Open access

Surface temperature inversion characteristics in dissimilar valleys, Yukon Canada

Publication: Arctic Science
2 June 2022

Abstract

Permafrost distribution in high-latitude continental mountains is a product of both latitudinal and elevationally controlled attributes. Frequently occurring surface-based temperature inversions (SBIs) significantly modify surface lapse rates (SLRs) annually. We aim to identify and quantify patterns of SBI characteristics in two proximal yet morphologically and vegetatively dissimilar central Yukon valleys. Elevational transect analysis is applied by using sensors in valley bottoms and 100 m upslope to determine in situ SLRs for the study period (August 2017 – August 2021). SLRs were shown to vary significantly between these dissimilar valleys. Climate reanalysis products (ClimateNA and Globsim) underestimated or almost entirely missed the presence of strong SBIs which produce annual average SLRs that range from 0.46 to 1.2 °C 100 m−1. The magnitude of these hyper-inversions was grossly underpredicted by previous surface air temperature modelling that attempted to account for SBIs across Yukon. Our results support the previously conceptualized framework that strong SBIs influence surface air temperatures and the pattern of permafrost distribution.

Résumé

La distribution du pergélisol dans les montagnes continentales des hautes latitudes est le produit d’attributs contrôlés à la fois par la latitude et par l’altitude. Les inversions de température en surface (ITS), qui se produisent fréquemment, modifient de manière significative les gradients verticaux de surface (GVS) chaque année. L’objectif des auteurs consiste à identifier et quantifier les caractéristiques des ITS dans deux vallées du centre du Yukon proches, mais dissemblables sur le plan morphologique et végétatif. L’analyse des transects d’élévation (ATE) est appliquée en utilisant des capteurs dans les fonds de vallée et à 100 m en amont pour déterminer les GVS in situ pour la période d’étude (août 2017 – août 2021). Il a été démontré que les GVS varient de manière significative entre ces vallées dissemblables. Les produits de réanalyse climatique (ClimateNA et Globsim) ont sous-estimé ou presque entièrement manqué la présence de fortes ITS qui produisent des GVS moyens annuels allant de 0,46 à 1,2 °C 100 m−1. L’ampleur de ces hyperinversions était largement sous-estimée par les modélisations antérieures de la température de l’air en surface qui tentaient de tenir compte des ITS dans tout le Yukon. Ces résultats appuient le cadre conceptuel précédent selon lequel de fortes ITS influencent les températures de l’air en surface et le patron de distribution du pergélisol. [Traduit par la Rédaction]

1. Introduction

As the impacts of climate change continue to be understood, it is more and more evident that high-latitude areas are undergoing complex and accelerated warming (Serreze and Barry 2011; Johannessen et al. 2016). Like other cold regions, high elevation areas are also undergoing a similar accelerated warming pattern (Wang et al. 2013; Pepin et al. 2015; Williamson et al. 2020). The mountains of Yukon represent terrain at both high elevation and high latitude. These areas are traditionally difficult to assess trends in climate due to limited accessibility and a general lack of climate observations (Palazzi et al. 2019). Although warming in northwestern North America is currently occurring rapidly (Porter et al. 2019), what is less evident is the nature and spatial distribution of warming over complex topography. Thus, a need exists for accurate representations of air and ground temperature over topography. For example, ecological studies require accurate landscape representations of temperature (Turner et al. 1989). Similarly, mountainous cryospheric elements including glaciers, icings, and permafrost also require this information (Gardner et al. 2009; Mernild and Liston 2010; Morse and Wolfe 2015; Smith and Bonnaventure 2017).
Air and ground temperatures primarily control the spatial distribution of permafrost (French and Williams 2017), which can have a profound impact on ecosystem services (Schuur and Mack 2018), community infrastructure, and water security (Hjort et al. 2018), landform structure and stability (Wang et al. 2009; Haeberli 2013; Ward Jones et al. 2019), hydrology (Quinton et al. 2011; Vonk et al. 2015) and act as a reservoir for carbon (Olefeldt et al. 2016) and exotoxins including mercury (Schuster et al. 2018; Schaefer et al. 2020). Permafrost distribution and thermal state are highly variable and difficult to predict across mountainous landscapes (Etzelmüller 2013). Mean annual air temperature (MAAT) is only one variable that influences permafrost distribution. Ground surface temperature and subsequently the temperature at the top of permafrost are influenced by many other variables that are often nonlinear across the landscape. These include snow depth (Garibaldi et al. 2021), substrate composition, and moisture level (Smith et al. 2010), slope angle (Haeberli et al. 2010), aspect (Cote 2002; Bonnaventure and Lewkowicz 2008), and vegetation cover (Shur and Jorgenson 2007; Kropp et al. 2020).
Perhaps the largest climatological difference regulating permafrost distribution in lower latitude mountains compared to those in high-latitude continental areas involves the combination of latitudinal and elevationally controlled permafrost (Bonnaventure et al. 2012). This results from the occurrence of persistent surface-based temperature inversions (SBIs), which influence MAAT across the landscape on an annual scale (Lewkowicz and Bonnaventure 2011; Lewkowicz et al. 2012). SBIs can occur frequently in high-latitude regions of northwestern Canada, particularly in the winter months, where they can persist for more than 80% of the time (Zhang et al. 2011). Bonnaventure and Lewkowicz (2013) conceptualized that persistent annual SBIs play a considerable role in permafrost distribution in these areas as average surface lapse rates (SLRs) can be gentle or inverted (positive) on average annually. This can result in permafrost having no lower limit and being nearly continuous in valley bottoms and mountain tops, while subsequently absent at mid-slope around the treeline where the top of the SBI most frequently occurs (Lewkowicz and Bonnaventure 2011).
SBI research in northwestern Canada is limited to high-latitude remote locations (Urban et al. 2013; Pepin et al. 2015). Therefore, traditional methods of measuring and monitoring SBIs utilizing radiosonde sensors (Bradley et al. 1992; Bourne et al. 2010; Zhang et al. 2011; Mayfield and Fochesatto 2013; N.C. Noad, P.P Bonnaventure, G. Gilson, H. Jiskoot, and M. Garibaldi, personal communication, 2022) only cover select locations in these immense regions. Another limiting factor is that radiosondes report on free-air lapse rates, which differ from SLRs. The use of remote sensing techniques (Liu and Key 2003; Liu et al. 2006; Devasthale et al. 2010; Chang et al. 2018) or climate reanalysis models (Palarz et al. 2018; Akperov et al. 2019; Luo et al. 2019; Shahi et al. 2020) have begun to fill gaps in monitoring SBI in remote regions. Climate reanalysis data sets of 2 m air temperature offer a way to examine past, current, and future surface air temperature (SAT)/SLR patterns in remote locations (Compo et al. 2011). Permafrost distribution modelling relies on the use of these data sets to understand the past, present, and future of MAAT (Obu et al. 2019; Tao et al. 2019; Qin et al. 2020), but there are limitations to the accuracy of this, particularly in mountainous areas due to their coarse spatial resolutions (Etzelmüller 2013).
This research applies an increasingly popular method to quantifying SBI characteristics and influence across topography through the setup of sensors along elevational transects from the valley bottom, up to surrounding mountain slopes. We have termed this methodology elevational transect analysis (ETA). These methods have allowed for the expansion of research on SBIs in remote areas of Finland (Pike et al. 2013; Williams and Thorp 2015), portions of Nunavut (Smith and Bonnaventure 2017; Garibaldi et al. 2021), and southern and central Yukon (Cote 2002; Lewkowicz and Bonnaventure 2011; Lewkowicz et al. 2012; Bonnaventure and Lewkowicz 2013). ETA gives a view of SBI conditions up a slope with a resolution not seen in other products or analyses. Unfortunately, both setup and data retrieval require considerable effort and time (e.g., Bonnaventure et al. 2012) and thus there are logistic and geographic limitations.
The objectives of this study are twofold. The first tests the hypothesis that dissimilar valleys experience frequent, strong, and persistent SBIs regardless of morphological and vegetative dissimilarities. Where previous studies have generalized the impact of SBIs on SLRs, we aim to highlight how this impact can differ even over short distances at the valley-to-valley scale. To do this, patterns of SBI characteristics in neighboring central Yukon valleys that are dissimilar will be quantified by using ETA. Second, the hypothesis that the impact of SBIs on SLRs in this region has heretofore been underestimated by modelling is tested. This has been completed by comparing observed SLRs to previously assumed regional SLRs and those derived using climate reanalysis products.

2. Study area

The study region includes an area of the complex mountainous terrain along approximately a 150 km stretch of the Dempster Highway in Yukon (Fig. 1). The Dempster Highway begins in a boreal forest ecoregion along its southern end and transitions quickly to a treeless tundra environment due to elevational and latitudinal climatic influences (Wahl 2004). This portion of the Dempster Highway is found within the Ogilvie Mountains, which is characterized by a series of narrow valleys predominantly covered with colluvium and alluvium surficial materials (Burn et al. 2015). The terrain in the study region was last glaciated during the Pre-Reid Glaciation (2.6 Ma to >200 Ka) but remained unglaciated in the subsequent more recent glaciations (Duk-Rodkin 1999). Most soils in the study region are classified as static cryosols (mineral soils without evidence of cryoturbation) (Scudder 1997). The study region has a subarctic continental climate, with modifications from intense cold-air drainage into valley bottoms that result in amplified cold winter temperatures (Burn et al. 2015). Permafrost in this region is classified as continuous (90%–100%), with low to moderate (<10%–20%) ground ice content (Heginbottom et al. 1995). Segregated and wedge ground ice were both mapped to be low to negligible in this area (O'Neill et al. 2019). Bonnaventure et al. (2012) modelled permafrost in this region at 30 × 30 m resolution. They found that permafrost probability in this study area ranges from being roughly 75%–100% at high elevations, while permafrost was least likely at mid-elevations, particularly on south-facing slopes (40%–50%). Valley bottoms, however, showed permafrost probabilities between 60% and 80% (Fig. 1).
Fig. 1.
Fig. 1. (a) Map of the subrange (≈120 km2) of the Ogilvie Mountains where the two valleys are located (≈10 km apart). The two valleys include the north valley (NV) and south valley (SV). In the NV, there are three sensors called the north valley weather station (NVWS), east-facing slope (EFS), and west-facing slope (WFS). In the SV, there are three sensor locations called the south valley weather station (SVWS), north-facing slope (NFS), and south-facing slope (SFS). Overlain on the map are the 30 × 30 m grid of permafrost probability (Bonnaventure et al. 2012), a hillshade (Natural Resources Canada 2022), and the Yukon Road Network polyline feature (Yukon Government 2022). The inset map in the right bottom corner situates the study area in Yukon by overlaying an Esri base map (Esri 2022a). (b) Map of all the transect locations with the comparator transects that include Engineer Creek transect (ECT) and Tombstone transect (TST). Overlays of this map include a hillshade (Natural Resources Canada 2022) and satellite imagery (Esri 2022b) to highlight the topography between the transects. Both maps have the “WGS 1984 Stereographic North Pole” projection, and the inset map has the “Yukon Albers” projection. The UTM coordinate system was used for each map. This map was created using ArcGIS® software by Esri. ArcGIS® and ArcMap™ are the intellectual property of Esri and are used herein under license. For more information about Esri® software, please visit www.esri.com.
The closest Environment Canada (EC) climate station to the study region is at Dawson, Yukon (∼160 km SW). The MAAT climate normal at Dawson from 1981 to 2010 was −4.1 °C with annual precipitation averaging 324.4 mm, 38% falling as snow (Environment Canada 2021a).
The study area consists of two mountain valleys located roughly 10 km apart in the same subrange of the Ogilvie Mountain Range (Fig. 1). The two valleys were selected as they have multiple differences between them, including vegetation cover, orientation, geometry, and fetch length (Fig. 2). South Valley (SV) is located on the southern edge of the subrange of mountains that surround Distincta Peak (Fig. 1) and drains to the east-south-east into a large north-south orientated valley that the Dempster Highway runs through. The SV bottom is at an elevation of 980 m a.s.l. with surrounding ridges that are 450 m higher to the north and 250 m higher to the south. SV is V-shaped and narrow with a gently sloping valley floor, approximately 250–300 m wide, running between the steep slopes (25°–30° on the south face slope and 30°–40° on the north face slope) (Fig. 2). Valley fetch length is 4.4 km. The surface geology of the SV consists of bedrock that is black weathered shale and laminated siltstone on the south face slope (SFS) and grey buff-weathering dolostone and limestone on the north face slope (NFS) (Yukon Geological Survey 2022). Vegetation cover is a primarily open boreal forest with Black Spruce (Picea mariana (Mill.) Britton, Sterns & Poggenb.) with thick mosses, lichens, Labrador tea (Rhododendron groenlandicum (Oeder) Kron & Judd), and grasses most dominant from valley bottom to roughly 100–150 m up the SFS (thinning moss on the steeper wall). Only small patches of mosses, lichens, and grasses are present on the NFS between large areas of exposed felsenmeer.
Fig. 2.
Fig. 2. Photos of each valley taken by the authors in August 2019 and 2021. (a) Is an areal photo of the SV looking up-valley from the mouth of the valley. (b) Is the weather station in the SV bottom. (c) Is an areal photo of the NV looking up-valley from the mouth of the valley. (d) Is the EFS microclimate station.
The north valley (NV) is located on the north part of the subrange and drains north into a larger basin (Fig. 2). The NV bottom has an elevation of 1090 m a.s.l with surrounding ridges roughly 350 m higher to the west and roughly 300 m higher to the east. The NV is U-shaped with a relatively flat 550 m wide valley floor between the steep slopes (30°–40°) with a fetch length of 5.7 km. Valley floor surficial geology is comprised of bedrock (black, weathered shale, and laminated siltstone) with grey and buff-weathered dolostone and limestone on surrounding slopes (Yukon Geological Survey 2022). There are no trees in the valley bottom with tundra-like vegetation (mosses and lichens) on the gentle slope (<1140 m a.s.l.). The steep slopes have patches of thin mosses and lichens surrounded by exposed felsenmeer.
Two valleys with comparator transects were also selected to assist in situating the SLRs observed regionally (Fig. 1b). Approximately 25 km north of the NV is the Engineer Creek transect (ECT) valleys and approximately 55 km south of the SV is the Tombstone transect (TST) valley. The ECT valley is covered with the dense boreal forest. The TST valley is also covered with boreal forest, which is not as thick as the ECT valley.

3. Methodology

3.1. Station deployment and data collection

To quantify the variability of SLRs over a local valley-to-valley scale, two valley locations were selected using an initial search in Google Earth Pro. The experimental design aimed to test SBI variability between valleys with different morphological and vegetative characteristics. Locations needed to be within 3 km of the Dempster Highway as deployment, and data retrieval were to be done on foot. In each valley, three sites were chosen to set up sensors (Table 1).
Table 1.
Table 1. Study area valley station geographic coordinates, elevation, and vegetation characteristics.
Weather stations (HOBO USB Micro Station Data Logger—H21-USB; Table 2) were set up in each valley bottom roughly 1000–1500 m away from the Dempster Highway. This distance assured that measurements were outside of the influence of anthropogenic effects from the highway including snow clearing. The weather stations recorded wind speed and direction, atmospheric/relative humidity, incoming solar radiation, 2 m air temperature, ground temperature, and temperature at depth sensors (Fig. 2; Table 2). Ground sensors were placed at roughly 5 cm beneath the surface, and depth sensors were positioned at the best approximation of the depth of the top of permafrost (TTOP) layer (range from 25 cm to 70 cm) (listed in Table 6). Depth sensors were installed using a frost probe to create a small hole down to the depth of the frost table or where coarse substrate impeded progress.
Table 2.
Table 2. Uncertainties and resolution of sensors used at microclimate and weather stations in each valley as listed by Onset.
Microclimate stations in the SV were set up 93 vertical m up each slope from the weather station. This vertical distance was selected to coincide with our best estimation of treeline's elevation on the SFS. Treeline was defined as the elevation at which the presence of trees became isolated to absent. This was the only location on both slopes that roughly resembled the vertical distance in the SV and had patches of moss that were capable of securely supporting the station on the felsenmeer slope. The microclimate stations consisted of a metal fence pole and a radiation shield (Onset RS1) housing the air temperature sensors and secured with four guidewires (Fig. 2). Sensors recorded air, ground, and top of permafrost temperature or maximum depth if substrate impeded (Depths listed in Table 6) and were all U23-001 HOBO dataloggers (Table 2).
Vegetation cover and snow cover are variable across the landscape. These factors influence the relationship between air and ground surface temperatures. Due to SBIs being most influential during the freezing degree-days, freezing n-factors were calculated by summation of all freezing degree-days for both ground and air sensors. The total ground freezing degree-days were divided by total air freezing degree-days at each of the six sites resulting in freezing n-factors (Riseborough et al. 2008; Garibaldi et al. 2021).
Two additional nearby elevational transects were also analysed to illustrate the occurrence of strong regional SBIs (Fig. 1b). Comparator transects have a consistent instrumental setup to the valley locations; however, valley bottoms only contain microclimate stations. ECT is located approximately 25 km north of the NV, has a southeast-facing aspect, and spans a vertical distance of 200 m (668–868 m a.s.l). TST is located approximately 55 km south of the SV, has a southwest-facing aspect, and spans 113 vertical m (966–1079 m a.s.l.).
At each sensor location, data were collected at a sampling interval of two-hours beginning at 0:00 PST on 28 August 2017, with a few exceptions to when the stations were deployed (EFS 15 August 2018, and TST 7 August 2019) (Table 3). Final measurements were taken between 8 and 10 August 2021 with a few exceptions due to equipment failure (SV weather station 20 July 2021, and EFS 11 December 2020) (Table 3).
Table 3.
Table 3. Gantt chart of when the sensors were deployed and the duration when data were collected.

3.2. Calculation of SBI characteristics

For the in situ and downscaled climate reanalysis data, SLRs were calculated from the raw temperature data by least-squares linear regression to determine a slope of the linear line representing temperature change between the valley bottom and higher elevation stations. Using these data, SBI metrics were calculated including the average of the inverted surface lapse rates (ISLRs) and frequency. ISLRs represent SBI strength (Bradley et al. 1992) by describing on average how much warmer the air temperature is with increasing elevation when an SBI is present. To calculate ISLRs the mean between all SLRs recorded when an SBI is present (SLRs that were >0 °C 100 m−1) was taken. SBI frequency (Ifreq) is the proportion of time that inversions are present (Bourne et al. 2010). When SLRs become ≥1 °C 100 m−1, we considered them to be hyper-inverted. Using these methods, SLR and SBI characteristics were calculated for every 2-h interval for each of the four elevational transects. Each inversion characteristic was calculated and averaged seasonally (based on meteorological seasons, e.g., winter is December, January, and February), annually, and during the complete study period.
The synchronicity of SLRs between each of the valley transects along with the comparator transects was reviewed. To calculate synchronicity between each transect, SLR readings between the compared transects that were all positive (inverted) or all negative (normal) were selected. The total number of readings that had SLRs that were synchronously positive or negative between the compared transects were divided by the total number of readings to give a proportion of time where SLRs sign (all positive or all negative) was synchronous. Then a statistical test was completed to calculate the Pearson correlation (PC) between SLRs on each transect. PC highlighted any significant positive or negative correlation between SLRs on transects. A strong positive PC would indicate that there is high synchronicity between SLRs of two transects.
A classification system developed by Whiteman et al. (2001) and modified by N.C. Noad, P.P Bonnaventure, G. Gilson, H. Jiskoot, and M. Garibaldi (personal communication, 2022) was used to describe inversion event length. Inversion events were classified as either transient or persistent based on length. Persistent inversions are defined to last 18 h or longer without reversal of lapse rates back to normal (≥0 °C 100 m−1), thus no SBI break-up event. Transient inversions refer to events that are less than 18 h in length and that do not last beyond the normal diurnal pattern (inversion development at night and breakup during the day) (Whiteman et al. 2001). Classifying SBI using this system aids in understanding how the length of inversions varies spatially. The assumption made while classifying the length of inversion events is that the inversion does not break up and redevelop within a 2-h increment. The result of such an assumption may be overprediction of how long some persistent inversions are.
Daytime and nighttime readings were selected based on the sunrise and sunset time calculated for each day of the year using the NOAA sunrise and sunset calculator (NOAA 2021a) (Table 4). The coordinates of WS01 were inputted into the calculator to represent both valleys. The times for sunrise and sunset predicted by the calculator were not different between the valleys due to their proximity. These sunrise and sunset times do not account for topographic shading. The output sunrise and sunset values were then used to select whether the reading was taken before/after sunrise or sunset, thereby separating all SLR data points into nighttime or daytime readings.
Table 4.
Table 4. The sunrise, sunset, daylight hours, nighttime, hours at each solstice, and equinox for the study area are based on Yukon Standard Time (UTC -7, permanent daylight savings time) (NOAA 2021a).

3.3. Inversion characteristics using climate reanalysis models

GlobSim Version 1.6 is a Python script program that downloads climate reanalysis 2 m air temperature data from their respective servers, interpolates the data across an area, and then scales it through empirical-statistical methods to be 2-h increment point data (10 × 10 m grid) for a sample of the study period (August 2017–2019) (Cao et al. 2019). This assessed how well the downscaling program accounted for the strong SBIs in this region. Globsim downloads data from ERA5, MERRA2, and JRA55 (Japan Meteorological Agency/Japan 2013; Global Modeling and Assimilation Office 2015; Copernicus Climate Change Service 2017, updated monthly)., The climate reanalysis data sets are vertically interpolated through pressure levels present for each climate variable in the original data set, matching the elevation of the input location. Each reanalysis product used for GlobSim has a different number of model levels for vertical readings, but these are typically interpolated into pressure levels that increase in altitude/elevation in 25 hPa increments for the lowest part of the atmosphere (1000–700 hPa) (Martineau et al. 2018; Graham et al. 2019). Vertical interpolation between the pressure levels is calculated using the geopotential height of each reading and situating it to the elevation of the local point to interpolate SLR. Using the NOAA (2021b) Pressure Altitude Calculator, rough estimates for the predicted elevations/altitudes at each pressure level were made. The valley and slope locations fall in between the pressure level of 900 and 875 hPa which are predicted to be at roughly 990 m a.s.l and 1220 m a.s.l. Downscaled climate reanalysis data sets of SATs are then scaled temporally to match the desired 2-h incremental data sets. Distincta Peak is at an elevation of 1760 m a.s.l. so SLRs between the SV bottom (993 m a.s.l.) span across pressure levels 900 hPa to a little higher than pressure level 825 hPa. This means that this transect incapsulates roughly four pressure levels.
A second program called ClimateNA (Version 6.3) (Wang et al. 2016) was used to obtain monthly average 2 m air temperatures at each site to calculate SLRs for a sample of the study period (August 2017 – August 2019). This program downscales monthly climate averages using a cluster of 800 × 800 m cells to scaleless point data. Elevations for the nearest eight surrounding cells are used to estimate a generalized regional SLR. ClimateNA has limitations in predicting variation of lapse rates across a region (Wang et al. 2016).

3.4. In situ data comparison to local MAAT interpolations

In situ Annual Mean Air Temperature (AMAT) data were compared to a modelled surface of MAATs that was developed using a series of logger transects and records from 18 EC weather stations across central Yukon and northern British Columbia (Lewkowicz et al. 2012). This 30 × 30 m resolution MAAT surface was utilized as one of the variables to regionally model permafrost probability accounting for dominant SLR patterns (Bonnaventure et al. 2012) (Fig. 1). In the creation of this MAAT surface, SLRs were defined using ETA from transects at multiple locations across the south and central Yukon. The surface is not a perfect comparison, as it was constructed for the 1981–2010 climatic normal and was applied using various assumptions. The most notable of these include interpolated surfaces for the position of treeline (where SLR is assumed to change in sign and magnitude), as well as the values of SLRs themselves. Treeline was interpolated in the model through the collection of sample points of treeline using Google Earth imagery across the modeled area. In the SV, the treeline is predicted by the model to be at roughly 1180 m a.s.l and was observed to be at around 1130 m a.s.l. Treeline in the NV is predicted to be at 1170 m a.s.l. but was in reality below 1100 m a.s.l. This is critical as even small deviations in these variables between the model and the observed data at the valley scale, will result in substantially different annual temperature values. This analysis aimed to not only examine the predicted and actual absolute temperature between the modelled MAAT surface and the in situ reading but more specifically predicted SLRs compared to observed in situ SLRs. The comparison of this MAAT surface with our in situ data will expose limitations in accounting for SBI influence in this region for permafrost probability mapping. By examining the limitations of previous modelling assumptions for this study region, improvements to air temperature and subsequent permafrost models can be made.

3.5. Statistical analysis

All statistical analyses were undertaken in SPSS version 27. Due to the number of sample points, the 2-h interval at a seasonal and annual scale was assumed to be normally distributed; thus, analysis of variance (ANOVA) and Tukey's post hoc statistical testing were used to compare the means of inversion characteristics between each transect (Abdi and Williams 2010).

4. Results

4.1. Study period climatic conditions

Compared with the normal, 2019 was the warmest year examined at Dawson, Mayo, and Old Crow (+1.9, +1.2, and +2.8 °C, respectively) (Table 5) (Environment Canada 2021a, 2021b). The coldest year over the study period at all sites was 2020 (−1.3, −1.8, and +0.6 °C, respectively). Recent amplified climate warming (upwards to 3.25 °C) has been observed in the region from 1948 to 2016 (Zhang et al. 2019). To account for rapid climate warming, average temperatures over a 10-year record (2010–2019) were compiled from the nearest EC stations (e.g., Garibaldi et al. 2021). Our observation years were similar to recent (2010–2019) average observations (including the Rock River station installed in 1994).
Table 5.
Table 5. Coordinates, elevations, MAATs (1981–2010) (Environment Canada 2021a), decadal AMAT (2010–2019), and AMATs for each year of the study period (2017–2021) (Environment Canada 2021b) are included for each of the nearest Environment Canada climate stations.

4.2. Mean annual site temperatures

AMAT ranged at 2.3 °C between the warmest and coldest sites (EFS excluded due to >1.5-year data gap) (Table 6). Colder AMATs were recorded at all SV locations compared to NV locations. In both valleys, valley bottom locations displayed the coldest AMAT while upslope locations were warmer. Mean annual ground surface temperature (MAGST) between the sites followed a different pattern than the AMAT. MAGSTs in the SV were generally warmer than the MAGSTs in the NV. For example, while the air temperature at the SV location was 2.0 °C colder than the NV location; the MAGST was 1.6 °C warmer. Calculated n-factors from the ground–air temperature relationship varied significantly between sites. In the SV freezing n-factors were low (SV = 0.29, SFS = 0.19, NFS = 0.39), while they were higher in the NV (NV = 0.57, WFS = 0.79, EFS = 0.87) implying more surface to atmospheric connectivity. Depth sensors were set up at the top of the permafrost/base of the active layer. Though this was not possible at the slope locations due to the coarse substrate. Thus, permafrost was only verified in each valley bottom location. That said, five of the six locations showed average annual depth temperatures below 0 °C supporting the notion of permafrost below the station. The one exception was the SFS at around treeline where the MAGTD was too warm (MAGTD = 1.8 °C) to support the notion of permafrost. Depth sensors on the slopes in the NV (WFS and EFS) could not be set up as deep as others due to the coarse substrate. Thus, it is hard to compare these sites to those in the SV. The valley bottom location in the NV was set up at the top of the permafrost layer like the valley bottom SV location. Comparing these two sites indicate that the depth temperature in the NV is 1.2 °C colder than in the SV.
Table 6.
Table 6. Annual mean air temperatures (AMAT), mean annual ground surface temperatures (MAGST), and mean annual ground temperature at depth (MAGTD) for each of the six sensor locations in °C. Also included are the depths from the surface each MAGTD sensor is positioned.

4.3. Study period, annual, and seasonal inversion characteristics from in situ sensors

Ifreq was greatest in the SV which resulted in the valley bottom having a substantially colder AMAT than surrounding sites (Table 7). SBIs in the SV persisted for longer durations on average (roughly 4–5 h), which led to a lower total number of separate annual SBI events throughout the study period. Inversion strength measured by ISLR was comparable between the four valley transects, ranging by 0.49 °C 100 m−1. ISLRs were significantly (P < 0.01) different between the four valley transects except for between the NFT and the EFT (P = 0.50). ISLRs also ranged to a greater degree (1.07 °C 100 m−1), with the two comparator transects outside the study valleys. SLRs in the SV had a significantly (P < 0.00) more inverted SLR on average than the NV. The SLRs in SV were most like the ECT, while the SLRs in the NV were more comparable to the TST (pattern not due to valley proximity). The monthly mean Ifreq on each transect was significantly (P< 0.04) different between the SV (south face transect, SFT, and north face transect, NFT) and NV (east face transect, EFT, and west face transect, WFT) transects.
Table 7.
Table 7. Total number of distinct inversion events, the Ifreq, the average length of inversion events, the average SBI strength measured through the mean inverted surface lapse rates (ISLRs), and the average SLR in each valley across the study period (August 2017 – August 2021). The * indicates that there are some gaps in the data collected (Section 3.1).
Both Ifreq and SLRs are marginally largest during year four of the study (August 2020 – August 2021) and marginally lowest in year 3 (August 2019 – August 2020) (Table 8). ISLRs were to a lesser degree larger in the second year, but this is only marginal compared with the other two characteristics. The mean of the SLRs and ISLRs on the SFT, NFT, and WFT differed significantly (P < 0.01) between the four years.
Table 8.
Table 8. Annual (based on each year from August to August during the study period) SBI characteristics at each of the four elevational transects set up between the two valley locations. Annual SLRs, ISLRs, and Ifreq are included. * Indicates that only part of the year of data was available (Section 3.1).
For each of the two SBI characteristics of ISLR and Ifreq, magnitudes were significantly (P < 0.05) greatest in the winter (December–February) (Fig. 3). This resulted in subsequent SLRs being most inverted during the winter. Moreover, the range of Ifreq and SLRs between the four elevational transects is greatest during the winter. The SLRs are only normal on a seasonal average in the summer (June–August) at three of the four transects. In all other seasons at each transect, SLRs were on average inverted or neutral.
Fig. 3.
Fig. 3. Seasonal SBI characteristics at each of the four elevational transects set up between the two valley locations. Seasonal ISLRs, Ifreq, and SLRs during the study period (August 2017–2021) are included.

4.4. SLR synchronicity between the valleys and nearby transects

The frequency that SLRs on multiple transects were the same sign (both positive or both negative) varied between different combinations of transects (Table 9). Overall, results here suggest that transects within the same valley have the most synchronicity while synchronicity with transects outside the same valley is variable. Variability of synchronicity is not necessarily related to transect proximity. For example, the ECT is located 10 km closer to the NV transect (25 km overall) than the SV, but the PC is roughly 0.2–0.3 greater (0.664–0.690 in the SV versus 0.393–0.414 in the NV) between the SV and ECT (Table 10). Overall, the PC was positive and statistically significant (P < 0.05) between most sites though there was a varying in how strongly positive.
Table 9.
Table 9. The probability that SLRs on each transect are inverted or normal (positive or negative) all at the same time.
Table 10.
Table 10. For the study period (August 2017 – August 2021), the Pearson correlation (PC) was calculated between each transect and the statistical significance of each correlation (including ECT and TST).
Transects in the SV are most correlated in the winter and are much less correlated in the summer (PC = 0.977 in the winter and 0.795 in the summer) (not shown). Similarly, the correlation between transects in the NV shares this same pattern (0.954 in the winter and 0.691 in the summer). The correlation between transects in the two valleys is relatively consistent throughout the year at around a PC of 0.600.

4.5. SBI event length classification

For each of the transects, there were more individual transient SBI events than persistent events (Fig. 4). These made up the highest proportion of the total number of SBI events in both valleys during the study period (August 2017 – August 2021). Persistent SBI events dominated the proportion (≥55%) of the total SBI hours in the SV, while they made up nearly half of the total SBI hours in the NV.
Fig. 4.
Fig. 4. The graphs on the right illustrate the number of transient and persistent SBI events that occurred on each transect, (a) SFT, (b) NFT, (c) WFT, and (d) EFT on seasonal and annual average during the study period (August 2017 – August 2021). The graphs on the left represent the proportion of total SBI hours attributed to each SBI length classification for the four elevational transects during the study period. Note that the east-facing slope transect consists of data only from August 2018 – December 2020.
Likewise, to patterns observed on an annual scale, persistent SBI events remained less frequent than transient SBI events seasonally, even in the winter. Though they represented a much lower proportion of the total SBI events, the persistent SBI events did account for a substantial number of total hours with an SBI present. In the winter, there is the greatest number of persistent events and a peak proportion of hours (up to 89% on the SFT) with an SBI present that is classified as persistent. Similarly, in the autumn season persistent SBIs, to a lesser degree, dominated the proportion (71% on the SFT) of total hours with an SBI present. Spring and summer seasons saw a dramatic drop in the number of persistent SBI events and subsequent proportion (≤33%) of hours represented by these events with the only exception being the SFT during the spring (54%).
The longest SBI event lasted for nearly 16 days (based on a 2-h temporal resolution), and this event occurred in the SV on each transect (December 2019) (Table 11). In the NV, the WFT had an SBI event last roughly 9 days and the EFT roughly 6.5 days. The mean persistent SBI event duration was substantially longer on the SV transects. Mean SBIs were much greater than the median SBI lengths and thus were skewed to be longer by a few outlying lengthy SBI events. Median values imply that most of these persistent inversion events are approximately 1 day (24 h) long.
Table 11.
Table 11. Attributes of persistent SBI events are described including the maximum length of persistent SBI events, the mean length of persistent SBI events, and the median length of persistent SBI events during the study period (August 2017 – August 2021).

4.6. Night vs. daytime inversions

Ifreq was significantly (P < 0.001) greater during the nighttime hours (0.75) throughout the year compared to the daytime hours (0.35) (Table 12). SLRs during the daylight hours on an annual average were gently inverted or gently normal depending on the transect. During the night, SLRs were hyper-inverted, averaging over 2 °C 100 m−1 in the SV while averaging closer to 1 °C 100 m−1 in the NV. This significant (P < 0.001) diurnal pattern of SLRs is exemplified by the summer average SLRs which were inverted during the night (0.42–0.94 °C 100 m−1) and normal during daylight hours (−0.17 – −0.58 °C 100 m−1) (Fig. 5). This significant diurnal pattern of SLRs breaks down with low sun angles and short days in the winter when SBIs most frequently persist beyond the normal diurnal pattern observed in the summer.
Table 12.
Table 12. The annual average Ifreq, ISLRs, and SLRs for both daytime and nighttime readings at each transect during the study period (August 2017 – August 2021).
Fig. 5.
Fig. 5. SLRs on each transect during night and day readings (a) winter and (b) summer for the study period (August 2017 – August 2021). Error bars indicate ±1 standard deviation from the 2-h increment data set. Under each transect name, there is a sample size for daytime (nD) and nighttime (nN) readings.

4.7. In situ measures compared to modelled MAAT

MAAT in the SV was predicted by the modelled MAAT surface (Lewkowicz et al. 2012) to be −4.6 °C, −4.5 °C, and −4.4 °C at the weather station, and NFS and SFS microclimate stations, respectively. Using the MAATs predicted by the model SLRs would be roughly 0.2 °C 100 m−1 in the SV. Thus, the influence of the SBIs on SLRs in this valley was grossly underpredicted. The actual annual SLRs in this portion of the valley as measured by in situ sensors were 0.9–1.2 °C 100 m−1. The NV was positioned above treeline in reality, but not according to the MAAT model (treeline predicted to be 1170 m a.s.l in the NV and 1180 m a.s.l. in the SV). Thus, partway up each slope SLRs were predicted to be inverted on an annual average, while above 1170 m a.s.l. SLRs reversed back to the assumed normal (−0.65 °C 100 m−1). As a result, MAAT was predicted by the model to be −4.78 °C at the NV bottom weather station, −4.83 °C on the EFS, and −4.90 °C on the WFS. Therefore, SLRs between the sites would have been roughly −0.05 to −0.12 C 100 m−1. According to our in situ data, SLRs on these two transects were still inverted at roughly 0.4–0.5 °C 100 m−1 on an annual average. This indicates that again, the influence of SBIs on SLRs in the NV was underpredicted by the model. These underpredictions of inverted SLRs resulted from assumptions that SLRs above treeline are normal on an annual average. This assumption can neither be confirmed nor refuted as the transects set for this research did not extend above treeline (future work will extend these transects above treeline).

4.8. Predicted SLRs using climate reanalysis data sets

On annual and seasonal scales, each of the climate reanalysis data sets ERA5, JRA-55, and MERRA-2 had SATs that were predicted to be nearly the same between all sites including the local high point, Distincta Peak (1760 m a.s.l.) (Fig. 1) (For the first 2 years of the study, August 2017–2019.) This is exemplified as winter seasonal SLRs for each model were on average near 0 °C 100 m−1 for each transect (Fig. 6). A large elevational transect from the SV valley bottom to Distincta Peak with a vertical distance of around 700–800 m (spanned four pressure levels in each model) was similarly predicted to be near 0 °C 100 m−1. ClimateNA predicts annual average SLRs that are closest to being representative on the WFT, SV bottom, and Distincta Peak transect. These SLRs are still predicted to be roughly 10 times smaller than the actual observed SLRs. No seasonal pattern of more strongly inverted winter SLRs was observed in the predicted SLRs.
Fig. 6.
Fig. 6. Comparison of winter season SLRs predicted using downscaled climate reanalysis data. The bars for each elevation transect are the actual average in situ measured wintertime SLRs.

5. Discussion

5.1. SBI characteristics in the two valleys

Inversion strength (represented by ISLRs) and Ifreq were greatest in the winter season, which is consistent with other research completed on high-latitude SBIs (Seidel et al. 2010; Zhang et al. 2011; Mayfield and Fochesatto 2013; Shahi et al. 2020). This pattern suggests the dominance of anticyclonic conditions in the region during the winter months (Mayfield and Fochesatto 2013) and the prevalence of cold-air drainage resulting in cold pooling due to limited incoming solar radiation (Daly et al. 2010). Similar to findings in Shahi et al. (2020), the spatial variation of ISLRs and Ifreq between the two valleys was greatest during the winter months. Furthermore, during the nighttime hours throughout the study period, SBIs were significantly more frequent than daytime readings resulting in inverted SLRs. This pattern is also consistent with findings observed in other literature regarding SBI characteristics on a diurnal scale (Whiteman 1982; Whiteman and Richland 2000).
Diurnal and seasonal variation of SBI frequency drives SLR variations spatially and temporally between the four sites. For example, Ifreq on the NFT was roughly 75% during the nighttime readings across the study period compared to roughly 35% during the daytime readings (Table 12). In comparison, ISLRs on the NFT were 2.8 °C 100 m−1 during the night and 2.4 °C 100 m−1 during the day. Thus, Ifreq between night and day is driven by diurnal and seasonal patterns, while ISLRs (SBI strength) are not driven as strongly by these same patterns.
The two SV transects had average SBI event lengths of 13.5–15 h, while in the NV SBI event length on average lasted only between 8.5 and 10 h (Table 8). Between the four transects, persistent SBIs were longer and more frequent in the SV leading to a more regular occurrence of inverted SLRs. Thus, these persistent SBIs play a substantial role in shaping SBI frequency and the variability of SLRs spatially on this valley-to-valley scale.
The evolution of persistent cold pooling and subsequent SBI events has been studied extensively (Wolyn and McKee 1989; Zhong et al. 2001; Zangl 2003; Yu et al. 2017; McCaffrey et al. 2019; Sun and Holmes 2019). For example, persistent cold-air pools and subsequent SBIs develop through cold-air drainage and (or) synoptic conditions that lead to warm air advection aloft or subsidence of air aloft producing a warmer air cap above the cold air layer near the surface (Lareau et al. 2013). Terrain that surrounds some valleys can block winds aloft from mixing to the surface which generates conditions optimal for SBIs to develop and persist (Pichugina et al. 2019). Cold-air pools and subsequent SBIs in the Columbia Basin of the northwest United States were found to persist until winds from approaching Pacific frontal systems mixed out the SBI layer (Whiteman et al. 2001). A study of the development and evolution of persistent cold-pooling events in high-latitude areas, particularly those in the mountainous areas of northwestern Canada, is needed to fill research gaps.
Overall, intense cold-air pooling and subsequent SBI development and persistence result in temperature patterns where low-lying and high-elevation locations are coldest and mid-elevations are warmest. Elevation-dependent warming is predicted to occur as climate change is accelerated in high elevations faster, and more intensely than in lower elevations (Pepin et al. 2015). Elevation-dependent warming has recently been observed in Southern Yukon (Williamson et al. 2020). Cold-air pooling and subsequent SBIs have been theorized to alter patterns of climate change through the decoupling of the lower atmosphere from synoptic conditions (Daly et al. 2010). This has led to reduced climate warming in the winter season at valley bottom locations susceptible to SBIs and atmospheric decoupling (Pepin et al. 2011). One study in the Italian Alps suggests the SBIs in the Po Valley may become more frequent as the climate warms (Caserini et al. 2017). In Australia, SBIs have broadly been observed to be decreasing in frequency and strength between 1961 and 2017 (Hiebl and Schöner 2018). All these studies review SBIs and their relationship to climate change in mid-latitude mountainous areas (American Rockies, European Alps, Australia). A global approach to predicting change in the number of hours with an SBI present between two time periods (1951–1980 and 1981–2010) found that there are some areas of increase and some of decrease depending on the location. Northwestern Canada was projected to have only as high of a 10% increase in SBI hours with most locations having a near-zero change (Hou and Wu 2016). More recently, SBI impact on SATs in northwestern Canada highlighted some evidence for reduction of inversion impact on SAT over time (N.C. Noad, P.P Bonnaventure, G. Gilson, H. Jiskoot, and M. Garibaldi, personal communication, 2022). While this pattern is far from confirmed, much future care and research into this topic are essential. If a change in the vertical temperature structure of the atmosphere is occurring, then valley bottom locations could be impacted by rates of warming that may differ from higher elevations.

5.2. Temporal and spatial comparison of SBI characteristics between the two valleys

The main spatial difference between the four transects is that SLRs between the two valleys were more inverted on an annual scale in the SV. The main temporal finding between the transects is that there is greater synchronicity between SLRs within each valley when compared to transects in other valleys. These two findings suggest that the microclimate factors within each valley (e.g., tree-covered SFS vs. treeless NFS) have the most limited effect on defining SLRs. The orientation of the valley likely plays a more significant role in defining SLR characteristics than the slopes with opposite (e.g., north-facings vs. south-facing) aspects within the valley. One hypothesis explaining this tendency could be that the NV is orientated in a direction that allows for channelling of wind through the valley increasing turbulence (Drobinski et al. 2003) and subsequently mixing cold pools out. Furthermore, the SV bottom is around 990 m a.s.l. while the NV is a little more than 100 m higher at 1100 m a.s.l. This could result in one location having surrounding topography that blocks synoptic winds from channelling through the valley (Pichugina et al. 2019). Another hypothesis is that there is a forest in the valley bottom and partway up the SFS. Increased surface friction may reduce wind speed and mixing of near-surface air with the free atmosphere (Wahl 2004; Lewkowicz and Bonnaventure 2011). Thus, conditions in the forest-covered SV could be more optimal for SBIs to develop and persist. Another hypothesis is that the difference in transect vertical length in each valley could result in variable SLR measurements between the two valleys. In the SV, transects were only about 140 vertical m in length while in the NV they were about 92 vertical m in length. This was due to a lack of locations where a microclimate could be supported on the EFS and WFS in the NV.
Synchronicity between the SFS and NFS is reduced in the spring, summer, and autumn seasons. One possible explanation is that orographic shading results in the SFS warming faster after sunrise than the valley bottom or the NFS. Furthermore, Ifreq during the day is greater on the SFT than on any other transect, including the NFT which indicates that positive SLRs may occur during the daylight hours on the SFT due to warming of the slope location quicker than the valley bottom. Similar observations were made by Williams and Thorp (2015) during the spring in Lapland, Finland. This would explain why the daytime annual average SLR that is inverted on the SFT (0.54 °C 100 m−1) is significantly (P < 0.001) different from the means of the other transects that have an annual average daytime SLR that is normal or neutral, including the NFT (−0.14 °C 100 m−1). In the winter season, aspect plays less of a role in defining SLRs synchronicity within the SV as the sun angle is low (1.5 °–17.0 °) (NOAA 2021a) and much of the SV (including the SFS location) is orographically shaded during the short days. Furthermore, SBIs are much more frequent and are significantly longer across the study area in the winter season and positive SLRs are likely more widespread throughout the valley. Thus, widespread conditions optimal for SBI development and persistence play a much stronger role that aspect in defining wintertime SLRs when compared to other seasons.
Synchronicity between transects in the SV and the ECT was greater than the synchronicity of any of those transects compared to the NV, which was located between the SFT and ECT. This could be evidence that the orientation of the valley the transects are situated in, plays a larger role than the proximity of the valleys to each other. This could also be due to where each of the valleys drains into. The NV drains into a large basin directly to its north where elevation continues to drop as you move northward (Fig. 1b). Meanwhile, the SV drains into another somewhat wider valley that is orientated southwest to northeast. A large plateau drains into this valley from just south of the outlet of the SV. The Dempster Highway follows this valley (down in elevation) towards the northeast until it meets up with the same drainage basin the NV drains into (this would be a local low elevation as the Dempster Highway follows the south edge of the basin westwards up in elevation towards the outlet of the NV). As an SBI develops, downslope winds are common (Gustavsson et al. 1998), meaning that winds in the NV can outlet into a large basin while the SV only outlets into another valley that already likely has cold air draining into it from the plateau to the south. This could effectively reduce the ability of air movement through the SV resulting in limited drainage of the pooled cold-air out of the valley, especially when compared to the NV. The NV is a valley at a higher elevation relative to the surrounding terrain and thus may more efficiently allow for downslope winds that drain cold air out of the valley and thereby limit cold-air pooling/SBI development and persistence. The ECT is in a location that has relatively low elevation compared to the surrounding terrain. It is located on a southeast-facing slope that borders a larger valley/basin that can only outlet through a narrow valley that is roughly 2 km northeast of where the transect is located (the Dempster Highway follows this narrow valley down in elevation out of this wide valley). Thus, once again, cold-air pools at this transect location may not have as efficient of an outlet as the NV does. This could explain why synchronicity is high between the SV and the ECT transect in the winter when SBIs dominate at both locations.
Overall, there are several potential reasons why there are spatial and temporal variations of SBI characteristics between each valley. Answering which of these factors is contributing to patterns seen in each valley is beyond the scope of this paper. Continued study of transects in these valleys and surrounding locations may provide insight into these conceptualizations. Furthermore, weather station data currently collected in these valleys will aid in providing some answers to these questions as research of SLRs in these valleys continues.

5.3. Modelled SLRs compared to observed SLRs

Minder et al. (2010) highlighted the errors that came when predicting SAT in complex terrain using assumed average SLRs in the Cascade Mountains of Washington. Cold-air pooling is intense in the SV, and studies have indicated that there is at times a susceptibility of interpolated temperature grids to underrepresent local cold-air pooling (Lundquist et al. 2008; Frei 2014). Modelled MAAT and associated SLRs (Lewkowicz et al. 2012) grossly underpredicted the influence of SBIs in these valleys. This resulted in valley slope locations being significantly cooler than actual measurements indicate. The valley bottom MAATs were predicted in the SV to be roughly 0.6 °C warmer than were recorded by in situ sensors. Furthermore, the model was based on the 1981–2010 climate normal, which was shown in our analysis of climate to be outdated. This would conceivably result in MAATs predicted to be colder than the current climate of this fast-warming region, but, in this case, they were predicted to be substantially warmer than they are. Thus, significant error in predicted temperatures across this mountainous landscape is introduced if SLRs are assumed to average the expected normal SLR (−0.65 °C 100 m−1) or even if predicted to be gently inverted on an annual basis. This error is also magnified in the SV where SBIs are more prevalent and intense throughout the year.
Another modelled source of temperature data in the valleys was downscaled climate reanalysis data. GlobSim and ClimateNA missed most of the impact SBIs had on SLRs (Fig. 6). In most cases, SLRs on annual and seasonal averages were predicted to be near zero for the lowest 100 m of the atmosphere. Another study where permafrost was modelled for high-latitude areas using reanalysis data sets (ERA-Interim) found similar issues when it came to valleys prone to frequent SBIs (Obu et al. 2019). Similarly, downscaled regional climate reanalysis data sets (MM5 Regional Forecast Model) in the Pacific Northwest Cascade Mountains when compared with in situ measures were more steeply negative (−0.57 °C 100 m−1) than the in situ measures (−0.39 to −0.47 °C 100 m−1) (Minder et al. 2010). These results raise questions regarding the validity of using downscaled climate reanalysis data sets for determining SLRs in areas of complex mountainous terrain, especially those prone to frequent SBIs.
Overall, results of this research confirm hyper-inversions within the first 100–150 m of the valley bottom (below treeline in the SV), resulting in strongly inverted annual average SLRs. SBI influence, while not as strong in the NV that is entirely above treeline, remains significant in producing inverted SLRs on an annual basis. Data collected and assumptions made in other studies suggest that annual average SLRs reverse back to normal (negative) above treeline (Wahl 2004; Lewkowicz and Bonnaventure 2011; Lewkowicz et al. 2012). Thus, strongly inverted SLRs on an annual average in the NV raises questions regarding the generalized assumption of annual SLR reversal to negative above treeline, especially for above treeline valleys. One recommendation we suggest may improve modelling developed by Lewkowicz et al. (2012) is the inclusion of a Topographic Position Index to assist in accounting for continued cold-air pooling and subsequent inverted SLRs in valleys that exist entirely above treeline. To verify existing assumptions regarding SLRs in this region, these transects need to be replicated and extended higher up each slope to test at what elevation, if any in the valley, there is a reversal of SLRs back to normal.

5.4. Implications of SBIs on permafrost distribution in the valleys

Ground depth temperature data observed at each of the valley bottom and slope locations indicate that permafrost is likely to present at each site excluding the SFS site (Table 6). According to the model developed by Bonnaventure et al. (2012), the SFS site is predicted to have the lowest permafrost probability (46%) (Fig. 1a). Permafrost probability is greatest across ridgetop locations (75%–100%) and the second highest in valley bottom locations (60%–80%). Mid-slope locations at the elevation of treeline predicted permafrost probability was lowest (40%–50%) where SLRs were assumed to change in sign and magnitude based on previous observations and modelling (Lewkowicz et al. 2012). Therefore, this model and conceptualization (Bonnaventure and Lewkowicz 2013) in SBI-dominated valleys explains permafrost findings at these point locations well. With only six data collection sites included in this study, there still exists the need for work to be done including ground-truthing and monitoring of permafrost temperatures at other slope locations within these valleys to further confirm this conceptualization.
Permafrost temperatures and probability is defined not only by SAT but also by other factors including snow depth (Garibaldi et al. 2021), substrate composition, and moisture level (Smith et al. 2010), slope steepness (Haeberli et al. 2010), aspect (Cote 2002; Bonnaventure and Lewkowicz 2008), and vegetation cover (Shur and Jorgenson 2007; Kropp et al. 2020). Therefore, SBIs can impact permafrost distribution; however, this impact must be viewed through the lens of additional factors also shaping permafrost distribution. For example, the NFS air temperature was significantly warmer than in the valley bottom; however, the ground temperature at depth was colder on the slopes than at the valley bottom location. This may be a result of the lack of solar radiation warming the ground on the NFS due to topographic shading. Additionally, deeper snow cover in the valley bottom location because of trees capturing snow could insulate the ground surface from wintertime SATs (Way and Lapalme 2021).
The interaction between air temperature and ground surface temperature can be described through n-factors. SBIs dominate throughout the cold months of the year and during the night hours. This indicates that the pattern of warmer conditions taking place at slope locations more often occurs during times when temperatures are freezing. This is especially important when analytically determining permafrost distribution through empirical models such as the TTOP model, which includes thawing and freezing degree-days (Riseborough et al. 2008). During the freezing season, permafrost temperatures are dominated by freezing n-factors that are related to snow depth reducing the interaction between SAT and TTOP (Garibaldi et al. 2021). With SBIs being most prevalent during the freezing degree-days of the year, freezing n-factors are of particular interest. Freezing n-factors close to a value of unity indicate that the air temperature strongly influences the ground temperature during freezing degree-days. The freezing n-factors in the NV ranged from 0.57 in the valley bottom to 0.87 on the WFS. In the SV, freezing n-factors were between 0.19 and 0.39, indicating a much less direct interaction between air temperature and ground surface temperatures on freezing degree-days. This indicates that while SBIs are less prevalent in the NV, they may influence permafrost temperatures more directly due to more interaction between air and ground temperatures than in the SV. Overall, further research confirming the influence of SBIs on permafrost thermal state and distribution in the valleys is needed as the interconnectivity between air temperature patterns and ground surface/permafrost temperatures is far from linear across the landscape.

6. Summary and conclusions

Fine-scale spatial and temporal variation in SBI characteristics in near-proximity dissimilar high-latitude valleys were reviewed in this study. Several key findings include:
Strongly inverted annual average SLRs led to air temperatures at each of the higher elevation sites that were significantly warmer than the valley bottom locations (up to 1.2 °C year−1), particularly during the winter season.
Ground surface temperatures and ground depth sensors were warmer at all higher elevation sites than the valley bottom locations in their respective valleys, except for the NFS location. Ground depth temperature on average was above 0 °C across the 2-year study period at the SFS location. This indicates that permafrost is likely absent at this site while being present in the valley bottom. Thus, permafrost distribution is likely influenced by SBIs supporting the results modelled by Bonnaventure et al. (2012) and conceptualized in Bonnaventure and Lewkowicz (2013).
SBIs in the region were found to be much more frequent and stronger than anticipated from previous research. The resulting annual average SLRs were inverted on each elevational transect (0.46–1.21 °C 100 m−1) and were much more strongly inverted in the SV, particularly on the SFT.
Observed SLRs using ETA were much more intensely inverted than those predicted in modelling created by Lewkowicz et al. (2012), which predicted average annual SLRs that were inverted to a maximum of 0.1°C 100 m−1.
The same previous modelling (Lewkowicz et al. 2012) predicted and assume that SLRs reverse back to normal above treeline. Our findings did observe warming towards treeline but also showed that in the treeless valley SLRs are also inverted in the first 100 m upslope from the valley bottom. The assumptions made by Lewkowicz et al. (2012) must be reviewed to determine when and if there is an inflection point of SLRs (positive SLRs to negative SLRs) in both valleys. Thus, transects need to be extended up higher in each valley to review these assumptions.
SLRs calculated for each transect using the climate reanalysis downscaling programs of GlobSim and ClimateNA were close to being neutral (0.0 °C 100 m−1). This indicates that these programs do not capture the hyper-inversions that occur frequently in the lowest parts of the valley. Thus, ETA is an important tool to quantify the influence SBIs have on annual average SLRs in these remote valley locations.
Overall, this research shows there is a significant impact of SBIs on SATs and the potential for impact on permafrost temperatures in complex mountainous high-latitude terrain. As hypothesized, the SBI characteristics and subsequent SLRs vary significantly between two proximal dissimilar valleys. SBIs were persistent and strong in both valleys, resulting in inverted SLRs on an annual average. That said, the SLRs were significantly less inverted on an annual average in the NV. Strong and persistent SBI resulted in the coldest locations being in the valley bottom rather than 100 m up each slope. While much has been found regarding the valley-to-valley scale variability of SBIs using the in situ elevational transect data in these valleys, continued expansion of this research is needed as many questions arose and remain unanswered. Mechanisms of variation of SBIs between the valleys both spatially and temporally need to be reviewed to explain the patterns observed. As this research focused mostly on air temperature, an immense opportunity exists to extend this research to examine how ground temperature and subsequently permafrost is impacted by strong SBIs in these valleys.

Acknowledgements

We would like to acknowledge that this research was conducted on the traditional territory of the Trʼondëk Hwëchʼin First Nation and thank them for allowing us to set up our sensors and for being supportive of this research. We appreciate the Yukon Territorial Government for giving us permission for fieldwork in the territory (Research Licence # 6800–20-1143). We acknowledge all the field assistants that helped set up the sensors in the fields and collect the subsequent data including R. Gibson, K. Bexte, M. Garibaldi, A.G. Lewkowicz, O. Kienzle, S. Vegter, and A. Musk. Thanks to my labmates in the Bonnaventure Lab for Permafrost Science for giving their insights on how to improve the research and prepare it for submission. In particular thanks to M. Garibaldi, L. Chasmer, and H. Jiskoot for reviewing the manuscript leading up to submission. We thank the University of Lethbridge for providing the necessary programs, support, and workspaces to complete the research behind this manuscript. Furthermore, we thank NSERC and the Northern Student Training Program for funding the fieldwork portion of this research.

Author contribution statement

NCN—Completed the data analysis as part of MSc requirements, wrote the first draft of the manuscript, and completed revisions as suggested by the peer reviewers.
PPB—Formulated the idea behind the manuscript, conducted several thorough reviews, and assisted in rewriting sections to prepare the manuscript for submission and resubmission.

References

Abdi H., Williams L. 2010. Newman-Keuls test and Tukey test. Encyclopedia of Research Design, 2: 897–902.
Akperov M.G., Mokhov I.I., Dembitskaya M.A., Parfenova M.R., Rinke A. 2019. Lapse rate peculiarities in the Arctic from reanalysis data and model simulations. Russian Meteorology and Hydrology, 44(2): 97–102.
Bonnaventure P.P., Lewkowicz A.G. 2008. Mountain permafrost probability mapping using the BTS method in two climatically dissimilar locations, northwest Canada. Canadian Journal of Earth Sciences, 45(4): 443–455.
Bonnaventure P.P., Lewkowicz A.G., Kremer M., Sawada M.C. 2012. A permafrost probability model for the southern Yukon and northern British Columbia, Canada. Permafrost and Periglacial Processes, 23(1): 52–68.
Bonnaventure P.P., Lewkowicz A.G. 2013. Impacts of mean annual air temperature change on a regional permafrost probability model for the southern Yukon and northern British Columbia, Canada. The Cryosphere, 7(3): 935–946.
Bourne S.M., Bhatt U.S., Zhang J., Thoman R. 2010. Surface-based temperature inversions in Alaska from a climate perspective. Atmospheric Research, 95(2): 353–366.
Bradley R.S., Keimig F.T., Diaz H.F. 1992. Climatology of surface-based inversions in the North American Arctic. Journal of Geophysical Research: Atmospheres, 97(D14): 15699–15712.
Burn C., Moore J., O'Neill B., Hayley D, Trimble J, Calmels F., 2015. Permafrost characterization of the Dempster Highway, Yukon and Northwest Territories. In: 7th Canadian Permafrost Conference. Quebec City, Canada.
Cao B., Quan X., Brown N., Stewart-Jones E., Gruber S. 2019. GlobSim (v1.0): deriving meteorological time series for point locations from multiple global reanalyses. Geoscientific Model Development, 12(11): 4661–4679.
Caserini S., Giani P., Cacciamani C., Ozgen S., Lonati G. 2017. Influence of climate change on the frequency of daytime temperature inversions and stagnation events in the Po valley: historical trend and future projections. Atmospheric Research, 184: 15–23.
Chang L., Feng G., Zhang Y., He X. 2018. Effect of cloud fraction on Arctic low-level temperature inversions in AIRS observations over both land and ocean. IEEE Transactions on Geoscience and Remote Sensing, 56(4): 2025–2032.
Compo G.P., Whitaker J.S., Sardeshmukh P.D., Matsui N., Allan R.J., Yin X., 2011. The twentieth century reanalysis project. Quarterly Journal of the Royal Meteorological Society, 137(654): 1–28.
Copernicus Climate Change Service (C3S). 2017: ERA5: Fifth generation of ECMWF atmospheric reanalyses of the global climate. Copernicus Climate Change Service Climate Data Store (CDS). Available from https://cds.climate.copernicus.eu/cdsapp#!/home [accessed 21 December 2020].
Cote M.M. 2002. The influence of elevation and aspect on permafrost distribution in central Yukon Territory. M.Sc. thesis, Department of Geography and Environmental Studies, Carlton University, Ottawa, ON.
Daly C., Conklin D.R., Unsworth M.H. 2010. Local atmospheric decoupling in complex topography alters climate change impacts. International Journal of Climatology, 30(12): 1857–1864.
Devasthale A., Willén U., Karlsson K.G., Jones C.G. 2010. Quantifying the clear-sky temperature inversion frequency and strength over the Arctic Ocean during summer and winter seasons from AIRS profiles. Atmospheric Chemistry and Physics, 10(12): 5565–5572.
Drobinski P., Dabas A.M., Haeberli C., Flamant P.H. 2003. Statistical characterization of the flow structure in the Rhine Valley. Boundary-Layer Meteorology, 106(3): 483–505.
Duk-Rodkin A.,1999 Glacial limits map of Yukon. Indian & Northern Affairs Canada/Department of Indian & Northern Development: Exploration & Geological Services Division, Geoscience Map 1999-2. Available from https://data.geology.gov.yk.ca/Reference/42050#InfoTab [accessed 23 April 2022].
Environment Canada. 2021a. Canadian climate normals (1981-2010). Canadain Centre for Climate Services. Available from https://climate.weather.gc.ca/climate_normals/index_e.html [accessed 10 September 2021].
Environment Canada. 2021b. Historical data. Canadian Centre for Climate Services. Available from https://climate.weather.gc.ca/historical_data/search_historic_data_e.html [accessed 10 September 2021].
Esri. 2022a. “Light grey canvass base” [basemap]. 1:14, 100, 000. “World Light Grey Reference”. Available from https://www.arcgis.com/home/item.html?id=291da5eab3a0412593b66d384379f89f [accessed 20 April 2022].
Esri. 2022b. “Imagery” [basemap]. 1:380,000. “World Imagery Map”. Available from https://www.arcgis.com/home/item.html?id=10df2279f9684e4a9f6a7f08febac2a9 [accessed 20 April 2022].
Etzelmüller B. 2013. Recent advances in mountain permafrost research. Permafrost and Periglacial Processes, 24(2): 99–107.
Frei C. 2014. Interpolation of temperature in a mountainous region using nonlinear profiles and non-Euclidean distances. International Journal of Climatology, 34(5): 1585–1605.
French H.M., Williams P. 2017. The periglacial environment. 4th ed. Wiley-Blackwell, Oxford, UK. 544p.
Gardner A.S., Sharp M.J.,  Koerner R.M., Labine C., Boon S., Marshall S.J., 2009. Near-surface temperature lapse rates over Arctic glaciers and their implications for temperature downscaling. Journal of Climate, 22(16): 4281–4298.
Garibaldi M.C., Bonnaventure P.P., Lamoureux S.F. 2021. Utilizing the TTOP model to understand spatial permafrost temperature variability in a high Arctic landscape, Cape Bounty, Nunavut, Canada. Permafrost and Periglacial Processes, 32(1): 19–34.
Global Modeling and Assimilation Office (GMAO). 2015. MERRA-2 inst3_3d_asm_Np: 3d,3-Hourly, Instantaneous, Pressure-Level, Assimilation, Assimilated Meteorological Fields. Greenbelt, MD, USA. Goddard Earth Sciences Data and Information Services Center (GES DISC).
Graham R.M., Hudson S.R., Maturilli M. 2019. Improved performance of ERA5 in arctic gateway relative to four global atmospheric reanalyses. Geophysical Research Letters, 46(11): 6138–6147.
Gustavsson T., Karlsson M., Bogren J., Lindqvist S. 1998. Development of temperature patterns during clear nights. Journal of Applied Meteorology (1988-2005), 37(6): 559–571.
Haeberli W., Noetzli J., Arenson L., Delaloye R., Gärtner-Roer I., Gruber S., 2010. Mountain permafrost: development and challenges of a young research field. Journal of Glaciology, 56(200): 1043–1058.
Haeberli W. 2013. Mountain permafrost—research frontiers and a special long-term challenge. Cold Regions Science and Technology, 96: 71–76.
Heginbottom J., Dubreuil M., Haker P. 1995. Canada permafrost [map]. National Atlas of Canada [accessed 21 August 2020].
Hjort J., Karjalainen O., Aalto J., Westermann S., Romanovsky V.E., Nelson F.E., 2018. Degrading permafrost puts Arctic infrastructure at risk by mid-century. Nature Communications, 9(1): 5147–5147.
Hiebl J., Schöner W. 2018. Temperature inversions in Austria in a warming climate – changes in space and time. Meteorologische Zeitschrift, 27(4): 309–323.
Hou P., Wu S. 2016. Long-term changes in extreme air pollution meteorology and the implications for air quality. Scientific Reports, 6(1): 23792–23792.
Japan Meteorological Agency/Japan. 2013 updated monthly. JRA-55: Japanese 55-year Reanalysis, Daily 3-Hourly and 6-Hourly Data. Research Data Archive at the National Center for Atmospheric Research, Computational and Information Systems Laboratory [accessed 23 December 2020].
Johannessen O.M., Kuzmina S.I., Bobylev L.P., Miles M.W. 2016. Surface air temperature variability and trends in the arctic: new amplification assessment and regionalisation. Tellus A: Dynamic Meteorology and Oceanography. 68(1): 28234–28212.
Kropp H., Loranty M.M., Natali S.M., Kholodov A.L., Rocha A.V., Myers-Smith I., 2020. Shallow soils are warmer under trees and tall shrubs across Arctic and boreal ecosystems. Environmental Research Letters, 16(1): 1501–1514.
Lareau N.P., Crosman E., Whiteman C.D., Horel J.D., Hoch S.W., Brown W.O.J., Horst T.W. 2013. The persistent cold-air pool study. Bulletin of the American Meteorological Society, 94(1): 51–63.
Lewkowicz A.G., Bonnaventure P.P. 2011. Equivalent elevation: a new method to incorporate variable surface lapse rates into mountain permafrost modelling. Permafrost and Periglacial Processes, 22(2): 153–162.
Lewkowicz A.G., Bonnaventure P.B., Smith S.L., Kuntz Z. 2012. Spatial and thermal characteristics of mountain permafrost, northwest Canada. Geografiska Annaler: Series A, Physical Geography, 94(2): 195–213.
Liu Y., Key J.R. 2003. Detection and analysis of clear-sky, low-level atmospheric temperature inversions with MODIS. Journal of Atmospheric and Oceanic Technology, 20(12): 1727–1737.
Liu Y., Key J.R., Schweiger A., Francis J. 2006. Characteristics of satellite-derived clear-sky atmospheric temperature inversion strength in the Arctic, 1980–1996. Journal of Climate, 19(19): 4902–4913.
Lundquist J.D., Pepin N., Rochford C. 2008. Automated algorithm for mapping regions of cold-air pooling in complex terrain. Journal of Geophysical Research–Atmospheres, 113(D22): D22107–D22n/a.
Luo H., Ge F., Yang K., Zhu S., Peng T., Cai W., 2019. Assessment of ECMWF reanalysis data in complex terrain: Can the CERA-20C and ERA-Interim data sets replicate the variation in surface air temperatures over Sichuan, China? International Journal of Climatology, 39(15): 5619–5634.
Martineau P., Wright J.S., Zhu N., Fujiwara M. 2018. Zonal-mean data set of global atmospheric reanalyses on pressure levels. Earth System Science Data, 10(4): 1925–1941.
Mayfield J.A., Fochesatto G.J. 2013. The layered structure of the winter atmospheric boundary layer in the interior of Alaska. Journal of Applied Meteorology and Climatology, 52(4): 953–973.
McCaffrey K., Wilczak J.M., Bianco L., Grimit E., Sharp J., Banta R., 2019. Identification and characterization of persistent cold pool events from temperature and wind profilers in the Columbia River basin. Journal of Applied Meteorology and Climatology, 58(12): 2533–2551.
Mernild S.H., Liston G.E. 2010. The influence of air temperature inversions on snowmelt and glacier mass balance simulations, Ammassalik Island, southeast Greenland. Journal of Applied Meteorology and Climatology, 49(1): 47–67.
Minder J.R., Mote P.W., Lundquist J.D. 2010. Surface temperature lapse rates over complex terrain: lessons from the Cascade mountains. Journal of Geophysical Research, 115(D14): D14122.
Morse P.D., Wolfe S.A. 2015. Geological and meteorological controls on icing (aufeis) dynamics (1985 to 2014) in subarctic Canada. Journal of Geophysical Research. Earth Surface, 120(9): 1670–1686.
Natural Resources Canada 2022. Canadian digital elevation model. 1:50,000. Canada. Available from https://maps.canada.ca/czs/index-en.html [accessed 15 April 2022]
NOAA 2021a. NOAA solar calculator. Earth System Research Laboratories. Available from https://www.esrl.noaa.gov/gmd/grad/solcalc/ [accessed 8 April 2021].
NOAA 2021b. Station pressure calculator. Available from https://www.weather.gov/epz/wxcalc_stationpressure [accessed 12 May 2021].
O'Neill H.B., Wolfe S.A., Duchesne C. 2019. New ground ice maps for Canada using a paleogeographic modelling approach. The Cryosphere, 13(3): 753–773.
Obu J., Westermann S., Bartsch A., Berdnikov N., Christiansen H.H., Dashtseren A., 2019. Northern hemisphere permafrost map based on TTOP modelling for 2000–2016 at 1 km2 scale. Earth-Science Reviews, 193: 299–316.
Olefeldt D., Goswami S., Grosse G., Hayes D., Hugelius G., Kuhry P., 2016. Circumpolar distribution and carbon storage of thermokarst landscapes. Nature Communications, 7(1): 13043.
Palarz A., Celiński-Mysław D., Ustrnul Z. 2018. Temporal and spatial variability of surface-based inversions over Europe based on ERA-Interim reanalysis. International Journal of Climatology, 38(1): 158–168.
Palazzi E., Mortarini L., Terzago S., von Hardenberg J. 2019. Elevation-dependent warming in global climate model simulations at high spatial resolution. Climate Dynamics, 52(5): 2685–2702.
Pepin N.C., Daly C., Lundquist J. 2011. The influence of surface versus free-air decoupling on temperature trend patterns in the western United States. Journal of Geophysical Research: Atmospheres, 116(D10): n/a–n/a.
Pepin N., Bradley R.S., Diaz H.F., Baraer M., Caceres E.B., Forsythe N., 2015. Elevation-dependent warming in mountain regions of the world. Nature Climate Change, 5(5): 424–430.
Pichugina Y.L., Banta R.M., Bonin T., Brewer W.A., Choukulkar A., McCarty B.J., 2019. Spatial variability of winds and HRRR–NCEP model error statistics at three Doppler-Lidar sites in the wind-energy generation region of the Columbia River basin. Journal of Applied Meteorology and Climatology, 58(8): 1633–1656.
Pike G., Pepin N.C., Schaefer M. 2013. High latitude local scale temperature complexity: the example of Kevo Valley, Finnish Lapland. International Journal of Climatology, 33(8): 2050–2067.
Porter T.J., Schoenemann S.W., Davies L.J., Steig E.J., Bandara S., Froese D.G. 2019. Recent summer warming in northwestern Canada exceeds the Holocene thermal maximum. Nature Communications, 10(1): 1631–1610.
Qin Y., Zhang P., Liu W., Guo Z., Xue S. 2020. The application of elevation corrected MERRA2 reanalysis ground surface temperature in a permafrost model on the Qinghai-Tibet plateau. Cold Regions Science and Technology, 175: 103067.
Quinton W.L., Hayashi M., Chasmer L.E. 2011. Permafrost-thaw-induced land-cover change in the Canadian Subarctic: implications for water resources. Hydrological Processes, 25(1): 152–158.
Riseborough D., Shiklomanov N., Etzelmüller B., Gruber S., Marchenko S. 2008. Recent advances in permafrost modelling. Permafrost and Periglacial Processes, 19(2): 137–156.
Schaefer K., Elshorbany Y., Jafarov E., Schuster P.F., Striegl R.G., Wickland K.P., Sunderland E.M. 2020. Potential impacts of mercury released from thawing permafrost. Nature Communications, 11(1): 4650–4650.
Schuster P.F., Schaefer K.M., Aiken G.R., Antweiler R.C., Dewild J.F., Gryziec J.D., 2018. Permafrost stores a globally significant amount of mercury. Geophysical Research Letters, 45(3): 1463–1471.
Schuur E.A.G., Mack M.C. 2018. Ecological response to permafrost thaw and consequences for local and global ecosystem services. Annual Review of Ecology, Evolution, and Systematics, 49(1): 279–301.
Scudder G.G.E. 1997. Environment of the Yukon. Available from https://www.researchgate.net/publication/252860187_Environment_of_the_Yukon [accessed 23 April 2022].
Seidel D.J., Ao C.O., Li K. 2010. Estimating climatological planetary boundary layer heights from radiosonde observations: comparison of methods and uncertainty analysis. Journal of Geophysical Research: Atmospheres, 115: D16113.
Serreze M.C., Barry R.G. 2011. Processes and impacts of Arctic amplification: a research synthesis. Global and Planetary Change, 77(1): 85–96.
Shahi S., Abermann J., Heinrich G., Prinz R., Schöner W. 2020. Regional variability and trends of temperature inversions in Greenland. Journal of Climate, 33: 9391–9407.
Shur Y.L., Jorgenson M.T. 2007. Patterns of permafrost formation and degradation in relation to climate and ecosystems. Permafrost and Periglacial Processes, 18(1): 7–19.
Smith S.L., Romanovsky V., Lewkowicz A.G., Burn C.R., Allard M., Clow G.D., 2010. Thermal state of permafrost in North America: a contribution to the International Polar Year. Permafrost and Periglacial Processes, 21(2): 117–135.
Smith S.L., Bonnaventure P.P. 2017. Quantifying surface temperature inversions and their impact on the ground thermal regime at a high Arctic site. Arctic, Antarctic, and Alpine Research, 49(1): 173–185.
Sun X., Holmes H.A. 2019. Surface turbulent fluxes during persistent cold-air pool events in the Salt Lake Valley, Utah. Part I: observations. Journal of Applied Meteorology and Climatology, 58(12): 2553–2568.
Tao J., Koster R.D., Reichle R.H., Forman B.A., Xue Y., Chen R.H., Moghaddam M. 2019. Permafrost variability over the Northern Hemisphere based on the MERRA-2 reanalysis. The Cryosphere, 13(8): 2087–2110.
Turner M.G., O'Neill R.V., Gardner R.H., Milne B.T. 1989. Effects of changing spatial scale on the analysis of landscape pattern. Landscape Ecology, 3(3-4): 153–162.
Urban M., Eberle J., Hüttich C., Schmullius C., Herold M. 2013. Comparison of satellite-derived land surface temperature and air temperature from meteorological stations on the pan-Arctic scale. Remote Sensing, 5(5): 2348–2367.
Vonk J.E., Tank S.E., Bowden W.B., Laurion I., Vincent W.F., Aleksychik P., 2015. Reviews and syntheses: effects of permafrost thaw on Arctic aquatic ecosystems. Biogeosciences, 12(23): 7129–7167.
Wahl H. 2004. Yukon overview: climate. In: Ecoregions of the Yukon Territory – Biophysical properties of Yukon landscapes. Edited by C.A.S. Smith, J.C. Meikle, C.F. Roots. Agriculture and Agri-Food Canada, PARC Technical Bulletin 04-01, Part I. PARC, Summerland, BC. pp. 19–23. Available from https://yukon.ca/sites/yukon.ca/files/env/env-ecoregions-yukon-territory.pdf [accessed 6 May 2021].
Wang B., Paudel B., Li H. 2009. Retrogression characteristics of landslides in fine-grained permafrost soils, Mackenzie valley, Canada. Landslides, 6: 121–127.
Wang Q., Fan X., Wang M. 2013. Recent warming amplification over high elevation regions across the globe. Climate Dynamics, 43(1-2): 87–101.
Wang T., Hamann A., Spittlehouse D., Carroll C. 2016. Locally downscaled and spatially customizable climate data for historical and future periods for North America. PloS One, 11(6): e0156720–e0156720.
Ward Jones M. K., Pollard W.H., Jones B.M. 2019. Rapid initialization of retrogressive thaw slumps in the Canadian high Arctic and their response to climate and terrain factors. Environmental Research Letters, 14(5): 055006.
Way R.G., Lapalme C.M. 2021. Does tall vegetation warm or cool the ground surface? Constraining the ground thermal impacts of upright vegetation in northern environments. Environmental Research Letters, 16(5): 54077.
Whiteman C.D. 1982. Breakup of temperature inversions in deep mountain valleys: part I. Observations. Journal of Applied Meteorology (1962-1982), 21(3): 270–289.
Whiteman C., Richland 2000. Mountain meteorology: Fundamentals and applications. Oxford University Press, New York. 376p.
Whiteman C.D., Zhong S., Shaw W.J., Hubbe J.M., Bian X., Mittelstadt J. 2001. Cold pools in the Columbia basin. Weather and Forecasting, 16(4): 432–447.
Williams R., Thorp T. 2015. Characteristics of springtime nocturnal temperature inversions in a high latitude environment. Weather, 70(S1): S37–S43.
Williamson S.N., Zdanowicz C., Anslow F.S., Clarke G.K.C., Copland L., Danby R.K., 2020. Evidence for elevation-dependent warming in the St. Elias Mountains, Yukon, Canada. Journal of Climate, 33(8): 3253–3269.
Wolyn P.G., McKee T.B. 1989. Deep stable layers in the intermountain western United States. Monthly Weather Review, 117(3): 461–472.
Yu L., Zhong S., Bian X. 2017. Multi-day valley cold-air pools in the western United States as derived from NARR. International Journal of Climatology, 37(5): 2466–2476.
Survey Yukon Geological, 2022. Yukon exploration and geology 2021. Edited by K.E. MacFarlane. Yukon Geological Survey, Whitehorse, Yukon. Available from. https://data.geology.gov.yk.ca/Reference/95935#InfoTab [accessed 23 April 2022].
Yukon Government 2022. Yukon Road Network. Polyline Feautre Class.  Available from https://mapservices.gov.yk.ca/arcgis/rest/services/GeoYukon/GY_Transportation/MapServer/60 [accessed 20 April 2022].
Zhang Y., Seidel D.J., Golaz J.-C., Deser C., Tomas R.A. 2011. Climatological characteristics of Arctic and Antarctic surface-based inversions. Journal of Climate, 24(19): 5167–5186.
Zhang X., Flato G., Kirchmeier-Young M., Vincent L., Wan H., Wang X., 2019. Changes in temperature and precipitation across Canada. In Canada's changing climate report Edited by E. Bush, D.S. Lemmen. Government of Canada, Ottawa, ON. pp.112–193. Available from https://changingclimate.ca/CCCR2019/chapter/4-0/ [accessed 22 April 2021].
Zangl G. 2003. The impact of upstream blocking, drainage flow and the geostrophic pressure gradient on the persistence of cold-air pools. Quarterly Journal of the Royal Meteorological Society, 129(587): 117–137.
Zhong S., Whiteman C.D., Bian X., Shaw W.J., Hubbe J.M. 2001. Meteorological processes affecting the evolution of a wintertime cold air pool in the Columbia Basin. Monthly Weather Review, 129(10): 2600–2613.

Information & Authors

Information

Published In

cover image Arctic Science
Arctic Science
e-First

History

Published online: 2 June 2022

Data Availability Statement

The in situ SLR data collected using ETA is available upon request from the corresponding author (nick.noad@uleth.ca). Second-party data were used for the climate reanalysis data sets. Information on accessing Globsim can be found at https://globsim.readthedocs.io/en/latest/. Information on accessing ClimateNA data sets can be found at http://climatena.ca/.

Key Words

  1. surface-based inversions
  2. surface air temperature
  3. permafrost distribution
  4. dissimilar valleys

Mots-clés

  1. inversions en surface
  2. température de l’air en surface
  3. distribution du pergélisol
  4. vallées dissemblables

Authors

Affiliations

Department of Geography and Environment, University of Lethbridge, Lethbridge, AB T1K 3M4, Canada
Philip P. Bonnaventure
Department of Geography and Environment, University of Lethbridge, Lethbridge, AB T1K 3M4, Canada

Competing Interests

The authors declare there are no competing interests

Funding Information

NSERC Discovery Grant Program (1506); NSERC Canadian Graduate Scholarship Master's Program (N/A); Northern Scientific Training Program (N/A).

Metrics & Citations

Metrics

Other Metrics

Citations

Cite As

Export Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

There are no citations for this item

View Options

View options

PDF

View PDF

Get Access

Media

Media

Other

Tables

Share Options

Share

Share the article link

Share with email

Email a colleague

Share on social media