Insights on seasonal solifluction processes in warm permafrost Arctic landscape using a dense monitoring approach across adjacent hillslopes

Solifluction processes in the Arctic are highly complex, introducing uncertainties in estimating current and future soil carbon storage and fluxes, and assessment of hillslope and infrastructure stability. This study aims to enhance our understanding of triggers and drivers of soil movement of permafrost-affected hillslopes in the Arctic. To achieve this, we established an extensive soil deformation and temperature sensor network, covering 48 locations across multiple hillslopes within a 1 km2 watershed on the Seward Peninsula, AK. We report depth-resolved measurements down to 1.8 m depth for May to September 2022, a period conducive to soil movement due to deepening thaw layers and frequent rain events. Over this period, surface movements of up to 334 mm were recorded. In general, these movements occur close to the thawing front, and are initiated as thawing reaches depths of 0.4–0.75 m. The largest movements were observed at the top of the south-east facing slope, where soil temperatures are cold (mean annual soil temperatures averaging −1.13 °C) and slopes are steeper than 15°. Our analysis highlights three primary factors influencing movements: slope angle, soil thermal conditions, and thaw depth. The latter two significantly impact the generation of pore water pressures at the thaw–freeze interface. Specifically, soil thermal conditions govern the liquid water content, while thaw depth influences both the height of the water column and, consequently, the pressure at the thawing front. These factors affect soil properties, such as cohesion and internal friction angle, which are crucial determinants of slope stability. This underscores the significance of a precise understanding of subsurface thermal conditions, including spatial and temporal variability in soil temperature and thaw depth, when assessing and predicting slope instabilities. Based on our observations, we developed a factor of safety proxy that consistently falls below the triggering threshold for all probes exhibiting displacements exceeding 50 mm. This study offers novel insights into patterns and triggers of hillslope movements in the Arctic and provides a venue to evaluate their impact on soil redistribution.


Introduction
) estimated an approximately 50% uncertainty in calculating the net ecosystem carbon balance when accounting for soil movements, emphasizing the intricate nature of understanding the impact of these processes on carbon emissions.
Solifluction processes in the Arctic encompass slow downslope movements of soil masses due to freeze-thaw processes (Harris et al 2011).These processes include frost creep, gelifluction, or plug-like flow (Washburn 1980, Matsuoka 2001).Solifluction is commonly observed in sandy to silty soils having low liquid and plasticity indices (Matsuoka 2001, Harris et al 2011), and is often observed during late spring and summer when thaw saturates the soil (Rowley et al 2015).Solifluction rates are controlled primarily by slope angle, thaw depth, and water content (Hjort et al 2014).It can occur on slopes with angles as low as 1 • , although it is more prevalent on slope angles ranging from 5 • and 20 •  (Rowley et al 2015).Two-sided freezing, which creates ice lenses near the base of the active layer that acts as a shear zone upon thawing, has been shown to enhance the likelihood of plug-like flow during thawing (Lewkowicz and Clarke 1998).
In non-Arctic environments, extreme precipitation events are commonly recognized as a triggering factor of slope instabilities.However, in Arctic environments, the influence of rainfall is comparatively small (Del Vecchio et al 2023), and the ice content of the soil emerges as one of the primary triggering factors for slope instabilities (Mithan et al 2021).Thaw consolidation of ice-rich fine-grained soils lead to raised pore water pressures (e.g.Morgenstern and Nixon 1971), where the excess pore-water saturates the weakened thaw layer and reduces the shear strength by lowering the soil's cohesion and internal friction angle (Harris andLewkowicz 2000, Harris et al 2011).Testing of soil-ice specimens showed decreasing shear strength with increasing temperature, with the most significant reduction occurring from −2 to 0 • C (Huang et al 2022), which correlates with the temperature range where the strongest change in ice content is often observed (Uhlemann et al 2021, Huang et al 2022), leading to an increase in unfrozen water content.
Despite recent improvements in the understanding of solifluction processes, field studies investigating the heterogeneity of these processes are still rare, leading to gaps in our understanding of their patterns, the volumes mobilized at small and large scales, and their triggering factors.Particular questions that we aim to answer here are: what are the features (timing, intensity, volumes, and location) and patterns exhibited by solifluction processes throughout the soil thawing season across adjacent hillslopes?What mechanisms act as triggers and what primary factors exert control over these phenomena?
We address these questions by using a novel dense monitoring approach to obtain vertically resolved, continuous observations of soil movement and temperature at tens of locations across multiple adjacent hillslopes throughout the thawing season.This spatially and temporally high-resolution data set enabled us to assess the heterogeneities in soil movement along different slope and permafrost conditions on a watershed scale, and evaluate triggering factors.

Site description/material
The study site is situated along the Nome-Teller highway, approximately 65 km northwest of Nome (Mile marker 47), on the western Seward Peninsula in Alaska (figure 1).The approximately 1 km 2 study area includes south, south-east and east facing slopes, separated by two main streams merging towards the bottom of the watershed (figures 1(a) and (c)).The mean slope angle is 13 • , ranging from 0 to 42 • .The site is predominantly covered by tussock tundra, dwarf shrubs and grasses with some patches of tall shrubs, especially near the streams (Del Vecchio et al 2023, figure 1(c)).Bedrock is outcropping at the highest elevations of the south facing slope (figure 1(a)).The geology comprises quaternary deposits and the Precambrian Nome Group, with the presence of metabasalt and impure marble and calcschist (figure 1(a), Till et al 2011).These units are covered by a thin layer of organic material, approximately 0.3-0.5 m thick, overlying silty material of varying thickness (about 2 m in average).Over the past 30 years, the mean annual air temperature has been −6 • C, and has increased by 0.06 • C yr −1 (Thornton et al 2016, Jafarov et al 2018).During the same period, the average rainfall and snowfall were 400 mm yr −1 and 300 mm yr −1 , respectively (Jafarov et al 2018).
The study area is located in a region characterized by discontinuous permafrost.The electrical resistivity tomography (ERT) profiles (figure 1(b), Uhlemann et al 2023) illustrate the main subsurface features, including the presence of permafrost, bedrock and taliks (perennially unfrozen layers).The high resistivity values in the high elevations of the south facing profiles indicate the presence of outcropping bedrock that is partially frozen, and permafrost bodies at the lower elevations, indicating permafrost is discontinuous in the watershed.In the remaining profiles, the high resistivity values highlight the presence of frozen bodies separated by taliks (figure 1(b)).

Monitoring
Movement and temperature monitoring were performed using an in-house developed sensor array Data collection spanned from 15 September 2021 to 15 September 2022 at 30 min interval.The temperature sensors factory-assured accuracy of 0.1 • C has been verified through lab experiments (Dafflon et al 2022).To further increase the accuracy of the temperature data, we applied a constant temperature correction for each sensor by extracting and removing the zero-curtain temperature value, observed during the freezing process during Fall 2022.The movement analysis was focused on the thawing season, specifically from 15 May to 15 September 2022.Before that period pressure build-up at the sensors caused artificially high readings that had to be removed.Among the 48 probes installed, 19 were not anchored into a frozen soil at the end of the monitoring period, however permafrost could still be present deeper than the length of the probes.
The roll, pitch, and yaw angles of each accelerometer are extracted from the accelerometric measurements, allowing the determination of the total angle, which we define as the angle between the vector defined by sensor reading and the gravitational field pointing downwards, without considering the orientation.The movement is calculated from the total angle considering the 10 cm interval between each accelerometer, achieving an accuracy of ±0.73 mm m −1 (Wielandt et al 2022).The movement extracted at each sensor corresponds to the temporal cumulative movement from the bottom of the probe to the sensor, considering the deepest sensor as the reference point (0 cm of displacement).The daily velocity at each sensor is determined from the daily gradient of these calculated displacements.To distinctly capture the variations in deformation both over time and along the probe's spatial axis, we also compute the daily gradient of the total angle measured by each sensor throughout the monitoring period.Fluxes are computed using the cumulative displacement (from bottom to the surface) of all sensors for the entire monitoring period.
Additional datasets used to investigate the mechanisms and controlling factors of soil movements include topographic data, weather forcing, and soil moisture measurements.A digital elevation model (DEM) with a spatial resolution of 1 m was derived from LiDAR data acquired in August 2021 by the National Center for Airborne Laser Mapping (Singhania 2020).Rainfall data was obtained from a meteorological station located 30 km south (Nome-Teller highway, mile marker 27, Busey et al 2017).Soil moisture content at each probe location was determined using a time domain reflectometry (TDR) meter (Fieldscout TDR150 Soil Moisture Meter) by averaging three measurements performed at 15 cm depth in mid-June 2022.The factory-assured accuracy for volumetric water content is ±3.0%, accompanied by a resolution of 0.1%.The location of each probe was measured using a real-time kinematic global positioning system (RTK differential GPS) with an estimated accuracy of ±2 cm on 20 June 2022 and 14 September 2022.

Slope stability analysis
Slope stability refers to the resistance of inclined surfaces against sliding due to external forces.Its assessment involves calculating the balance between the resisting forces that hold the slope in place, i.e. the shear strength, and the driving forces that attempt to make it move.The Mohr-Coulomb failure criterion (Jaeger With γ the unit weight (kN m −3 ), h the thaw depth (m) and β the slope angle ( • ) (figure 2).
In permafrost environments, McRoberts and Morgenstern (1974) calculated u following: With γ w the water unit weight (kN m −3 ), R the thaw-consolidation ratio, ds the soil particle density (g cm −3 ) and w the water content (m 3 m −3 ), with γ ′ corresponding to the submerged unit weight.
In alpine permafrost environments, Nater et al (2008) calculated c and φ ′ following: with w i the ice content (m 3 m −3 ), T the soil temperature ( o C), c u the cohesion at a reference temperature (Pa), and φ ′ u the internal friction angle for dry soil ( o ).Nater et al (2008) estimated the temperaturedependence (<0 o C) of the ice content w i with: where a and b are empirical constants.
In this study, we assess slope failure susceptibility using equations ( 1)-( 6) to calculate F s and name it F s proxy.This F s proxy implies several simplifications and limitations associated with the limited datasets we have and the still limited understanding on the evolution of soil mechanical properties associated with ice-water phase change and thus on the representation of processes leading to soil instability (Fu et al 2021).In particular, our F s proxy is aimed at assessing the susceptibility to failure at annual scale by considering the annual average state of soil and notably the ice content and thaw depth.Thus, for each probe, we define T as the annual mean temperature at 1 m depth (MAT1), h as the maximum thaw depth observed at each location (on 15 September 2022), w based on the TDR measurements by assuming homogeneous conditions throughout the unfrozen layer and thawing season, and β using the slope angle derived from the DEM (at 10 m scale).The other parameters were chosen within the range reported in the literature for unfrozen silty soils (Hampton andWinters 1983, Bommer et al 2012) and from nearby soil tests (Lathrop et al 2021) (table 1).
A comprehensive sensitivity analysis of our Fs proxy was conducted by investigating the influence of the four main parameters (soil water content, MAT1, slope angle, maximum thaw depth) using the Sobol indices and SALib (Herman and Usher 2017, Iwanaga et al 2022).The Sobol method, also known as analysis of variance, characterizes the total variance of the model by summing up the variances of the individual inputs (Sobol 2001).This approach allowed us to assess the impact of each input on the model's outcome while excluding interactions with other parameters, focusing solely on the first-order indices.

Movements across the watershed
During the thawing season (May-September) in 2022, observed soil movements were large in places, with local depth-integrated fluxes reaching 0.23 m 2 yr −1 , but highly variable (figure 3).The largest soil movements were located high on the south-east facing slope, where MAT1 is lower (minimum −3.22 • C, average −1.13 • C) than further down the slope (maximum 0.76 • C, average −0.62 • C) where there was minimal soil movement.The temperature is the highest in south facing slope (average MAT1 = −0.71• C), and the most stable.Very small movements were observed at the bottom of this slope where permafrost is still present.The temperature is the lowest in east facing slope (average MAT1 = −1.50• C).Some soil movements were observed at the top of this slope, although rates were smaller than at the top of the south-east facing slope.No soil movements were observed at the bottom of this slope, which is characterized by small slope angles (<14 • ) and thin active layer (ERT, figure 1).

Depth-resolved sliding dynamics and their triggers
Here we explore the mechanisms driving soil movement in the watershed using observed rainfall, soil temperature, displacement, velocity, and total angle gradient (figure 4).We first focus on four probes located on the south-east (P1, P2, P3) and east facing slopes (P4), as depicted in figure 1(a), which are representative of the diverse range of behaviors and soil settings observed across the site.P1 and P2 are situated in the middle of the slope, where the presence of taliks is more pronounced, with resistivities below 3000 Ωm (figure 1(b)).P1 is embedded within a frozen body, while P2, located only 20 m away, is installed in a depression where a talik is present below the seasonally frozen layer.The movements at the surface measured for P1 and P2 were 70 mm and 24 mm, respectively, with fluxes of 0.06 m 2 yr −1 and 0.02 m 2 yr −1 , respectively.Upon reaching a thaw depth of 1.6 m in mid-July, P2 ceased to be anchored in the frozen layer (figure 4), displaying increased velocity after rain events on 18 July and 7 August.Movements at location P1 are triggered 4 days before the first rain event in June (see the gradient, figure 4(a)), and experience a clear acceleration following the rain event on 18 July.P3 and P4 are located at the top of the watershed, where colder soil conditions and a thinner active layer prevail (maximum thawing depth of 0.8 and 0.7 m, respectively).The surface displacements measured for P3 and P4 were 334 mm and 119 mm, respectively, with fluxes of 0.23 m 2 yr −1 and 0.06 m 2 yr −1 , respectively.
The gradient of the total angle indicates the location and the timing of movements.Its analysis reveals that movement occurs predominantly along the thawing front (figure 4).Across the watershed, soil movement begins after the thawing front has reached depths of 0.4-0.75m (figures 6 and 7).A shear zone is formed in locations with movement greater than 50 mm (figure 7(b)).In these cases, the shear surface is located at depths ranging between 0.4 and 1.1 m and deepens by an average of about 30 cm during the thawing season (figure 7(b)).Two-sided freezing is observed at locations with large soil movements (e.g.P1, P3, and P4; figure 5), as evidenced by the upward slope in the zero-degree isotherm at depth.This phenomenon, which generates upward freezing, facilitates the formation of ice lenses at the permafrost table, which can contribute to the development of a shear zone upon thawing (Matsuoka 2001).
The spatio-temporal variability in thawing (figure 6(b)) is a strong driver of heterogeneity in movements across the watershed.The east facing slope, characterized by colder permafrost, thaws more slowly (indicated by green to yellow colors at shallow depth in figure 6(b)) than the south facing slope.The south-east facing slope thaws both slowly and quickly at different locations.Soil movements are triggered late in the thaw season on the east facing slope where thawing is slow (August, figure 6(c)), and are slightly smaller than the movements triggered earlier on the south-east facing slope (June-July, figure 6(c)).The south facing slope does not exhibit significant movements even with a rapid thawing process.
Rain events (red dashed lines in figure 6(c)) do not seem to impact soil movement at the top of the watershed, where velocities are large and relatively constant.At some locations where soil movement is otherwise limited (e.g.P2), these rain events are associated with transient spikes, but not long-term increases, in velocity.Almost all of the 19 probes not anchored in the frozen layer exhibited very small displacements (16 mm on average), indicating that locations with a deep active layer or no permafrost are largely stable.However, GPS measurements reveal an additional 5-10 cm of soil movement at six of these probe locations.At these locations, and elsewhere across the watershed, shear zones may be formed deeper than can be sensed by the probes.

Controlling factors on slope stability
To assess the potential drivers of solifluction processes, we investigate the relationship between surface displacements and MAT1, slope angle, maximum thawing depth, and moisture content.We focus on 29 probes that are embedded in frozen soil at the end of the monitoring period (figure 8).We find that soil movements larger than 50 mm are only triggered when the slope angle is greater than 15.5 • , the moisture content is greater than 33%, MAT1 is below −0.94 • C, and the maximum thaw depth is between 0.65 and 1.3 m (figure 8).
To provide a quantitative measure of the impact of the various soil parameters on soil stability, we perform a sensitivity analysis of the Fs proxy (equation ( 1)).The slope angle has the largest  influence, followed by MAT1 and maximum thaw depth (figure 9(a)).We evaluated the impact of MAT1 and slope angle on Fs by calculating Fs using equation (1).Soil moisture was set at an average value of 40%.Subsequently, measurements taken at the 29 locations, where probes were anchored in frozen soil year-round, were integrated for comparison.Below 0 • C, both MAT1 and slope angle fluctuations jointly influence Fs (figure 9(b)).As MAT1 decreases, critical Fs values (<1) are attained at progressively smaller slope angles.Above 0 • C, the slope angle becomes the main controlling factor.Figure 9(b) also emphasizes that among the 29 locations considered, those with displacements exceeding 50 mm at the surface have an Fs below 1 (figure 9(b) and supplementary 1).

Discussion
A key finding of this study is that shallow movements within this watershed occur along well-defined shear zones that develop with the thickening of the thaw layer (figure 7(b)), characteristic of plug-like flow (Lewkowicz andClarke 1998, Del Vecchio et al 2023).We find that the initiation of soil movement is linked to the progression of the thaw front, aligning with the conclusion of Harris et al (2011), who demonstrated that a specific depth of thawing is required for solifluction movement to occur.Increasing water content near the thaw-freeze interface likely drives a reduction in cohesion and an increase of the internal friction angle (figure 10).Soil movement is triggered if  the cumulative pore pressure induced by the unfrozen water column (figure 2(h)) and the weight of the soil column surpass a critical threshold.In this watershed, we find that the shearing surface is 0.4-0.75m deep when slopes are larger than 15.5 • .These values align with the literature (Lewkowicz and Clarke 1998, Harris et al 2011, Bommer et al 2012) and notably with Harris and Lewkowicz (2000), who suggested that movements could be initiated on slopes exhibiting a shear surface at a depth of 0.6 m.
The maximum thawing depth and the MAT1 exert control over the observed fluxes.Fluxes exceeding 0.03 m 2 yr −1 occur either with MAT1 values of −2.4 • C and below coupled with a shallow shear surface (−0.9 m and shallower), or MAT1 values of −1.65 • C and above associated with a deep shear zone (base at −1.1 m and deeper; supplementary 2).
In addition, two-sided freezing, which is often associated with the plug-like flow solifluction (Matsuoka 2001), is observed across the watershed (figure 5).The formation of ice lenses at the top of permafrost creates a shear zone upon thawing that allows the entire layer to move along this shear zone as a cohesive unit.The depth and the composition of this shearing surface likely control the movement's pattern and extent.
Beyond these shallow processes, the possibility of deep slow creep within the permafrost (Dallimore et al 1996, Foriero et al 1998), with rates ranging from millimeters to centimeters per year, arouses interest in a multiyear analysis perspective, where cumulative movements may pose potential hazards.Looking at the observed fluxes and at the precision and capacity limitations of our instrumentation and GPS measurements, coupled with the fact that the studied period covers only a single thawing season, we are unable to distinguish the possible presence of such slow movements from the solifluction processes.
Our data show that permafrost table depth, soil temperature, and thaw front progression vary across the studied hillslopes, influencing the timing and locations of hillslope movements.Spatial soil temperature variability across the watershed is a reliable indicator of the spatio-temporal heterogeneity of these movements.The warmest, south facing slope has a substantial movement history indicated by multiple lobes on the surface (supplementary 3, Del Vecchio et al (2023)), but is currently experiencing minimal to no movement.The south-east facing slope, with intermediate MAT1, is currently undergoing the most substantial movements.The coldest, east facing slope is displaying movement, although to a lesser degree due to the presence of colder permafrost and shallower thawing depth.This indicates a progressive soil deformation pattern with progressing permafrost thaw, intricately associated with the parabolic susceptibility to slope instability as depicted in figure 10 (right panel).A plausible hypothesis is that deformation rates are decreasing on the south facing slope, since thaw depths are already very deep.In contrast, deformation rates may be peaking on the south-east facing slope, and beginning to accelerate on the east facing slope.
While studies such as those by Kokelj et al (2015) or Matsuoka (1996) emphasize a correlation between rainfall and soil movements in permafrost terrain, our findings show that the influence of rainfall on the observed movements is limited, at least during the monitoring period.The locations associated with increase in soil movement following rain events show variable timing after the rain events without clear patterns across the landscape.This is likely due to the fact that soil saturation is high across the watershed, leading to limited infiltration of rainwater, and hence negligible changes in pore water pressures.However, large rainfall events may hasten soil movement via increased heat transfer, accelerated permafrost thaw (Mekonnen et al 2021), and increase in porewater pressure.
This study, like other studies (e.g.Kinnard and Lewkowicz 2005, Lewkowicz and Harris 2005, Bommer et al 2012), shows that in Arctic environments soil temperature plays a pivotal role in triggering and regulating shallow slope movements.Soil ice content and thawing depth (h) control the critical soil parameters governing slope stability, including pore pressure (u), cohesion (c) and internal friction angle (φ ′ ) (figure 10, left panel) and subsequent slope movements (figure 10, right panel).The computed Fs proxy shows that, besides slope angle, the risk of hillslope movements varies strongly as a function of temperature, and that the use of ground temperature may help identify when soil thaw is associated with high susceptibility to slope movements.

Conclusion
The results of this study indicate that most of the shallow movements observed across warm permafrost hillslopes occur along the thaw-freeze interface.These soil movements are triggered upon reaching thawing depths between 0.4-0.75m, after which the entire layer moves cohesively as a unified entity.Notably, this study finds the largest movements in colder areas, which can be attributed to ice-lens formation via two-sided freezing in these locations.Furthermore, the study underscores the complexity of processes involved in slope instability through the estimation of a Fs proxy, underlining the importance of capturing the depth profile of soil temperature, water and ice content, and slope to estimate landscape susceptibility to slope instability.The outcomes of this study exhibit promising potential for predicting slope susceptibility to movement across Arctic regions.A multi-year analysis will provide additional information on trends in thaw depths and their influence on soil movement.Moreover, a more holistic understanding could be obtained by integrating data from multiple sites that encompass different geological characteristics, exposure to various environmental factors, and diverse slope setups.Overall, this study contributes to our understanding of the main factors controlling slope movements, providing valuable insights for future mitigation strategies, hazard assessment, and estimates of soil carbon stocks and fluxes.
Climate warming induced thawing of permafrost (Lewkowicz and Way 2019, Dobricic et al 2020, Smith et al 2022), and the increasing frequency of extreme events such as wildfire (Lewkowicz and Harris 2005, Lipovsky et al 2006) or major rainfall events (Kokelj et al 2015) significantly impact landscape and ecosystem processes (Olefeldt et al 2016).Soil movement is a key consequence of permafrost thaw, and poses natural hazards (Patton et al 2019, Lader et al 2023), causes damage to infrastructure (Darrow et al 2016, Hjort et al 2022), and influences the carbon budget (Turetsky et al 2020, Vascik et al 2021).The exact impact of soil movement on carbon fluxes remains unclear (Pautler et al 2010, Lafrenière and Lamoureux 2013, Beamish et al 2014, Patton et al 2019, Turetsky et al 2020).Turetsky et al (

Figure 1 .
Figure 1.(a) Map of the study site with main streams, ERT profiles and probe locations.Geological units Im-CS, MB and Qu correspond to impure marble and calc-schist of the Nome group, metabasalt of the Nome group and quaternary surficial deposits, respectively (Till et al 2011).(b) 3D map of inverted ERT profiles and probe positions across the watershed (Uhlemann et al 2023).Colors represent the soil deformation and temperature monitoring locations on south (orange), south-east (green), and east facing (blue) slopes.(c) Photograph of the watershed taken from the South facing slope on 22 June 2022 (view direction indicated by the white icon in (a)).

Figure 2 .
Figure 2. Schematic of parameters impacting slope stability.T is the temperature, β is the slope angle, h the thaw depth, γ the unit weight, w the water content, c the cohesion and φ ′ the internal friction angle.
et al 2009) is widely used (Harris and Lewkowicz 2000, Harris et al 2008, Bommer et al 2012, Wang et al 2022) to assess the soil shear strength by considering soil cohesion (soil's internal strength, c in Pa), the internal friction angle (angle threshold before particle starts sliding, φ ' in • ) and, the pore pressure (u in Pa).The Factor of safety (F s ) is commonly used to assess the ratio of the resisting forces to the driving forces (Nater et al 2008, Bommer et al 2012):

Figure 3 .
Figure 3. Map showing the fluxes m 2 yr −1 for each probe location and associated MAT1.

Figure 6 .
Figure 6.Deformation characteristics of all 48 probes.(a) movement at the surface as function of thawing depth.Black lines highlight 0.4 and 0.75 m thawing depths.(b) Timing of the thawing front depth.(c) Velocity at the surface as function of time.Red dashed lines highlight the main rainfall events (figure3).

Figure 7 .
Figure 7. (a) Velocity profiles for the entire monitoring period.Thick black line corresponds to the average value.(b) Normalized velocity profiles for all probes exhibiting more than 50 mm of displacement during the monitoring period.Thick black line corresponds to the average value.

Figure 8 .
Figure 8. Cross-plots of main parameters controlling the movements of the 29 probes embedded in frozen soil with their associated total surface displacement.(a) Cross plot of slope angle and moisture content.(b) Cross plot of MAT1 and maximum thaw depth.Dotted lines indicate thresholds in main parameters needed to trigger soil movement, red shadow box indicates area where movements could be expected.

Figure 9 .
Figure 9. (a) Sensitivity analysis of the four main parameters used in this study in the calculation of the factor of safety, including soil water content, MAT1, slope angle and maximum thaw depth.(b) Isolines showing the variation of the factor of safety in relation to slope angle and temperature fluctuations.Additionally, the 29 locations where probes were embedded in permafrost are represented with their associated surface displacement as color.Probes experiencing displacement greater than 50 mm are circled in red.

Figure 10 .
Figure 10.Schematic highlighting the complexity in how properties influencing the susceptibility (S) to slope instability change with soil temperature.Shaded dotted areas highlight relatively high uncertainty in defining cohesion and internal friction angle due to temperature variations, water phase change, and material properties.

Table 1 .
Soil properties used for the computation of the factor of safety (Fs).