Holocene gigascale rock avalanches in Vaigat strait, West Greenland—Implications for geohazard

In a warming climate, landslide occurrence is projected to increase in the Arctic (IPCC, 2019). Recent landslides in Alaska in 2015, Chile in 2007, and Greenland in 1952, 2000, and 2017 have demonstrated the hazard (Dahl-Jensen et al., 2004; Sepúlveda et al., 2010; Higman et al., 2018; Svennevig et al., 2020, 2023). These events involved high-velocity subaerial rock avalanches (sensu Hungr et al., 2014; Hermanns et al., 2021), which generated large waves via displacement of water as they impacted the sea. Displacement waves are also called tsunamis (Hermanns et al., 2013).

To better understand the postglacial occurrence and magnitude of rock avalanches within high-latitude regions, we examined the geomorphological landslide record of the Vaigat strait and compared it to historic accounts previously used as “worst-case scenarios” (Dahl-Jensen et al., 2004). We did this by using a new digital elevation model (DEM) compilation of high-resolution topographic and bathymetric data alongside seismic reflection data and onshore observations. The SPLASH formula was then used to estimate displacement wave runup (Oppikofer et al., 2018).

On the west coast of Greenland, located between Disko island (Qeqertarsuaq) and the Nuussuaq peninsula (Fig. 1), the Vaigat strait (Sullorsuaq) consists of a 2600-m-deep, steep-sided glacial valley ~110 km long and 15–30 km wide. The valley floor is below the present-day sea level, forming a 10–25-km-wide strait down to 600 m water depth. The bedrock geology consists of the Nuussuaq Basin, which is composed of Cretaceous–Paleocene mudstones and poorly lithified sandstones overlain by extensive Paleogene volcanic rocks and intruded by associated sills and dikes (Chalmers and Pulvertaft, 2001). The strait was filled with ice during Quaternary glaciations and deglaciated between 12 and 10 ka (Weidick and Bennike, 2007; Hogan et al., 2012). Numerous nonspecific landslide deposits have been identified during previous onshore geological mapping (Pedersen et al., 2001, 2007). Furthermore, the area has previously been identified as a landslide “hotspot,” with numerous Holocene landslide deposits mapped and landslide displacement waves observed in 1952 CE and 2000 CE (Dahl-Jensen et al., 2004; Svennevig, 2019; Svennevig et al., 2023).

We based our geomorphological interpretations on a new DEM compilation (~10 m resolution onshore and 25 m offshore), which, alongside seismic data and limited onshore fieldwork, provided an overview of gigascale landslides in the study area. The Supplemental Material¹ provides further details on these data sets and seismic facies interpretation.

We calculated landslide volumes via methods based on geometrical estimation of the onshore rockslide scarp volume, as well as of the subaqueous deposits, and we discuss these methods in the Supplemental Material. Estimated displacement wave height was calculated using the SPLASH formula (Table 1; Table S3; Oppikofer et al., 2018).

Based on DEM observations of long runout, a blocky/hummocky carapace, and a large volume, along with geophysical facies indicative of mass-movement deposits (chaotic to transparent convex to ponded unit geometry) in the seismic and Parasound data, we identified nine complexes of rock avalanches in Vaigat (Fig. 1; Fig. S1; sensu Hungr et al., 2014). As we do not know how many individual rock avalanches comprise each landform, we term them rock avalanche complexes. Here, we present a brief discussion of the largest three, named the Ivissussat, Uppalluk, and Ujarasussuk rock avalanche complexes (Fig. 1). Key geometric parameters from these complexes are shown in Table 1 and Table S3.

The Ivissussat complex covers an area of 85 km², with a total length of 13.6 km and maximum width of 9.1 km. The source area is in the volcanic and sedimentary bedrock succession on the mountain of Ivissussat Qaqqaat, where the head scarp is situated at an elevation of ~1790 m. A maximum water depth of 546 m was recorded at the toe of the deposit (Fig. 1E). The deposit morphology consists of hummocks, blocks, and instances of smoother lobate landforms. Individual blocks have lateral dimensions up to 1 km, relief of ~100 m, and flank gradients of 36° (Fig. 1C). Further, more than 50 isolated hummocks were observed with diameters of 100–300 m and vertical relief up to 100 m.

Located 5 km farther southeast, the Uppalluk complex covers 123 km², with a total length of 18.5 km and maximum width of 9.7 km (Fig. 1C). The head-scarp elevation is ~1580 m, and the bedrock of the source area consists of the same volcanic and sedimentary succession as for the Ivissussat complex (Fig. 1E). Water depth at the toe is 526 m. As with the Ivissussat complex, the morphology is characterized by numerous blocks and hummocks with lateral dimensions up to 300 m and vertical relief commonly between 50 and 100 m. The deposits at Uppalluk also display indications of marine sediment deformation, such as in front of the 2.5-km-wide “lobe 1” at the toe of the slide (Fig. 1E).

The Ujarasussuk complex, located on the northern Disko coast, is the largest in the study area, affecting an area of 133 km², with a total length of 19.0 km and maximum width of 9.6 km (Fig. 1D). The clear arcuate rockslide scar extends from the basalt cliff at 1120 m through the sedimentary succession into the offshore area (Fig. 1F). Water depth at the toe is 523 m. Its morphology is distinct from the two other complexes, with more clearly defined offshore lobes (n = 5; see Fig. 1F; Table S3). Surrounding the distal part of the largest lobe, there are numerous isolated hummocks (n = 159), some of which are extremely large with lateral dimensions reaching 900 m. Onshore, inside and adjacent to the rockslide scar, there are several blocks of basalt of the same dimension and morphology as the offshore hummocks, and these show evidence of rotation during slide movement (Figs. 1F and 2C).

Best-estimate volumes for the marine complex deposits are 1.7 km³ for Ivissussat, 5.2 km³ for Uppalluk, and 8.4 km³ for Ujarasussuk (Table 1). The different methods used to calculate volumes yielded different results, as discussed in the Supplemental Material and shown in Table S3.

Based on the morphologies described above, along with a comparison to subaqueous landslide deposit morphologies elsewhere, we interpret these features as subaerially sourced rock avalanche complexes (e.g., Day et al., 2015; Owen et al., 2018; Hughes et al., 2021; see facies interpretation in Table S2). Considering the regional geology and the presence of blocks of volcanic rock of similar morphology on the coastal slope, it is probable that some of the hummocks with high vertical to lateral dimension ratios and steep sides are large blocks of volcanic rock transported >15 km in a rock avalanche (Figs. 1E, 1F, and 2C; Table S2). For example, “hummock 1” of the Ujarasussuk complex is 500 m wide (0.08 km³ volume) and is located 17 km from the source, >3 km distance up the facing submarine slope (Figs. 1F and 3A). Similar-magnitude processes have been suggested for the giant Flims rockslide (Calhoun et al., 2015). The blocks with low vertical to lateral dimension ratios visible in the complexes at Ivissussat and Uppalluk have the appearance of secondary translational seabed failures (Canals et al., 2004; Hermanns et al., 2014). The more clearly defined, smooth-surfaced, convex-up lobes (e.g., Ujarasussuk lobes 1 and 2; Figs. 1F and 2B) are interpreted as finer-grained debrites. These were likely sourced from the poorly lithified Cretaceous–Paleogene clastic sediments onshore stratigraphically below the volcanic succession.

An examination of the angle of reach–volume relationship of the rock avalanche complexes compared with a reference data set from Hermanns et al. (2021) indicated that each rock avalanche deposit could have resulted from a single large-magnitude event or a few events with individual minimum volumes of 0.1–1 km³ (Fig. 3B). Higher-resolution data are required to determine the potential number of events in each complex.

However, the DEM provided some evidence of multiple events (i.e., lobes at Ujarasussuk; Fig. 1F), and the historical record of landslides observed in Vaigat post–1952 CE demonstrates that, to some degree, the complexes are formed from several events. Furthermore, Parasound data from the Ivissussat complex (Fig. 2A) showed evidence of landslide activity below, within, and above a parallel-bedded unit, which is known to represent deglacial to Holocene sediments (Hogan et al., 2012). Evidence of multiple events was also seen inside individual landslide units in the seismic sections (Figs. 2A and 2B). This demonstrates that at least the Ivissussat complex has experienced minor activity in recent times. The main deposit is, however, at the base of deglacial to Holocene sediments leading us to conclude that the main phase of rock avalanche activity is in early Holocene times. A similar pattern has been observed in Norwegian fjords (Böhme et al., 2015). The absolute ages of the giant rock avalanches in Vaigat remains a topic of further investigation.

The large volumes and blocks along with the high energy necessary to produce the long runouts point to a significant displacement wave potential for the rock avalanches complexes. If the rock avalanche complexes were single rock avalanche events, then they could have produced waves with 160–280 m runup on the opposite shore of Vaigat (Table 1; following the equation of Oppikofer et al., 2018). Allowing for formation of the complexes by multiple events, they are still significantly larger than recent rock avalanches that produced displacement waves at high latitudes (Table S3; Fig. 3). By way of reference, the Paatuut landslide in 2000 CE, with a volume of 0.03 km³, generated a displacement wave with a nearfield runup of 50 m (Dahl-Jensen et al., 2004). This recent event is two orders of magnitude smaller than the individual lobe 1 of 3.3 km³ at the Ujarasussuk complex (Table S3).

While there are larger known landslides associated with active volcanoes as well as other cubic-kilometer-scale subaerial rock avalanches (Korup et al., 2007; Penna et al., 2011), the rock avalanche complexes in Vaigat described here are among the largest on Earth and had the potential to generate extremely large displacement waves.

From a hazard perspective, a key question is whether events of similar large magnitude to those of the past are likely to occur in the future, especially given that future warming is projected to exceed earlier Holocene fluctuations (Marcott et al., 2013). Mechanisms for preconditioning rock slope failures in this time interval (<3 ka after deglaciation) elsewhere have been suggested to be glacial debuttressing, increased seismic activity caused by isostatic uplift, and rapidly changing climate conditions (Ballantyne et al., 2014; Böhme et al., 2015; Sæmundsson et al., 2021). Potential preconditioning factors for the rock avalanche complexes in Vaigat remain unknown. Two recent, and order-of-magnitude smaller, landslides in Vaigat (Fig. 1B) led Svennevig et al. (2022, 2023) to conclude that they were preconditioned by permafrost degradation and that destabilizing effects of warming permafrost in this region may have penetrated >80 m inside the slopes. However, whether such processes played a role in the gigascale rock avalanche complexes and whether the present warming can affect larger slope areas in the near future remain topics of further study.

Nevertheless, given the rate of climatic change the polar regions are experiencing because of accelerating global warming (IPCC, 2019), when assessing the threat from landslide displacement waves at high latitudes, we believe it is important to consider events not only from the short historical record, but also the Holocene geological record. When society overlooks the geological record of events, the hazards that we face are severely underestimated. Examples of this are shown by the identification of tsunami lag deposits of Holocene age on the Sendai Plain, Japan, and the discovery of submarine landslides of late Holocene age in the western North Atlantic, both of which shorten the estimated return period of high-magnitude events (Goto et al., 2011; Normandeau et al., 2019).

We documented rock avalanches of deglacial to Holocene age that caused displacement waves in Vaigat, Greenland, of which individual components are at least an order of magnitude larger than those observed historically. Though further work is required to constrain age and preconditioning factors, these rock avalanches are a vital consideration for this type of hazard assessment in high-latitude regions. We suggest that they show the need to reassess this type of hazard in this and in similar regions, especially in the light of the warming climate.

The governments of Denmark and Greenland funded the “Study of the risk for serious landslides in Greenland 2019–2022,” for which the original technical work was undertaken. The National Aeronautics and Space Administration (NASA) Oceans Melting Greenland (OMG) project provided vital bathymetry data, and the German Science Foundation (DFG) funded the expedition MSM05/03, during which Parasound data and further bathymetric data were acquired. We are grateful to Louise Mary Vick, Michael Clare, Mauro Soldati, Marten Geertsema, Jeff Coe, and two anonymous reviewers, whose constructive reviews significantly improved this paper.