Howe Sound

Howe Sound provides an example of the variety of mass movement styles in BC fjords. Submarine slides in Howe Sound are mostly located seaward of mountainous river valleys (Fig. 1), the fjord-head Squamish delta and on submerged Quaternary deposits. The fjord sidewalls are typically characterized by steep fan deltas and intermittent subaerial debris flows, which directly feed submarine flows. One of the deadliest terrestrial debris flows in Canadian history occurred at the mouth of one such creek at Britannia Beach and an extensive subaqueous landslide deposit can be observed in the adjacent fjord. Anthropogenic influence is also significant at Britannia Beach, where a mine site disposed of mine tailings for decades into Howe Sound (Jackson et al. 2008). The subaqueous delta of the Squamish River has grown into the fiord and has developed an extensive channel system. Several older slides are apparent along the eastern edge of the Porteau Cove terminal moraine and Keats Island. These likely occurred shortly after deglaciation when supporting ice receded, leaving over-steepened coarse sediment piles susceptible to sliding.

Mass movement and turbidity current events on the Squamish delta front have been examined in great detail (Hughes Clarke et al. 2012, 2013). A training dyke, installed in 1971 to create the Squamish Port facility, has concentrated the full Squamish River discharge into a 500 m section of the fjord-head shoreline on the NW side of the fjord. The training of the river and removal of the natural ability for the river to avulse (switch course) is similar to the situations at the Fraser Delta and at other important developed areas. Hughes Clarke et al. (2012) provide important insight into small scale but frequent failure modes of a delta, which initiate dozens of turbidity current events per year during the spring flood season. These failure modes include slope over-steepening of the delta front and settling river plumes (Hizzett et al. 2018; Hage et al. 2019).

Douglas Channel and Kitimat Arm

Significant mass transport bodies were also identified in focused studies of Douglas Channel (Shaw and Lintern 2016), Kitimat Arm (Shaw et al. 2017; Stacey et al. 2018). In Kitimat Arm, detailed analysis of the morphology, seismic stratigraphy and chronology showed that the classic Kitimat slide, mapped by Prior et al. (1982) and thought to have triggered the 1974 and 1975 tsunamis, was formed from several separate events, at least 1974 and 1975, but also a large slide, which occurred at c. 620 cal years BP (Stacey et al. 2019). An older, buried slide deposit with an age of c. 2.59 cal ka BP was identified, providing constraints on the natural return frequency of submarine slides and slide-generated tsunamis at this fjord-head.

An important outcome of Conway et al.’s (2012) inventory was the identification of two very large landslide features in Douglas Channel. These are the largest features seen in all of the BC inlets. Both occur on the east side of Douglas Channel, and each is in the order of 0.03 km³. They are interpreted to be rotational slides (sackungens) formed by the failure of scallop-shaped blocks that have fallen off the steep-sided fjord walls (Conway and Barrie 2018). The core of these blocks is probably Cretaceous diorite and, being a rigid block, has the potential to have caused tsunamis of up to 40 m in amplitude, or 80 m from trough to peak (Thomson et al. 2012). The run-up on the shoreline would likely have been greater, but was not modelled. The nature of rigid block tsunamis is that they generally have short wavelengths, which means that they lose (dissipate) energy rapidly as they travel. The model indicates that these huge tsunamis at the source are reduced to just a few tens of centimetres by the time they reach the head of the inlet at Kitimat. The ocean currents created by these events would be on the order of 15 m s⁻¹ or 30 nautical miles per hour near the source in Douglas Channel, and similarly would diminish to tenths of m s⁻¹ by at the entrance to Kitimat Arm, 10 to 15 minutes after the initial failure (Thomson et al. 2012).

Fraser River Delta

The Fraser River Delta is a strategically important location for Canada, being adjacent to the city of Vancouver and key infrastructure including an international airport and major coal and container port. The delta front is characterized by small to medium (0.01 km³) retrogressive slides at the top of an active submarine channel and minor gullies on the adjacent slope in front of the most active distributary channel, Main Channel (McKenna et al. 1992; Hill 2012). Recent analysis of bed deposits on the levees of the active submarine channel has refined understanding of the nature of turbidity current activity in the channel (Stacey 2014). Slide-generated flows with a thickness greater than 10 m overspill the main channel on average every 10 to 15 years. Lower density flows occur more frequently and appear to be related to large flood years in the Fraser River.

Unconfined turbidity currents have also been observed on the open slope of the delta through the Ocean Networks Canada cabled observatory (Lintern and Hill 2010; Ayranci et al. 2012; Lintern et al. 2016). These flows occur almost every year during the river flood period; recent research has focused on the triggering mechanism, which likely relates to a combination of high sediment influx from the river, sediment resuspension by tidal currents and potentially groundwater flow and pore pressure effects.

Pacific continental slope

The continental slope off the west coast of Vancouver Island is relatively poorly covered by multibeam sonar. However, those areas surveyed have high potential for tsunamigenic slope failures, preconditioned by the presence of shallow gas or gas hydrates and triggered by megathrust earthquakes (Riedel et al. 2018). The work to map these areas is ongoing and it is anticipated that more submarine landslides will be discovered. A recent example is shown in Figure 2, where a gas seep and possible deep-seated failure plane are visible in seismic profiles at the shelf edge. The plane delimits a wedge of material 300 m wide at the surface and 150 m deep, in relatively well-consolidated sediment and rock, which has the potential to fail in this seismogenic zone.

Seafloor data from an area slightly further offshore, at the toe of the slope, reveal eight subaqueous mass movement deposits (Naegeli 2010). Scholz et al. (2016) determined that these features were likely early to mid-Holocene in age and that anticipated earthquake accelerations would not be sufficient to cause slope instability without significant preconditioning through excess pore pressure. Hamilton et al. (2015) have investigated turbidite beds associated with one of these slides, the Slip Stream Slide (formerly Humpback Slide), and determined that between 10.8 and 6.6 cal ka BP, the recurrence interval for the beds and inferred megathrust earthquakes was 460 ± 140 years. This rate is compatible with other estimates of palaeoseismic megathrust earthquake occurrence rates along the subduction zone.

Farther north, Greene et al. (2018) and Brothers et al. (2019) mapped the Queen Charlotte Fault, a seismically very active transform fault that shows abundant evidence for fluid migration as well as volcanic activity. The fault defines the trend of the continental slope on this part of the margin. Except for one large translational slide (greater than 13 km²), mass movements on the upper slope, above the fault trace are relatively smaller (most less than 2 km²) and the upper slope is characterized by canyons and gullies. Downslope of the fault trace on the margins of a terrace feature, there are several larger (up to 18 km²) square-headed block slides.

Offshore Nova Scotia

Recent research on subaqueous landslide deposits offshore Nova Scotia has mainly concentrated on the Scotian Slope. Studies that utilize industry 2D and 3D seismic reflection data provide new insights into failure processes such as their distribution along the margin (Campbell et al. 2015), erosional effects (Mosher and Campbell 2010; Deptuck and Campbell 2012; Campbell and Mosher 2016) and tsunami potential (Mosher 2009). Offshore exploration wells provide ground truth and age control for these studies, albeit the samples generally comprise drill cuttings and sidewall cores, and biostratigraphic age resolution is limited compared to radiometric methods.

Late Pleistocene glacial process have a significant influence on the modern seabed morphology of the NW Atlantic margin (Mosher et al. 2017; Normandeau et al. 2019a). On the Scotian Slope, a number of studies have looked at subaqueous landslide processes that were prevalent during deglaciation of the margin (e.g. Huppertz et al. 2010; Piper et al. 2012, 2016). Although these studies examine deglacial processes not currently active on the Atlantic margin, they are good analogies for what may happen in the future in Arctic settings.

Study of Holocene submarine landslides may hold the most relevance in terms of assessing modern slope stability and understanding future hazard. Recent discoveries of large (>14 000 km²), previously unidentified Holocene-age submarine slides on the Scotian Slope (Normandeau et al. 2019b, c), have led to a re-evaluation of the submarine landslide hazard on the Atlantic margin. At least four landslides, including the most recent 1929 Grand Banks event, have occurred on this part of the margin within the last 4 ka, establishing a shorter return period than previously understood, i.e. approximately 1 ka. MacKillop et al. (2019) suggest that cyclical loading (e.g. earthquakes) or preconditioning factors (e.g. elevation of pore pressures) are critical for slope instability on the Scotian Slope. In the light of this new information, a re-evaluation of the rest of the margin is warranted.

Offshore Newfoundland and Labrador

Research has been conducted near the 1929 Grand Banks earthquake area. Giles et al. (2010) showed in 3D seismic reflection data that very large submarine landslides occurred in the region throughout the Cenozoic. Several publications were focused on the 1929 event specifically. Stevenson et al. (2018) reconstructed sediment concentration of the 1929 flow event using two distinct methods and showed that the concentration was orders of magnitude higher than observed elsewhere in smaller flows. Schulten et al. (2019a, b) put forward that the 1929 submarine landslide consisted of two failure mechanisms; massive slumping along deep faults and décollements followed by widespread surficial sediment failure. Løvholt et al. (2019) used the new failure mechanism to refine numerical models of the 1929 tsunami.

Along the continental slope off the SE Grand Banks, Rashid et al. (2017, 2019) examined the interplay of the Labrador Current and mass movement, in turn presenting a late Pleistocene to present chronology of subaqueous landslides and new ideas on how ice-rafting deposits and contourites impact the geotechnical parameters of the seabed.

Flemish Pass is a large saddle, separating the Grand banks from Flemish Cap and is an area of active hydrocarbon exploration. Using 3D seismic reflection data, Hawken (2017) examined the interaction between gas hydrate bottom-simulating reflectors (BSRs) and mass movement on Sackville Spur, a large contourite drift in northern Flemish Pass. She noted that the onset of mass movement deposition in the Pleistocene correlates to decreased sediment drift development and increased deposition in the base of Flemish Pass. There is no direct spatial association between the BSRs and mass-transport deposits (MTDs), although the onset of pockmarks and failure scarps are stratigraphically equivalent, suggesting hydrate dissociation and sediment failure may have a common triggering mechanism. In another study, Cameron et al. (2014) describe a large retrogressive slide complex with associated mass transport deposits on the northern flank of Flemish Cap. This multiple failure complex extends about 65 km along slope and individual arcuate failures are up to 10 km wide with scarps over 70 m high. Run-out distances extend up to 20 km downslope. Detailed sedimentological, geotechnical and geochronological studies of the slide complex and adjacent slope established that the failures occurred in the late Pleistocene and that the present slope appears to be stable. Recurrence intervals in this part of Flemish Pass are approximately 10 ka and most likely correspond to the recurrence of triggering earthquakes. Migration of deep fluids along the Cumberland Belt Fault Zone may have preconditioned the region and contributed to past failures.

In Orphan Basin, Li et al. (2012a) examined Pleistocene turbidites and canyon formation, showing that the origin of the canyon is synchronous with the onset of cross-shelf glaciation and that the occurrence of muddy turbidity flows is correlative with the movement of the ice stream. Using 3D seismic reflection data, Li et al. (2012b) described fundamental differences between glacigenic and blocky mass transport deposits; whereas blocky mass transport deposits have characteristic parallel to sub-parallel basal grooves and chaotic internal acoustic facies, glacigenic debris flows show a smooth basal surface and transparent internal acoustic character. Piper et al. (2019) reported that four large failures occurred since 85 ka and 18 failures were large enough to be recognized in seismic reflection profiles and/or cores. They postulated that the widespread mass movements were earthquake triggered and that on such a margin with modest sedimentation rates and low seismicity, mass movement recurrence provides a robust estimator of past seismicity. Further north on the NE Newfoundland shelf and slope, Shaw et al. (2012) and Roger et al. (2013) examined glacial-related mass movement processes and trough mouth fan formation.

St Lawrence Estuary

There has been considerable effort over the last 20 years to map and characterize submarine mass movements, canyons and gravity flow systems in the St Lawrence Estuary (e.g. St-Onge et al. 2004; Cauchon-Voyer et al. 2008, 2011; Gagné et al. 2009; Normandeau et al. 2015; Pinet et al. 2015; Poncet et al. 2010). Pinet et al. (2015) completed a comprehensive analysis of submarine slide morphology based on multibeam bathymetry. They documented 94 mass movements in this relatively small area, with a median area of 3.3 km² up to a maximum of 174 km². These deposits are preferentially located along the topographic break at the margin of the glacially incised Laurentian Channel. They consist of lateral spreading, slumps, debris flows and blocky submarine landslides (Campbell et al. 2008). One of the largest landslides in the St Lawrence Estuary was well studied by Cauchon-Voyer et al. (2008, 2011) in the Betsiamites area and had multiple phases of activity at 7.25 cal ka BP, AD 1663 and AD 1860 or AD 1870. These events are hypothesized to have been triggered by earthquakes originating from the Charlevoix-Kamouraska Seismic Zone. In addition, Turmel et al. (2019) showed that the 7.25 cal ka BP event might have produced a 1.5 m tsunami wave, which is considered low risk to coastal population since the tidal range is 5 m. St-Onge et al. (2004) suggested a reduced frequency of earthquakes in the Saguenay Fjord based on reduced evidence of landslide deposits after 4 ka BP. Although they are hypothesized to be earthquake-triggered, the ages of most of these slope failures are unknown and Pinet et al. (2015) state that the potential relationship with seismicity is difficult to establish for most of them without more detailed work. The preliminary hypothesis is that the recent ones may have been triggered by the M_w7.2 1663 earthquakes (Cauchon-Voyer et al. 2008) but further work is required to confirm their ages. There are few data to support suggestions of preconditioning of the sediments by high sedimentation rates and/or shallow gas (Duchesne et al. 2003; Cauchon-Voyer et al. 2008). Poncet et al. (2010) showed that one of the landslides in the upper St Lawrence Estuary had potentially triggered a tsunami although no physical evidence has been discovered so far.

The St Lawrence Estuary is also characterized by a significant number of small submarine canyon and channel systems, mostly on the north shore of the estuary (Normandeau et al. 2015). These channels and canyons were formed by turbidity currents and were mostly active during deglacial times (Normandeau et al. 2017). Most of them originate from glacier or river sediment supply in the past, are now detached from these sources and so are no longer active. Canyons supplied by longshore drift and triggered by intense storms are episodically active (Gagné et al. 2009, Normandeau et al. 2013, 2015), but, curiously, the most active canyons in the estuary are those that are sediment-starved but located on the steepest slopes (Normandeau et al. 2014). Recent direct monitoring of turbidity currents in this system revealed that storms trigger the turbidity currents by suspending small volumes of sediment in the canyon heads (Normandeau et al. 2020).

The results of this research effort have advanced our knowledge and understanding of the potential hazard posed by submarine slides and mass movements in the St Lawrence Estuary but further work on timing, preconditioning and trigger processes is required to quantify the hazard.

Baffin Bay and Baffin Island Fjords

The seismic hazard in Baffin Bay is high, especially in northern Baffin Bay where one of the largest passive margin earthquakes ever recorded occurred in in 1933 (M7.3; Bent 2002). Consequently, the potential submarine mass movement hazard on the Canadian continental margin of Baffin Bay can also be considered high until more understanding of the occurrence and timing of recent events is obtained. The difficulty of accessing this northern region has hindered systematic mapping of the margin and both multibeam and sub-bottom data coverage are sparse offshore. However, mapping of the fjords has revealed numerous submarine landslides and evidence of gravity flow processes (Hughes Clarke et al. 2015; Brouard and Lajeunesse 2019a, b; Normandeau et al. 2019d).

The Pleistocene history of the margin is characterized by episodes of shelf-crossing glacial ice advance, leaving complex ice proximal deposits including extensive glacigenic debris flow deposits (Campbell and Bennett 2013; Bennett et al. 2014; Jenner et al. 2018). Whereas the acoustic characteristics of these glacigenic debris flow deposits can be distinguished from non-glacigenic mass flow deposits, the complexity of the late Quaternary stratigraphy requires high-resolution surveys and detailed analysis to make possible a rigorous assessment of the submarine mass movement hazard. For example, the seafloor northern Baffin Bay in the vicinity of the 1933 earthquake has been examined and, whereas glacigenic debris flow deposits are widespread (Li et al. 2011), evidence of modern slope failure is absent (Bennett et al. 2014). Further south, off Clyde River and Qikiqtarjuaq, submarine mass movement deposits were reported by Aksu and Hiscott (1989) and Hiscott and Aksu (1994), which, upon further examination, are likely glacigenic in origin based on their acoustic character and location seaward of transverse troughs.

Some fresh slide scars and mass flow complexes have been identified on the continental slope and on the margins of some shelf-crossing glacial troughs (Campbell and Bennett 2013). In addition, mass movement and gravity flow complexes have been widely observed in multibeam surveys of fjords and fjord-head deltas of Baffin Island (Hughes Clarke et al. 2015; Broom 2019; Brouard and Lajeunesse 2019a, b; Normandeau et al. 2019d). Of particular note, at least 243 submarine slides were observed on the margins of Frobisher Bay, within 20 km of the town of Iqaluit, capital of Nunavut (Deering et al. 2015, 2018, 2019; Todd et al. 2016). Broom (2019) showed that Holocene submarine mass movement recurrence in a large basin north of the Hamlet of Pond Inlet is once every 800 years. Gravity flow processes present on delta fronts are associated with glacial retreat. A predictive framework of mass movement on fjord deltas for the entire eastern Baffin Island has been developed and shows that about 43% of the river mouths in Baffin are likely to have active gravity flow processes on their delta (Normandeau and Dietrich 2019). Therefore, the fjords and nearshore environments of Baffin Island appear to be particularly active areas for slope failures and work is underway to understand their distribution, triggers and recurrence.

Beaufort Sea

The presence of significant mass wasting on the continental slope of the Beaufort Sea has been recognized since the 1980s (Hill et al. 1982). The Beaufort Sea is dominated by sediment supply from the Mackenzie River and is therefore characterized by locally high postglacial sedimentation rates (Hill et al. 1992). The seismic hazard is considered moderate (Adams and Atkinson 2003) with the two largest recorded events a M6.5 earthquake in 1920 and a M5 event in 1975 (Hyndman et al. 2005).

Recent detailed studies of the shelf-edge failure identified by Hill et al. (1982) shows that it is a complex feature consisting of at least five phases (Cameron and King 2019), initiating along a single deep failure plane, followed by subsequent, smaller retrogressive failures. It is contiguous with, and is cut by, a second slide complex that originates in mid-slope. This failure complex has at least four phases and has an estimated volume of 40 km³. A second, larger failure complex with a minimum volume of 100 km³ has been identified along slope in a mid-slope position (Cameron and King 2019). This failure complex also has a minimum of four phases. It has proven difficult to establish the age and timing of the individual failure events, but based on ¹⁴C and ²¹⁰Pb chronology, they can all be constrained to the last 1 ka.

The mid-slope location of two of these failure complexes is unusual, but is likely to be closely related to preconditioning by migrating fluids. Recent work has provided strong evidence for the presence of thick but degrading permafrost and gas hydrate beneath the continental shelf and for significant migration of gas towards the continental slope (Paull et al. 2007, 2015; Riedel et al. 2017). Both the Beaufort shelf and slope are characterized by pingo-like features and active mud volcanoes associated with significant venting of fluids (Paull et al. 2007, 2015). Whereas the composition of the fluids indicates multiple sources at depth, it is highly likely that the Beaufort shelf edge and slope sediments are preconditioned for failure by these migrating fluids. These factors suggest that the submarine mass movement hazard is considerable in this area and further work is necessary to better constrain recurrence intervals.

Case study

By its nature, the discovery of landslide deposits is a perpetual work in progress. To give an indication of effort required, and how technologies have evolved, we describe the case history of Kitimat Arm in Douglas Channel. This is a small and shallow area, relative to the coast of Canada, but with multiple attempts over decades to quantify the hazard. In 1975, a tsunami occurred in Kitimat Arm which triggered investigations. Fortunately, there was an eyewitness who, while placing a boulder on the seabed to build a crib wharf, saw the boulder and the whole seafloor disappear beneath his working barge. Had it not been for this, investigators would hardly have known where to start to look for the source of the wave. A series of investigations by consultants and scientists identified events in 1952, 1968, 1971, 1974 and 1975. So, a very specific lead allowed direct observation of several events in a confined area. This was adequate to determine the cause of the specific 1975 tsunami; however, it left questions about the hazard in this and other fjords.

The next attempt at the Douglas Channel was by Conway et al. (2013) during a coast-wide assessment of multibeam data which had been collected up to 2011. The data for Kitimat Arm consisted of perhaps two days of ship time, and of the wider Douglas Channel perhaps a week of ship time. These were not targeted specifically at this hazard. This surficial assessment revealed a further six recent mass transport deposits in Kitimat Arm, and over 80 outside the Arm in the larger Douglas Channel. So, the multibeam assessment certainly provided much information that submarine mass movements are common in the area during recent times. In terms of the hazard, it provided basic geometry or size of recent failures.

From 2013 to 2018, the area was targeted to obtain frequency and magnitude information about mass movements, to inform risk assessment with respect to public safety and a developing coast (Lintern et al. 2019). The area was resurveyed using a sub-bottom profiler and cored to determine the age of failure. This led to the discovery of a further 25 buried deposits (Shaw and Lintern 2016; Shaw et al. 2017; Stacey et al. 2020). It provided researchers with information to calculate frequency and magnitude of mass transport events, and thus links could be made to natural triggers. Importantly though, the sub-bottom data indicated specific areas where landslides repeat throughout history, and Kitimat Delta (the source of the 1975 tsunami) has failed several times throughout the Holocene (Stacey et al. 2018).

As this effort is directed towards advising GSC managers on steps forward for the country's coastlines, it is worth summarizing the effort to examine this one stretch of coastline. It would take a modern ship with multibeam, piston coring and sub-bottom capabilities a maximum of two weeks to determine this high level of information about the hazard in waterways totalling 250 km long and an average of 3 km wide (750 km²). This area indicated over 80 MTDs, 6 of which were investigated to be a present hazard to coastal infrastructure. In addition to the ship work, it is estimated that obtaining this information required, in total, three months of laboratory work (core lab and geotechnical lab), six months of spatial computing (GIS, seismic interpretation), one year of writing results, two months of public consultation and many months of planning in between. In the current configuration a capable scientific research vessel is used for coring, instrument deployment (cameras and other ground-truthing methods) during the day and multibeam and sub-bottom surveys run simultaneously at night. The surveys from the night prior are often processed by hydrographers and ready to plan the next day's science. Thus, it is a very efficient manner of conducting the work. It is tempting to think that some of this work could be transferred to autonomous underwater vehicle (AUV) surveys. In Canada, however, the large distances of uninhabited coastline would require a ship as a base for the AUV operations, and a hydrography team to be on hand regardless. Having an AUV team on many ships would come at the expense of berths for coring and other geoscience expertise. Nevertheless, the GSC's new marine geoscience programme is now commencing the use of AUVs for seabed mapping.