, or disturbance, like the logged and burned taiga at site 7, of 'ecosystem-protected permafrost' (Shur and Jorgenson 2007).

While regional climate warming is the major driver of changes observed causing vertical permafrost degradation, ecosystem variability, particularly related to vegetation and soil properties, determine where permafrost is more resilient or susceptible to warming climate signals. For instance sites in degradation stages 1 and 2 may recover as ecosystem-driven permafrost dependent on biophysical (e.g. vegetation and organic matter accumulation) and micro-climatologic factors. Progressive climate warming however, means degrading permafrost is unlikely to re-aggrade.