We acknowledge the support of the Natural Sciences and Engineering Research Council of Canada via NSERC PermafrostNet (NETGP 523228-18) and RGPNS-2020-04783. We thank Marten Geertsema for his support and help with this project.
1In steep mountains, topography controls local ground conditions such as snow and ice cover and the subsurface thermal regime. It determines how regional climate translates into microclimates, for example, affecting air temperature, incoming solar radiation, and the redistribution of sediment and snow (Gruber and Haeberli, 2009). In regions where permafrost occurs with dense research and monitoring activity (e.g., Gruber et al., 2004b; Hipp et al., 2014; Magnin et al., 2019), the resulting spatial patterns are well understood, whereas patterns need to be newly investigated (Bodin et al., 2010) or inferred (Gruber et al., 2017) in other regions.
2Better understanding the similarity of a region of interest that has little previous research with another region that has dense research can help transfer knowledge with more confidence or to justify the investment into new research. For example, while ice loss from glaciers and permafrost is visibly correlated with slope instability in high-mountain regions globally (Deline et al., 2021), detailed studies are often constrained to intensely studied areas in Europe (e.g., Deline et al., 2011; Ravanel and Deline, 2011; Ravanel et al., 2010).
3This study originates from the need to better quantify permafrost characteristics in the mountains of western Canada so that the relationship of permafrost change and landslides could be better understood. Deciding whether to newly conduct local research, or to extrapolate from past studies elsewhere, however, is a generic challenge. As an example of how such decisions can be contextualised, we quantify similarities of western Canada with two mountain areas with a high density of permafrost research: the European Alps and the Scandes. Only sparse permafrost research is available for western Canada (Gruber et al., 2015). Even though Hasler et al. (2015) revealed some relationships between permafrost, topography, continentality, and surface characteristics for parts of British Columbia (BC), our ability to understand the entire mountainous environment of western Canada from this research is limited. Findings from European mountains, however, are likely not directly applicable to western Canada, where there are important differences such as a large latitudinal range, greater continentality and temperature inversions. For this reason, we compare the climate of the European Alps and the Scandes to western Canadian mountain regions with permafrost. Our objective is to understand where permafrost conditions can be compared and where new research is needed.
4We start by reviewing the relationship of climate, topography, and ground temperature in mountain regions with permafrost. With this basis, we then compare regional climate on a coarse scale, with approximately 30 km x 30 km grid cells, where continental-scale climatic differences become apparent and the fine-scale effects of topography are minimal.
5Research evaluating the permafrost-environment relationship in mountain permafrost regions has often used the mean annual ground surface temperature (MAGST). Even though it is not a direct observation of permafrost, it is inexpensive to measure and it reflects the most important part of the spatial variability of ground temperatures at depth. This section reviews knowledge about environmental controls on MAGST as a basis for understanding how regional climate and topography together produce differing patterns of MAGST.
6South-facing slopes have higher MAGST than north-facing slopes in the northern hemisphere (Gubler et al., 2011, Hipp et al., 2014, Hasler et al., 2015, Magnin et al., 2019) and this difference is greater on steeper slopes. In BC, Hasler et al. (2015) found a north-south difference in MAGST of 4 ºC in vertical bedrock, 1.5–3 ºC on open gentle slopes, and no difference in dense coniferous forest. Similarly, as a result of changing snow and radiation conditions in the Alps, Gubler et al. (2011) found north-south differences of up to 5 ºC in 40º slopes compared to 2 ºC in 10º slopes.
7In steep bedrock, there is a relationship between latitude and the magnitude of this north-south difference (Gruber, 2012a). At higher latitudes, direct solar radiation decreases on south-facing slopes and increases on north facing slopes due to the longer summer days and longer winter nights, effectively reducing the north-south temperature difference (Magnin et al., 2019). For example, north-south differences in MAGST are 0.5–1.5 ºC on Ellesmere Island at 80º (Lewkowicz, 2001), and are 5–8 ºC in the Alps at 46º (Gruber et al., 2004b, Magnin et al., 2015).
8These differences between aspects also extend to active-layer thicknesses. In the steep rock walls of the aiguille du Midi of the French Alps, Magnin et al. (2015) found the shallowest active layer on north-facing slopes (2.2 m) and deepest on south-facing slopes (5.5 m).
9Usually, air and ground temperatures decrease with elevation. As a result, permafrost in mountainous areas is often described in terms of the elevation of its lower limit. For example, in southern Norway, the lower limit of permafrost is estimated between 1050 and 1750 m above sea level, whereas in northern Norway it is estimated between 150 and 1400 m above sea level (Gisnås et al., 2017).
10The dependence of air temperature on elevation is often quantified by lapse rates. In the Alps, lapse rates on the order of 5–6 ºC/km have been measured and used for permafrost modelling studies (e.g., Ravanel et al., 2017, Boeckli et al., 2012). In the Scandes, lapse rates used for modelling permafrost are 2.6–8.6 ºC/km (Magnin et al., 2019) with an average of 4.4 ºC/km (Steiger et al., 2016).
11In the Arctic and in cold mountain areas, lapse rates can be inverted for extended periods, with lower temperatures in the valley bottom than at higher elevation (Bradley et al., 1992). In southern Yukon, Lewkowicz and Bonnaventure (2011) found that more continental regions have smaller lapse rates or inversions below treeline. As a result, MAGST can be lower in valley bottoms than up on higher slopes.
12In southern Norway, Hipp et al. (2014) observed the highest summer rock temperatures at the highest measuring point. This is likely due to a relationship between elevation and short-wave incoming radiation (SWin), where a decrease in shading and atmospheric attenuation with elevation increases solar radiation (Gruber et al., 2004b).
13Continentality increases with distance from the ocean or large bodies of water. It is often quantified as the annual range of mean monthly air temperatures. Continentality correlates with SWin and the amount of snow fall, both important predictors of permafrost conditions in mountains (Gruber and Haeberli, 2009).
14Bonnaventure and Lewkowicz (2008) found that permafrost of cloudy maritime sites in BC near the border of Alaska had no relationship to potential SWin but that more continental sites did. In New Zealand, Sattler et al. (2016) attributed low-elevation permafrost in debris-covered slopes to summer cloudiness of the maritime climate.
15Gisnås and coworkers (2017) found that lower limits of permafrost in northern Norway decreased in elevation with distance from the ocean. In southern Norway, this was attributed to decreasing snow cover and changing surface materials (more block fields) and not to decreasing mean annual air temperature (MAAT) (Farbrot et al., 2011).
16Snow further complicates topographic effects on MAGST due to its combined effects of insulation, high albedo, and uptake of latent heat during snowmelt. Pogliotti (2011) found that these effects depend on the thickness of the snowpack, where thin snowpacks tend to promote cooling, and thicker snowpacks promote warming. In shaded areas, snow mainly acts as an insulator in the winter and prevents heat from escaping, thus warming the ground (Goodrich, 1982). However, for slopes exposed to strong radiation, the high albedo of snow can have a cooling effect on the order of 2–3 ºC on MAGST (Hasler et al., 2011). Finally, when snow is in proximity to steep slopes, it can have a warming effect on neighbouring slopes by reflecting radiation onto them (Allen et al., 2009, Lewkowicz, 2001).
17Slope angle is closely related to the effect of snow because steep slopes more frequently shed their snow cover via avalanching and sluffing. Correspondingly, homogeneous slopes with angles above 50º are often assumed to be snow-free (Winstral et al., 2002), even though many slopes in fact are heterogenous and subject to intermittent snow cover even when considerably steeper (Wirz et al., 2011). In snow-free slopes between 90° and 50°, lower slope angles increase MAGST as a result of increasing SWin (Gruber et al., 2004b). However, the preferential re-distribution of snow to gentler slopes can have a cooling effect as it protects from SWin when snow cover is thin; Gubler et al. (2011) found that a steeper slope (40º) can be warmer than a shallower (10º) south-facing slope.
18Global ruggedness and permafrost maps (Gruber, 2012b) are used for identifying the mountain permafrost regions to be compared at a resolution of 30 arc-seconds (<1 km). The ruggedness index measures how flat or mountainous a landscape is and is provided in five classes. The permafrost zonation index is derived from a global digital elevation model and global MAAT data for the period of 1961–1990. Each cell has a permafrost zonation index between 0 and 1. For this study, mountainous areas with permafrost are those that have the ruggedness classes “mountainous” or “rugged”, and a permafrost zonation index ≥ 0.1.
19Climate data for comparing mountain regions is derived from the atmospheric reanalysis ERA5, produced by the European Centre for Medium-Range Weather Forecasts. It provides hourly data for the period 1979 to present with a grid spacing of 0.25º (about 30 km) (Hersbach et al., 2019). Reanalysis data are generated from assimilating climate observations into an atmospheric model and include a variety of variables related to radiation, precipitation, and temperature.
20Differences between the European Alps, Scandinavian mountains (the Scandes), and western Canadian mountainous permafrost regions (Fig. 1) are explored in relation to MAAT, elevation, precipitation, solar radiation, continentality and latitude. The primary tools used are ArcGIS Pro and R.
Figure 1. The three regions of study: western Canada, the European Alp and the Scandes
21Monthly surface-level ERA5 data was downloaded and processed for the regions of interest for the period of 1986–2005, for which data has been easily available for processing. Given the wide range of spatial differences of climates, the results of the present study are deemed insensitive to the choice of multi-decade analysis interval.
22Data were downloaded for air temperature (“2m temperature”), precipitation (“Total precipitation”), elevation (“Geopotential”), and incoming solar radiation (SWin: “Surface solar radiation downward”, SWdir: “Total sky direct solar radiation at surface”). These are used to calculate MAAT, continentality (as difference between maximum and minimum monthly air temperature), total annual precipitation, mean annual incoming solar radiation (SWin), and mean annual incoming direct solar radiation (SWdir), all averaged over the 20-year period.
23For further analysis, only the ERA5 grid cells that contain the previously identified mountainous areas with permafrost are retained. These points have the attributes calculated from ERA5, along with latitude, longitude, and regions of interest Scandes, European Alps and western Canada. To facilitate analysis, the western Canadian region is further subdivided into nine sub-regions (Fig. 2). These are delineated manually based on mountain ranges and similar climates, focusing on variation in the climate variables used during analysis: air temperature, precipitation, solar radiation, and continentality.
24Because the permafrost zonation index is related to the probability of finding permafrost, but not an accurate map of the presence of permafrost, some marginal cells may not contain permafrost.
Figure 2. Sub-regions of western Canada used for comparison with the Alps and Scandes
25The mountain regions with permafrost of western Canada cover a much larger area than that of the Scandes and the European Alps and encompass larger ranges in temperature, continentality, latitude, elevation, and solar radiation.
26Western Canadian mountains with permafrost often have lower MAAT than the Scandes and European Alps (Fig. 3). The general decrease of MAAT with latitude is visible in Figure 3a, and the warming effect of the North Atlantic Current for the Scandes. All western Canadian mountains are at higher latitude than the European Alps.
27Continentality has a range of about 28 ºC (17ºC to 45 ºC) in western Canada, compared to the Scandes with a range of about 12 ºC and the Alps of about 3 ºC (Fig. 3b). Coastal sub-regions of western Canada have similar continentality to the Alps, however, MAAT there is lower than in the Alps. The areas of the St-Elias Mountains sub-region are unique in having low MAAT, low continentality and high elevations, some of the permafrost areas here are peaks standing above the accumulation area of large glaciers.
28Some of the warmer areas of the Canadian sub-regions south of 60º N have similar MAAT to the European Alps and the Scandes. At the same elevation, the Scandes have lower air temperatures than South Interior and South Coast Mountains and higher temperatures than most sub-regions north of 60º N, other than a couple of areas from South Yukon. In Figure 3c we can see the sub-region points resembling lines of temperature decreasing with elevation. The slopes of these lines represent the lapse rates, which become weaker in colder and dryer conditions.
Figure 3
Air temperature versus (a) latitude, (b) continentality, and (c) elevation for mountain permafrost areas of the nine sub-regions of western Canada (coloured), the Scandes and the European Alps (grey) at a resolution of about 30 km x 30 km.
29SWin includes both direct and diffuse radiation and, as such, the influence of atmospheric scattering and clouds. SWdir represents only the solar radiation that comes directly from the sun, after attenuation by the atmosphere. Derived at the coarse ERA5 grid, it gives an indication of the relative difference in SWin between north and south facing slopes at fine scale.
30Most mountain areas with permafrost in western Canada receive amounts of solar radiation between those of the Scandes and the Alps, with several Canadian areas having much lower MAAT (Fig. 4). Higher air temperatures are generally associated with more solar radiation, even though the Scandes and areas of the St-Elias Mountains do not follow this trend. In terms of solar radiation regimes, the northernmost sub-regions of western Canada are similar to the Scandes, and the southern areas and parts of the St-Elias Mountains are similar to the Alps.
31Areas of western Canada with similar amounts of SWin as the Scandes or the Alps often have higher SWdir. This reduces similarity with the southern Scandinavian mountains for the western Canadian sub-regions above 60 ºN, and increases similarity with the Alps for South Interior.
Figure 4
The relationships of SWin (a) and SWdir (b) with latitude and of SWin with MAAT (c) for mountain areas with permafrost for nine sub-regions of western Canada (coloured), the Scandes and the European Alps (grey) at a resolution of about 30 km x 30 km.
32Most mountain areas with permafrost in western Canada have less precipitation than the Scandes and Alps (Fig. 5). These are the cold sub-regions north of 60º N. The sub-regions with higher precipitation are those situated along the Pacific coast, such as the St-Elias Mountains, and the North and South Coast mountains.
33North and South Interior partially overlap in precipitation and MAAT with the lower precipitation areas of the Scandes, with South Interior showing some higher air temperatures, and North Interior showing some lower air temperatures. Higher precipitation areas of South Interior and the warmer areas of the Coast mountains overlap with the colder parts of the Alps. Parts of the North and South Coast Mountains overlap with the Scandes and Alps in MAAT and precipitation.
Figure 5
Air temperature versus precipitation for the mountain areas with permafrost of the nine sub-regions of western Canada (coloured), the Scandes and the European Alps (grey) at a resolution of about 30 km x 30 km. Note that the vertical and horizontal axes are inverted to resemble cooling air with increasing elevation and the patterns of precipitation decreasing eastward over continents and mountains.
34Ground temperatures and their spatial patterns in mountains are controlled by air temperature, precipitation (snow cover) and radiation. Our results illustrate regional differences with a series of two-variable scatter plots, drawn from six variables analysed (latitude, MAAT, continentality, total annual precipitation, mean annual incoming solar radiation, and mean annual incoming direct solar radiation). These are intuitive to interpret while being quantitative and richer than earlier conceptual models (Schumskii, 1964; Haeberli and Burn, 2002; Gruber and Haeberli, 2009; Gruber, 2012b).
35While the example shown here is specific to comparing permafrost and ground thermal regimes of mountains in western Canada with European mountains, the methods and workflows can equally be used to support the transfer of knowledge to other areas globally (e.g., Wani et al., 2020) and to case studies that have a stronger component related to glaciers and snow (e.g., Shugar et al., 2021).
36The large mountain areas with permafrost in western Canada often have lower MAAT than the Alps and Scandes and likely lower ground temperatures, overall. Areas of western Canada have lower ground temperatures than the Scandes, which are warmed by the North Atlantic Current, at similar latitudes and/or elevations. The warmer areas of southern BC and Alberta could have ground temperatures similar to the Alps, despite their difference in latitude.
37Lapse rates get smaller with lower MAAT and increasing continentality, and cold air drainage and pooling can amplify the effect of inversions in valley bottoms during the winter months. With the large range of continentality and low MAAT found in western Canada, a much larger range in lapse rates than in the Scandes and the European Alps is expected. This has implications for the effect of elevation on permafrost in mountains, with the more northern sub-regions of western Canada likely showing a weaker relationship between elevation and ground temperatures.
38Precipitation is an important variable to consider on slopes where snow accumulates. Where local conditions are comparable, the interior areas of BC and Alberta and the interior areas of the Scandes may have similar ground temperatures. While areas in southern BC and Alberta are similar to the warmest areas of northern BC, southern Yukon, and the Scandes in terms of MAAT, only the interior areas of BC have precipitation similar to the dryer areas of the Scandes, those furthest from the coast.
39Similarly, although there is overlap in MAAT between much of the Alps and the lower elevation areas of BC and Alberta, only the higher precipitation areas of the South Interior and the areas of the Coast Mountains with higher MAAT have also overlap with precipitation. This suggests that where local slope and snow conditions are comparable, these areas may have similar ground temperatures to those of the Alps.
40Greater precipitation mostly results in thicker snowpacks and higher ground temperatures. However, where solar radiation is strong and snowpacks thin, for example in steep bedrock slopes, the high albedo of snow can also have a net cooling effect (Pogliotti, 2011; Hasler, 2011).
41Ground temperatures of snow-free slopes in western Canada cannot be easily compared with insight from the Scandes and Alps as few areas overlap in both SWin and MAAT. Southern BC and Alberta, with similar MAAT to the Scandes receive more radiation, suggesting that they would have higher ground temperatures in snow-free slopes. The South Interior and South Coast Mountains are expected to have ground temperatures similar to the Alps in snow-free slopes.
42These general temperature regimes are modulated by the interaction of topography and SWdir. The most northern sub-regions (North Yukon, parts of Ogilvie Mountains) and coastal sub-regions (North Coast Mountains, part of St-Elias Mountains) of western Canada are expected to have north-south differences in ground temperature similar to the southern Scandinavian mountains, about 3.5 ºC in steep bedrock. The South Interior and parts of the South Coast Mountains sub-regions are expected to have similar north-south differences to the Alps, about 6–8ºC in steep bedrock. The remaining western Canadian sub-regions (parts of Mackenzie Mountains, most of South Yukon, North Interior, and parts of the South Coast Mountains) are expected to have north-south temperature differences greater than the Scandes but smaller than in the Alps.
43The effects of radiation on differentiating ground temperature will be most pronounced in steep bedrock and reduced by the influence of a thick snowpack and vegetation canopies.
44As an example for informing the application of knowledge from established research areas into new regions, we have investigated similarities and differences between areas with permafrost in the European Alps, the Scandes, and mountains of western Canada. For this we have compared consistent and quantitative regional climate data, focused on variables controlling the ground thermal regime.
45For our application example, we conclude that a direct transfer of knowledge about ground temperature regimes and spatial patterns from the Scandes and Alps to western Canada is inappropriate. This is because the known climatic drivers of ground temperatures and their spatial differentiation are quite different in these regions: (1) The areas in western Canada receive more radiation than those in the Scandes, and less than in the Alps. (2) The areas in western Canada are more continental than the Scandes and the Alps. (3) The areas in western Canada extend into much colder conditions that the Scandes and the Alps. (4) Overlap in climatic variables is concentrated in small areas. For these reasons further research is needed in western Canada to better understand permafrost in its mountains.
46The availability of reanalysis data and the method shown here can now better inform the transfer of cryosphere knowledge between mountains globally. Even though some variables in reanalysis data are subject to large biases and difference between products (Cao et al., 2020, Gruber, 2012b), they present a unique and homogeneous data source for cryosphere regions, which tend to be remote and sparsely measured. A growing number of studies show their value for permafrost and other cryosphere research (e.g., Cao et al., 2019).