Keywords

1 Introduction

Canada’s national railway network is the dominant mode for exporting natural resources (e.g., coal, oil, grain, potash, forest products) to deep-water marine terminals; and for intermodal goods entering continental North America from global markets. Unfortunately, across much of the continent (Fig. 1), railway infrastructure and operations are confined to transportation corridors with slopes susceptible to a broad range of landslide processes (Geertsema et al. 2009a, b; Bobrowsky and Dominguez 2012). Landslide form and function are dependent upon regional physiographic setting (montane belts, high plateaus, prairies, lowlands), environmental conditions (past and present), and local geology (solid and drift). Regionally, the diverse bedrock geology, wide range of surficial deposits, along with high relief in mountains, plateaus and deeply incised valleys (Fig. 1) favours a broad range of rapid- to slow-moving landslides (Table 1).

Fig. 1
figure 1

Physiography, geology, and permafrost zones of Canada, showing extent of national railway network, major ports, population centres, and location of study area (after Bostock 2014)

Full size image
Table 1 Surficial materials and landslides of southwestern BC (terrain classification after Howes and Kenk 1997; Hervás and Bobrowsky 2009; Highland and Bobrowsky 2008). See Fig. 2 for terrain code legend (after Deblonde et al. 2018)
Full size table

Where transportation corridors traverse unstable slopes, varying degrees of damage to vulnerable infrastructure, or service delays caused by landslides have potential local and national economic, social, and environmental consequences. A landslide-resilient national transportation railway network requires sustainable, cost-effective management of service operations to meet future socioeconomic needs while protecting the natural environment and public. In a scenario of future extreme weather events and climate change, railway infrastructure and operations are expected to face unique challenges in design, monitoring, adaptation, mitigation, reclamation and restoration. An understanding of the geographic distribution and temporal range of earth materials and geological hazards is fundamental to effective hazard management and risk reduction safety (cf. Jespersen-Groth et al. 2009; Laimer 2017).

Here, we present a monitoring framework for railway disaster risk-reduction that incorporates: (1) fundamental spatial and temporal information on geological, geophysical, and geotechnical properties of landslides at national, regional and local scales; with (2) site-specific benchmark monitoring acquired using an array of remote sensing platforms; and (3) in-situ change-detection monitoring technologies and methodologies.

2 Study Area

Because of its socioeconomic primacy, the focus of attention is on landslide processes adversely affecting the national railway corridors connecting the deep-water ports of Vancouver and Squamish in southwestern British Columbia (BC) to the rest of Canada (Fig. 1). From these coastal ports, the Canadian National (CN) and Canadian Pacific (CP) railway corridors traverse mountainous and rolling plateau terrain, running through deeply incised valleys and along the steep shorelines of fjords and lakes (Fig. 2a–c). Gentle to steep bedrock slopes along these transportation corridors are mantled by surficial deposits susceptible to a range of rapid- to slow-moving landslides (Table 1).

Fig. 2
figure 2

CN and CP railway corridors traversing southwestern BC: a Topography and physiography, showing location of cross-sections along CN and CP transportation corridors, key urban settlements, and major physiographic elements. b Surficial geology (modified from Fulton 1995; Deblonde et al. 2018). c Landslide susceptibility (modified from Bobrowsky and Dominguez 2012). d A-A’—CN secondary railway corridor to north-central British Columbia. e B-B’—CN primary railway corridor to eastern Canada. f C–C’—CP primary railway corridor to eastern Canada. Approximate rail grades shown as black dashed (CN) and dotted (CP) lines. FR—Fraser River; TR—Thompson River

Full size image

CN and CP railway transportation corridors are usually <1 km from river channels, fjord coastlines, and lakeshores to maintain an optimal grade (Fig. 2d–f). Located so, railway tracks and infrastructure are prone to damage by landslides, floods, wildfires, and other geological hazards. Local physiography, weather conditions, geological hazards, and land-use activities present local and national economic, social, and environmental challenges (cf. Schuster and Fleming 1986; Haque et al. 2016; Blais-Stevens 2020).

The strategic importance of these vital corridors, along with the need to understand and manage the safety risks related to landslides, make on-site investigations a strategic priority for governments, academia, and the rail industry (Bunce and Chadwick 2012; Hendry et al. 2015). Combining field-based landslide investigation with multi-year geospatial and in-situ time-series monitoring leads to a more resilient railway national transportation network able to meet Canada’s future socioeconomic needs, while ensuring protection of the environment and resource-based communities from landslides related to extreme weather events and climate change.

2.1 ICL Project 202 Landslide Laboratory

Thompson River with its several slides and history afford a study of known slides that, with certainty, experience dynamic behaviour during monitoring efforts. Landslides along a ten-kilometre stretch of the Thompson River valley between the communities of Ashcroft and Spences Bridge (Fig. 3) have negatively impacted critical railway infrastructure, arable land, fisheries, and other natural resources since the 1880s (Stanton 1898; Evans 1984).

Fig. 3
figure 3

Landslides of the Thompson River valley and railway transportation corridor; prime ground control point, survey base station (B); weather station (W). Topographic layer merges World Imagery with a DEM of Canada generated from Aster GDEM v2 worldwide elevation data (1 arc-second resolution)

Full size image

In the late nineteenth century, terraces were intensively irrigated for agricultural land use and toe slopes were incised and over-steepened during railway construction (Stanton 1898; Evans 1984; Clague and Evans 2003). Prehistoric mass movements were consequently reactivated as sudden onset, rapid retrogressive flow-slides during the fall and winter months. The economic importance of this transportation corridor, along with the need to understand and manage the safety risk related to the landslides that threaten the route, mandate the Thompson River valley a monitoring priority for Natural Resources Canada (NRCAN) and the Geological Survey of Canada (GSC).

3 Methods and Results

Monitoring unstable slopes and infrastructure at risk is a cost-effective hazard management practice in southwestern BC that also provides important geoscience information to help develop appropriate stakeholder mitigation and adaptation measures (Bunce and Chadwick 2012; Huntley et al. 2021a). Here, we outline landslide mapping and change-detection monitoring protocols based on the successes of ICL-IPL Project 202 in the Thompson River valley railway transportation corridor that incorporate diverse historical time-series and near real-time geoscience datasets.

3.1 Monitoring to Understand Landslide Form

Descriptions of local topographic, bathymetric, surficial and bedrock geological conditions, including earth materials and landforms, and their hydrological properties are essential for understanding landslide compositions, structures, and behaviours. Geospatial relationships between landslide distribution and specific terrain features, and the environmental conditions triggering instability are determined from field-based, on-site geological and geophysical studies, combined with geotechnical, petrophysical, and geochemical laboratory analysis, which are then codified and quantified in geographic information systems (GIS).

3.2 Remote Sensing and Field Surveys

Understanding form and function begins with terrain analysis and landslide inventory. Terrain polygons and features characteristic of landslide activity are delimited on optical satellite imagery (e.g., Landsat, WorldView), conventional air photographs, and unoccupied aerial vehicle (UAV) photogrammetry. Terrain and landslide classifications are benchmarked by ground observations of slope gradient, surficial materials, material texture, material thickness, slope morphology, moisture conditions, ongoing geomorphic processes, and land cover (Figs. 4b, 5; Huntley and Bobrowsky 2014; Huntley et al. 2017a, 2019a, 2020a; b, c; Holmes et al. 2018, 2020). Terrain and landslides are codified following British Columbia (Howes and Kenk 1997) and GSC mapping standards (Deblonde et al. 2018).

Fig. 4
figure 4

Landslide monitoring best practices along the national railway transportation corridor in the Thompson River valley, southwestern BC. a RS2 InSAR change detection, showing average linear displacement rate rastered at 3 cm yr-1, with purple polygons delimiting 4-sigma confidence levels (Huntley et al. 2021e). b Surficial geological mapping describing sediments and landslide characteristics (Huntley et al. 2020a). c, d Shallow geophysical surveys provide data for interpretation of sub-surface geology and landslide structure (Huntley et al. 2019a). e Boreholes, instrumented to monitor groundwater pressure and slope displacement, provide additional in-situ geological, geophysical, geochemical, and geotechnical data on subsurface conditions (Hendry et al. 2015; Sattler et al. 2018). f UAV photogrammetric data used to generate DSMs and DEMs of landslides, and for change-detection monitoring (see Huntley et al. 2021a). g GPR, single-beam acoustic and multi-beam bathymetric datasets provide information on portions of active landslides submerged beneath Thompson River (Huntley et al. 2019a, 2021a). h Trihedral aluminum corner reflectors installed on active landslides enhance the resolution of persistent scatterer InSAR change-detection (Journault et al. 2018). i Surface displacement data derived from UAV overflights and multi-beam bathymetry data; plotted with RTK-GNSS and d-GNSS displacement data (stable d-GNSS unit—yellow dot; active d-GNSS unit—black and white dot; inactive Geocube™—black dot; active GCP—blue dot) (Huntley et al. 2021a). j Repeat RTK-GNSS surveys of GCPs on active landslides benchmark displacement estimates from InSAR and UAV datasets (Huntley et al. 2021a). k Geocube™ d-GNSS network installed at Ripley Landslide and South Slide provide near real-time displacement measurements (Huntley et al. 2020b). l Fibre Bragg Grating and Brillouin Optical Time Domain Reflectometry strain monitoring of lock-block retaining wall at Ripley Landslide (Huntley et al. 2017c). m Weather station continuously records precipitation, snowfall, air temperature, wind speed and direction (Huntley et al 2020a)

Full size image
Fig. 5
figure 5

© UK Research and Innovation 2020). Terrestrial ERT survey—red lines (A-E); waterborne ERT and EM-31 survey—yellow lines (2–5); PRIME installation—green lines (oriented N-S and E-W, see Fig. 6)

GIS output generated from field studies and laboratory results. Surficial hydrogeological units and landforms of Ripley Landslide, in the Thompson River valley, southwestern BC, as revealed by field-based mapping and geophysical surveys (after Huntley and Bobrowsky 2014; Huntley et al 2019c, 2020a). Locations of logged and monitored boreholes, observation pits, field observations, and ERT survey lines indicated on surficial geology map. ERT pseudosections produced by the British Geological Survey (

Full size image

For the Thompson River valley, landslide distribution is influenced by the distribution of forests, shrub lands, grasslands, wetlands, water, croplands, and railway infrastructure. Vegetation cover generally contributes to increased slope stability through the action of binding roots and water absorption in soils. However, extensive, deep-penetrating roots on steep slopes also contributes to the mechanical and chemical weathering of soils and parent materials by providing additional materials for remobilization if triggered by other factors (e.g., seismic shaking and precipitation events). Lying leeward of the Coast Mountains, the valley experiences semi-arid conditions with annual precipitation as rainfall and snow <2700 mm (Dominguez-Cuesta and Bobrowsky 2011). Greater cyclic and seasonal ground moisture changes due to evapotranspiration varying strongly between winter and summer will also drive deterioration.

Nine surficial material classes are recognized; distinguished by genetic origin, material texture and thickness, slope gradient and morphology, moisture conditions, and active geomorphic processes (Fig. 2b and Table 1). Bedrock influences slope hazards where surficial cover is thin, for example on terrain with slopes >35°. Elsewhere, drift cover is thicker and surficial materials have a greater influence on slope instability. Large retrogressive rotational-translational earth-debris slides were initially triggered by deep incision of Pleistocene fill in the Thompson River valley during the early Holocene (Clague and Evans 2003).

Subsequently, the river has influenced slope stability and landslide susceptibility by changing (a) the pore water pressure in the slope mass, and along rupture surfaces; (b) the supporting force on landslide toes; and (c) through cut-bank erosion, thereby affecting the geometry of the landslide. Back-tilted blocks and rapid debris slumps move over weak, curvilinear rupture zones in glaciolacustrine clay-silt units confined between overlying till and underlying bedrock (Porter et al. 2002; Eshraghian et al. 2007, 2008; Huntley et al. 2020a).

3.3 Geophysical Surveys and In-Situ Monitoring

Geophysical surveys, borehole logging, and laboratory analysis of geotechnical and geochemical properties of earth materials, including stratigraphic layers and structures (e.g., joints, faults, shear planes and tension cracks) provide additional information on internal geological structures and failure mechanisms, as well as how and where groundwater flows through landslide bodies and surrounding bedrock (cf. Whiteley et al. 2019).

Electrical resistivity tomography (ERT), frequency modulated electromagnetic conductivity (FEM), ground-penetrating radar (GPR), seismic primary wave refraction (PWR), and multichannel analysis of shear waves (MASW) surveys provide the most useful information on the distribution earth materials, in addition to the distribution and quantity of groundwater (Figs. 4c, d, 5 and 6; Huntley et al. 2017a, 2019a). Geophysical methods adopted for hydrogeological logging of observation wells include down-hole measurement of natural gamma radiation (GR), induction conductivity (IC) and magnetic susceptibility (MS) (Fig. 5). Subsurface borehole monitoring (Fig. 4e), combining ShapeAccelArray inclinometry with piezometer head levels, indicate that the landslides studied in the Thompson River valley are failing along sub-horizontal, weak, basal shear surfaces in highly plastic clay beds that extend under the river (Porter et al. 2002; Eshraghian et al. 2007, 2008).

Fig. 6
figure 6

In-situ ERT time-series monitoring observations, results and interpretation for the Ripley Landslide, in the Thompson River valley, southwestern BC. a 3D block diagram of merged ERT data captures a static proxy image of subsurface resistive and conductive earth materials (see Fig. 5). b Precipitation, expressed as a weekly average rainfall and effective rainfall; and temperature expressed as a weekly average from September 2017 to May 2018 (Holmes et al. 2020; Sattler et al. 2020). c, d Resistivity profile of PRIME cross-sections (north–south, east–west) with observations of ground conditions and changes in resistivity (Ωm) from December 5, 2017. e, f Ground and resistivity conditions February 6, 2018. g, h Ground and resistivity conditions for May 2, 2018 (after Holmes et al. 2018; Sattler et al. 2018; Huntley et al. 2019c; British Geological Survey ©UK Research and Innovation 2020)

Full size image

At Ripley Landslide, detailed borehole logging shows that the central and northern parts are translating sub-horizontally (2.1 \(^\circ \)–2.5 \(^\circ \)), whereas the southern portion near a lock-block retaining wall has a steeper (28°) slide surface (Fig. 5; Macciotta et al. 2014; Hendry et al. 2015; Schafer et al. 2015).

3.4 Aerial and Bathymetric Surveys

UAVs allow flexible, inexpensive acquisition of low-altitude aerial images, while various off-the-shelf photogrammetric Structure from Motion (SfM) software packages enable production of high-resolution digital surface models (DSM), digital elevation models (DEM) and orthophoto mosaics from such images. These UAV surveys help characterize landslide surface morphology and the spatial extent of displacement (Fig. 4f).

DSMs and corresponding orthophoto mosaics provide very high spatial resolution datasets (2 cm) enabling detailed topographical and textural information (cf. Huntley et al. 2020a, 2021a). UAVs, when equipped with a LiDAR sensor, penetrate the vegetation cover and provide elevation data in areas where the photogrammetry based on SfM method is not likely to produce a representative ground surface. Merged mosaics and DSMs capture the surface condition of landslides, and the extent of bare earth and vegetation growth (e.g., grasses, shrubs, and trees). Metre-scale anthropogenic features (e.g., train tracks, lock-block retaining walls, culverts) are resolvable in high-resolution raster imagery.

Bathymetric surveys better characterize the geometry of submerged landslide toe slopes and identify reaches with channel incision and erosion. GPR, ERT, and acoustic single-beam and multi-beam river surveys (Fig. 4g) reveal variations in channel bed composition ranging from sand and silt draping bedrock to coarse gravel and boulders overlying clay-rich valley fill (Huntley et al. 2019a, b, c, 2021a). Shallow waters (riffles) with rapids lie adjacent to stable terrain, separated by deep scour pools (up to 5 m below river level) adjacent active slide toes.

3.5 Monitoring to Determine Landslide Function

Although some landslides in the Thompson River valley failed and moved rapidly in the past, today all are slow-moving compound features within Pleistocene valley fill (Porter et al. 2002; Eshraghian et al. 2007, 2008; Bunce and Chadwick 2012; Huntley et al. 2020a, b; Fig. 4a). Temporal relationships between landslides and the environmental conditions triggering instability are determined from time-series monitoring and GIS analyses of satellite Interferometric Synthetic Aperture Radar (InSAR), UAV photograms, and ground-based real-time kinematic ground-based global navigation satellite system (RTK-GNSS) surveys (Eshraghian et al. 2008; Bunce and Chadwick 2012; Huntley et al. 2020a, b; Fig. 4a).

3.6 InSAR Change-Detection Monitoring

Satellite remote sensing is an effective first approach for determining the risks posed from ground hazards than other site investigation (Fig. 4a). InSAR provides the best opportunity to determine the extent of landslide activity in the Thompson River valley (Journault et al. 2016, 2018; Huntley et al. 2017b). Slope deformation is monitored using persistent scatterer InSAR techniques, with landslide mapping, modelling, back-analysis of deformation velocities, and long-term deformation trends derived from datasets (e.g., Macciotta et al. 2014; Journault et al. 2016, 2018; Huntley et al. 2017c). Trihedral aluminium corner reflectors, permanent coherent artificial InSAR targets, improves the precision and accuracy of subsequent image processing of images (Fig. 4h). Different line-of-sight (LoS) viewing geometries allow for the projection of vertical and horizontal displacement. Ground movement measured by space-borne InSAR produces results with precision comparable to RTK-GNSS measurements, but with the advantage of monitoring displacement over large areas (Huntley et al. 2017c; Journault et al. 2018).

InSAR monitoring indicates at least five large volume rotational-translational landslides remain active and have potential to adversely impact CN and CP railway infrastructure and services (Fig. 4a; Journault et al. 2016, 2018; Huntley et al. 2017b, 2021a, b, c). Satellite InSAR platforms with repeat visit times of weeks (e.g., RADARSAT-2 [RS2] and SENTINEL-1 [S1]) to days (e.g., RADARSAT Constellation Mission [RCM]) provide rapid monitoring capability with cm-scale precision and accuracy when periodically benchmarked with ground-based RTK-GNSS measurements and UAV photogrammetry. The combination of relatively high spatial and temporal resolution offered by RCM resolves rapid movement over small areas that would otherwise be aliased by the coarser spatial and temporal resolution of RS2 and S1 (Huntley et al. 2021b, c, d). Ground movement is shown to be generally concentrated within the main body of most sliding masses (Fig. 4a). Rates of displacement detected by InSAR vary seasonally, with slower displacement rates occurring during the May to August interval, and higher values from September to April.

3.7 UAV Change-Detection Monitoring

Rigorous change detection requires an accurate and precise elevation model to serve as a reference base topographic map. Repeat UAV surveys have aimed to better characterize the spatial extent, magnitude and direction of landslide movement in the Thompson River valley (Huntley et al. 2021a, d). Planimetric displacement of landslides are first mapped using image co-correlation analysis processing of time-sequenced hill-shaded UAV DSMs. Parameters during hill-shading that align the orientation of the landslide with the simulated sun are used to draw out features for displacement detection. Areas of vegetation and recent track ballast work on hill-shaded images are masked prior to processing to reduce the areas with substantial change not related to slide movement. Two images are created for E/W (X) and N/S (Y) displacement, while elevation changes (Z) are derived from the DSM. These values are added, and then squared to produce a single raster containing 3D displacement values that are all positive, with larger values representing more displacement (Fig. 4i; Huntley et al. 2021a, d).

3.8 RTK-GNSS Change-Detection Monitoring

Repeat RTK-GNSS ground surveys aim to better characterize the magnitude and direction of landslide movement over time (Fig. 4j). GNSS techniques have been successfully employed to determine the three-dimensional coordinates of moving points on landslides (Macciotta et al. 2014; Rodriguez et al. 2018). Ground control points (GCP) are established across the slope using stable boulders and anthropogenic features on, and adjacent to the landslide. A reference base station was established on a stable post-glacial terrace near Black Canyon, 3 km north of the Ripley Landslide (B, Fig. 3). GNSS monuments (Bunce and Chadwick 2012; Macciotta et al. 2014) provide continuous, near real-time monitoring of surface displacement, but only at three trackside locations.

Repeat RTK-GNSS surveys provide point position data across much of the slide body. However, this method generates limited information on the seasonal variation in displacement rates and amounts. Moreover, periodic and continuous monitoring of GCPs and railway infrastructure with small and slow annual displacements (<10 cm year−1) is particularly challenging in an environment with a semi-arid intermontane climate, and an extreme temperature range of −30 to +40 °C. A Geocube (GeoKylia)™ continuous differential (D)GNSS network installed at the Ripley Landslide addresses both issues of spatial and temporal coverage (Fig. 4k). Two areas of maximum displacement recorded by the DGNSS network coincide with maximum displacement indicated by InSAR and UAV analysis. The larger northern zone spans the CN and CP tracks, while a smaller zone at the south end of the landslide near the lock-block retaining wall (Macciotta et al. 2014; Rodriguez et al. 2018; Huntley et al. 2020b, 2021a).

3.9 In-Situ Instrumental Change-Detection Monitoring

Elevated soil moisture was recognized as a driver of slope failure in the Thompson River valley since the first slope stability study in the area (Stanton 1898). Regional and local groundwater conditions contribute to high pore pressures and slope instability along the transportation corridor. Since clay impedes groundwater flow, pore pressure increases along the surface of clay layers, resulting in reduced material strength and decreased slope stability (Porter et al. 2002; Clague and Evans 2003). Piezometers on active landslides record upward hydraulic gradients and elevated pore pressures, confirming toe slopes are in a discharge zone of a regional groundwater flow system (Eshraghian et al. 2008; Hendry et al. 2015).

A strong correlation between river stage and pore pressures suggests Thompson River controls the distribution of groundwater in the slide mass by acting as a lower hydraulic boundary for the regional groundwater regime within the fractured bedrock; and by allowing horizontal connectivity through higher conductivity layers of fluvial cobbles, gravels and sands (Hendry et al. 2015; Schafer et al. 2015).

Proactive infrastructure monitoring and evaluation (PRIME) with continuous ERT and soil moisture acquisitions capture subsurface responses in interannual and seasonal variations in precipitation, in addition to temperature, surface runoff, and snowmelt (Fig. 6a–g; Sattler et al. 2018, 2020; Holmes et al. 2018, 2020; Huntley et al. 2019b, c). Noticeable increases in rate of movement occur when high, prolonged flows of Thompson River sustained by snowmelt are followed by low flows through the autumn and winter months (Macciotta et al. 2014; Schafer et al. 2015; Tappenden 2016). The greatest displacement rates occur from November to March when transitional ground conditions allow snowmelt and rainfall to penetrate deep into the still frozen (or thawing) slide body by way of tension cracks, planar fractures, and bedding surfaces (Fig. 6c, e, g). Resistivity and soil suction show similar seasonal trends, responding cyclically to changing weather conditions from December to March (Fig. 6d, f, h).

Fibre Bragg grating (FBG) and Brillouin optical time domain reflectometry (BOTDR) indicate peak strain rates in the lock-block retaining wall at the south end of the Ripley Landslide also occur in the fall and winter months (Fig. 4l; Huntley et al. 2017c).

In the Thompson River valley, InSAR interferometry, UAV photogrammetry, RTK-GNSS, piezometers, ERT monitoring, and fibre optic reflectometry all show peak movement occurs through winter to spring, indicating factors other than changes in river and groundwater levels influence landslide activity. To understand climatic controls, local weather conditions, including wind speed and direction, rainfall and snow depth, air temperature, ground temperature, and soil matric suction are monitored (W, Figs. 3 and 4m). Weather station data confirm that major precipitation events occur mostly between fall and spring when landslide activity increases (Fig. 6b). Fluctuations in temperature over the winter months also contribute to intervals of landslide activity with thawing soil moisture recharging groundwater (Holmes et al. 2018, 2020; Huntley et al. 2019a, b, c; Sattler et al. 2020, 2021).

4 Discussion and Synthesis

4.1 Landslides and Consequences

Across Canada, relief, slope, aspect, distance from hydrological features, environmental conditions, surficial geology, and land cover conspire to produce a wide range of rapid and slow-moving landslides (Table 1) with the potential to impact railway infrastructure and operations. Landslides have been responsible for numerous casualties, injuries and deaths, costly damage to transportation infrastructure and property, socioeconomic losses, and environmental degradation since the late eighteenth century (Blais-Stevens 2020). Since the late 1800s, the operational and economic consequences of slope failure along the national railway transportation corridors have depended on the scale and rate of movement, and temporal relationship between landslide activity and train timetables.

In the Thompson River valley transportation corridor (Fig. 3), small (<10 cm year−1), incremental surface displacements detected by InSAR, UAVs and periodic RTK-GNSS surveys contribute to minor track misalignment, requiring short-term (seasonal) reorganization of train schedules to allow the safe addition of ballast and realignment of tracks, and to avoid significant time-tabling impacts (i.e., intrinsic consequences). In contrast, frequent, rapid, large and widespread ground movements are a concern for railway companies, government agencies and local communities. Greater surface displacement will damage bridges, culverts, retaining walls and access roads, and cause major track misalignment with the potential for train derailments and service disruption. Potential accompanying negative environmental impacts (i.e., extrinsic consequences) include the loss of natural resources, including fish, wildfowl, game animals, cattle and crops, and potable water for communities.

Global climate change is anticipated to lead to more extreme regional weather events across Canada, along with an increase in the frequency and magnitude of landslides, floods, wildfires and other geological hazards in all provinces and territories (Sauchyn and Nelson 1999; Couture and Evans 2006; St. George 2007). For example, increasing precipitation and higher sustained river flows will change groundwater pressures on slide rupture surfaces extending below rivers and lakes, contributing to increased instability of slopes along the national railway transportation corridors. Climate-driven geological hazards can potentially compromise the safe and secure transport of rural resources and intermodal goods across the continent. A cascade of resulting negative intrinsic and extrinsic consequences for transportation and energy infrastructure, supply chains for goods and services, as well as the environment, will challenge the integrity and resilience of national transportation infrastructure and local communities (Evans and Clague 1997; Geertsema et al. 2006a, b; Blais-Stevens 2020).

The extreme storm system of November 14, 2021, in southwestern BC and its aftermath illustrates such a cascade of events. Following a dryer-than-normal summer and wetter-than-normal fall along all major transportation corridors through the Cascades and Coast Mountains (Fig. 2), wildfire-disturbed slopes received >100 mm of rain in 24 h (and up to 300 mm over 48 h) during an extreme “atmospheric river event” (https://weather.gc.ca/ [URL 2021]). Along with the loss of life (human and livestock), railways, highways, pipelines, power transmission networks, light industrial, and rural–urban infrastructure (e.g., potable water conduits and sewage treatment plants) sustained significant damage because of widespread storm-driven landslides and flooding. The costly relocation of whole communities and reorganization of transportation and shipping schedules led to national socioeconomic losses through service disruption and broken supply chains extending well beyond the winter of 2021–2022. Long-term negative environmental consequences include widespread contamination of agricultural lands and damage to natural fish and wildlife habitats.

4.2 Landslide Mitigation and Monitoring Best Practices

Geotechnical mitigation solutions for landslides can carry significant economic costs and environmental consequences, especially when measures fail to stabilize the slopes of concern (Bunce and Chadwick 2012). In closing, we propose a best-practice solution incorporating three levels of investigation to describe the form and function of the wide range of rapid and slow-moving landslides occurring across Canada in a range of physiographic settings (Fig. 7). Geospatial and temporal change-detection monitoring of active landslides and at-risk infrastructure is a cost-effective hazard management practice that provides important geoscience information to help develop appropriate early warning, mitigation, and adaptation measures in response to extreme weather events driven by climate change.

Fig. 7
figure 7

Conceptual model outlining three levels of investigation to classify, determine susceptibility, and identify landslides of concern for monitoring. Inventory maps, geospatial change-detection, and in-situ time-series monitoring define the interactions between landslides of varying sizes, displacement amounts, and timing of activity; and contribute to estimates of infrastructure vulnerability, anticipated damage, and cascade of consequences contributing to landslide risk. Mitigation solutions reduce the risks to railway transportation corridors running through terrain susceptible to landslides

Full size image

4.3 Level I Investigation (Fundamental Geoscience)

All landslide studies begin with the acquisition of fundamental geoscience information on earth materials, landforms, and their behaviours. GIS landslide inventories are based on the interpretation of high-resolution optical and hyperspectral satellite imagery, air photos, and UAV photogrammetry (Fig. 7). On-site investigations of slope and drainage, bedrock and structures, surficial geology and geomorphology, and vegetation and land cover, when combined with borehole logging, geophysical surveys, petrological and geochemical analyses, and geotechnical studies provide fundamental geoscience information on terrain units and landforms susceptible to mass-movement landslides (Fig. 7).

Landslide inventories and susceptibility maps (e.g., Fig. 2c) allow stakeholders to recognize slopes of concern for further investigation and identify the most suitable geological indicators for level II and III monitoring (Fig. 7). GIS and derivative maps that delimit unstable and potentially unstable terrain are invaluable tools for landslide monitoring and risk management since they can be applied to calculate spatial relationships between landslide distribution, specific terrain features, and environmental conditions expected to correlate with landslide incidence.

Although non-systematic datasets have unequal reliability, accuracy and precision, they nevertheless provide useful insight into landslide distribution at various scales and help focus attention on areas of interest where rigorous qualitative analyses are warranted (Fig. 7).

4.4 Level II Investigation (Geospatial Benchmarks)

For landslides of concern and terrain susceptible to mass-wasting identified in a Level I investigation, ground-based RTK-GNSS measurements with cm-scale precision and accuracy periodically benchmark displacement values determined from satellite InSAR and UAV change-detection datasets (Fig. 7).

As highlighted by the case study in the Thompson River valley, slow-moving mass movements (e.g., rock slumps, debris slides and earth flows) are most effectively monitored by combining InSAR, UAV and RTK-GNSS monitoring. Shallow geophysical and bathymetric surveys also contribute an understanding of the composition and behaviour of a range of slow-moving landslides. Like other parts of Canada, the landscape and harsh environment of southwestern BC presents logistical challenges for installation, and instrumentation of geophysical and continuous GNSS measurements. Permanent GNSS stations and GCPs are less appropriate for monitoring rapid mass movements in steep, mountainous areas with rock avalanches, debris flows and lahars, or in low relief permafrost terrain with active layer detachments and thaw flows where they are prone to damage and loss of data.

4.5 Level III Investigation (Time-Series Monitoring)

Steep, mountainous terrain also presents logistical challenges for installation and in-situ time-series monitoring of rapid-moving rock avalanches, debris flows and lahars in steep, are prone to damage and loss of data. In a Level III investigation, lithological, geophysical, geochemical and geotechnical properties, and other environmental conditions (e.g., river and groundwater levels, soil moisture, weather variables), recognized as drivers of landslide activity and benchmarked during levels I-II, are selected for in-situ time-series monitoring.

Level III investigations are most suitable for slower moving landslides where boreholes can be drilled for monitoring and unstable slopes instrumented safely. In the Thompson valley transportation corridor, Level III InSAR, UAV and RTK-GNSS change-detection datasets conclusively show that geotechnical solutions to stabilize slopes based on current physical models are only partly successful (Journault et al. 2016, 2018; Huntley et al. 2017b, 2021c, e). To better understand the impacts of climate, landslide behaviour is further investigated though time-series monitoring of slopes and infrastructure (e.g., fibre optic reflectometry), boreholes (e.g., piezometers, inclinometers, and acoustic emissions), and weather variables (e.g., rain gauges, snow sensors, thermistors, anemometers, soil moisture meters).

The key goal of geological indicator time-series monitoring is the derivation of thresholds for driving factors that influence landslide dimensions and displacement values (Fig. 7).

4.6 Stakeholder Actions (Climate-Resilient Infrastructure)

The severity of damage to vulnerable railway infrastructure (e.g., tracks, ballast, tunnels, bridges and retaining walls) is a consequence of the areal extent and volume of the landslide(s), temporal changes in landslide activity, and the magnitude of displacement detected during level I–III investigations (Fig. 7). In addition, climate-driven mass-wasting will affect, to varying degrees, sensitive ecosystems, fish, wildlife and water resources, agricultural lands and livestock, communities and people living in proximity to railway infrastructure, and ultimately, supply chains and the national economy (Fig. 7).

Environmental thresholds derived from multi-year datasets capturing spatial and temporal changes in landslide activity in proximity to critical transportation infrastructure enable stakeholders to build robust monitoring and early warning systems, cost-effective mitigation measures, and reliable risk management strategies (Fig. 7). Geospatial and in-situ time-series datasets improve the safety, security and resilience of transportation infrastructure to climate-driven geological hazards, reducing risks to the economy, environment, natural resources and public safety.