The Largest Landslides on Earth part of Vignettes:Vignette Collection
In landscape evolution, landslides are thought to play a prime role in controlling hillslope erosional response to tectonic, climatic, and anthropogenic forcing (e.g. Korup et al., 2007; Hewitt et al., 2008). Besides, landslides are a major geological hazard and frequently occur in the wake of strong earthquakes or high-intensity rainstorms such as tropical cyclones. Several studies have underpinned an empirical relationship between landslide frequency and magnitude (which entails a spectrum from less than a cubic meter to several thousands of cubic kilometers), although most of these studies typically concern a fraction of the full size spectrum only, i.e. the one containing the numerous smaller slope failures. Judging from reported frequency-magnitude relationships, the largest of these landslides are extremely rare with average return periods well beyond a human lifetime. So what is the largest landslide on Earth and in what geo(morpho)logical setting does it occur? Answering this question calls for reviewing many different case studies and inventories, i.e. databases that contain varied information on the causes, triggers, geometry, and consequences of landslides. Such comparison reveals that the largest landslides on Earth mainly occur below sea level. One of the largest fully submarine slope failures, the Storegga Slide, is on the continental shelf 120 km west of Norway (Haflidason et al., 2004). It occurred around ~8.1 ka, and involved the geologically instantaneous detachment of 2400 to 3200 km3 of sediment, covering an area of 95000 km2. Increasingly detailed bathymetric mapping of the ocean floors has revealed many well-preserved landslide deposits of comparable volumes. While the largest of these submarine landslides are at least one order of magnitude larger than their terrestrial cousins, even bigger landslides have been found on planets and moons in the solar system. With the exception of the early Eocene 3400-km2 Heart Mountain landslide, Wyoming and Montana (Aharonov and Anders, 2006), the largest terrestrial landslides typically involve the catastrophic flank collapse of volcanic edifices, producing highly mobile debris avalanches. One of the largest reported debris-avalanche deposits is Las Cumbres, Mexico (~81 km3). These landslides often travel beyond the 10-km mark and bury up to several thousands of km2 with sheets of hummocky volcanic debris. Upon entering drainage channels, debris avalanches may transform into lahars and hyperconcentrated flows, which further increase their mobility and impact range. Debris avalanches from volcanic islands may create tsunami upon hitting the ocean surface, while continuing their runout along the submarine flanks of the volcano into the deeper sea. The largest terrestrial landslides of non-volcanic origin can often be found in tectonically active mountain belts prone to frequent earthquake shaking, high-intensity rainstorms, and slope undercutting, and thus high relief with steep hillslopes. However, the frequent occurrence of such dynamic loads on hillslopes cannot be reconciled with the rare occurrence of the largest of landslides. In other words, the stability thresholds required for triggering are either (a) exceeded very rarely, e.g. during very large earthquakes, or (b) as a consequence of gradual deterioration of hillslope strength, allowing detachment of large landslides at lower stress levels. Moreover, many of the largest landslides are not necessarily tied to high local relief, and a significant fraction has occurred in rather flat terrain, involving the detachment of many cubic kilometres of mass along low-gradient basal shear planes. Most of the largest terrestrial non-volcanic landslides involve bedrock regardless of whether it is igneous, metamorphic or sedimentary. Indeed, loosely consolidated Quaternary sediments may also be prone to large-scale catastrophic collapse, and the title of "largest terrestrial non-volcanic landslide" is claimed by, among others of similar estimated volume, the Pleistocene Baga Bogd earth-block slide, Mongolia (~50 km3; Philip and Ritz, 1999). Currently we know of some 200 terrestrial landslides that have mobilised >1 km3 material each. Four such landslides occurred in the 20th century alone, while many regions of the world remain under-explored with regard to the geomorphic legacy of such extreme slope failures. In contrast, there are regions that seem to be particularly susceptible to extremely large landslides, such as North Island, New Zealand. There, the dissected hill country composed of soft upper Tertiary silt- and mudstones features clusters of huge (>1 km3) low-gradient landslides. These can be best appreciated from satellite imagery or shaded digital elevation models, as they form conspicuous deviations from regional trends of drainage pattern and host a number of ponds on their surface or margins. The upper Indus Basin, Pakistan, hosts the densest cluster of large (>1 km3) catastrophic rock-avalanche deposits known to date, and many of these landslides have dammed major mountain rivers, creating extensive valley fills, and profoundly altering long-term fluvial sediment transport. Although rare, it is estimated that single large landslides may have accounted for 1-10% of Holocene erosion rates in mountainous catchments during the Holocene. The contribution of large river-damming landslides to sediment yields can be substantial, once fluvial erosion is able to dissect the dam and entrain the landslide debris. On a global comparison, such point-source contributions of single large landslides rival the fluvial specific yields (expressed as the mean annual mass of sediment trasnported per unit upstream drainage area) following regional landslide episodes involving up to tens of thousands of smaller landslides, or even the extremely high fluvial and hyperconcentrated yields of pyroclastic debris in the wake of volcanic eruptions. Sediment delivery is often extremely pulsed and exceeds 105 t km-2 yr-1 annually during the rapid erosion of debris from large catastrophic bedrock landslides. Over the course of a few hours, however, debris flows during catastrophic breaching of river-blocking landslides may produce unmatched sediment yield peaks of >109 t km-2 yr-1, thus constituting the most effective short-term sediment transport events in mountain rivers.
The Enigma of Inner Gorges in the Alps part of Vignettes:Vignette Collection
Inner gorges form at the interface between hillslopes and river channels. Seen in cross section, they form an often near-symmetrical convex break the steepness of the hillslope profiles. This distinctive valley-in-valley topography may occur in both rock and debris slopes (Kelsey, 1988). Many slot gorges, and also some canyons, can be considered to be inner gorges. Several mechanisms have been proposed for the formation of inner gorges cut into bedrock. These include relief rejuvenation by fluvial incision in response to rapid base level drop, linear erosion along weakened fault-zone rocks, repeated glaciations during glacial-interglacial cycles, frequent pore pressure-driven landsliding focused at hillslope toes, or incision during catastrophic outburst flows from natural dam failures. To show that disentangling these formative mechanisms is not a merely academic exercise, but a major challenge to understanding alpine landscape evolution, we consider here inner gorges in the European Alps: In the northward draining rivers of the central European Alps inner gorges occur mainly in calcareous sedimentary rocks, where they can be several hundred meters deep and extend for many kilometers upstream of major confluences. In contrast, inner gorges in crystalline rocks are less numerous and deep. In the eastern Swiss Alps, inner gorges occupy some 10% of the trunk drainage network, although the total percentage is difficult to estimate, given that many gorges have been artificially impounded for hydropower generation (Korup and Schlunegger, 2007). Some of the most prominent and widespread inner gorges occur in soft Mesozoic calcareous schists, which grade into the lithologically very similar lower Tertiary flysch rocks of the Penninic Nappes. The river longitudinal profiles of these inner gorges formally resemble the theoretically predicted transient response of detachment-limited bedrock rivers to rapid base level fall by headward knickpoint migration. Thus it seems reasonable to assume that these gorges were cut into a formerly ice-sculpted valley floor since the Last Glacial Maximum, thus acting to return river channels to base level from positions of hanging valleys. Potential motors for such a postulated incision pulse comprise base-level falls induced by downwasting of trunk valley glaciers, differential glacio-isostatic rebound, and neotectonic activity. The hypothetical postglacial incision of these inner gorges requires local downcutting rates in excess of 20 mm/yr, and hillslopes to adjust by frequent landsliding toward development of a threshold state. Simple rock-slope stability models indeed suggest that many of the gorge walls are at a mechanical threshold limited, however, to the lower 25% of total hillslope relief (Korup and Schlunegger, 2007). However, the required fluvial bedrock incision rates rival or even exceed those reported from tectonically much more active and rapidly uplifting mountain belts such as the Himalayas! So are the inner gorges in the Alps testimony to comparable geomorphic process, if not tectonic, activity? Anecdotal evidence from the days prior to extensive river training refers to frequent debris flows from many of the drainage basins, in which inner gorges slice their way into soft calcareous schists and flysch. Moreover, the volume of well preserved debris fans fed by these basins, and perched on dated alluvial surfaces, require towering rates of sediment yield averaged over most of the Holocene. These yields are of the order of 104 m3/km2/yr, which we would typically associate with rapidly eroding terrain rather than the seemingly quiescent Alps, where mean postglacial denudation is of the order of 1 mm/yr (Schlunegger and Hinderer, 2003). It remains unknown, however, whether such soaring yields were sourced from bedrock erosion or valley fills. Along these lines, there is some compelling evidence that argues in favor of a pre-Holocene origin for at least some of the inner gorges in the Alps. There are several independent, though largely qualitative, observations of older sediment fills comprising indurated and strongly weathered morainic or glacifluvial material in some of the gorges (e.g. Tricart, 1960; De Graaff, 1996). Rock knobs and truncated spurs at tributary junctions attest to several episodes of gorge cutting, with former bedrock channels being left stranded above river confluences. Clearly, both such sediment fills and "avulsions" in bedrock would call for even higher (and previously unreported) bedrock incision rates, were we to be supportive of a purely postglacial age of these gorges. Why is this discussion about the onset of gorge incision so significant? Overall, any pre-Holocene age of the inner gorges implies that a fluvially sculpted gorge topography must have survived at least the last glaciation. This could have been achieved by subglacial sediment fill during glacial cycles. Importantly, this would suggest that glacial erosion was not able to fully eradicate the fluvial bedrock topography formed during interglacials. Alternatively, much of the gorge relief could have been formed by subglacial meltwater, and much of the current gorge topography could be inherited from non-fluvial processes. This may complicate analyses of bedrock river longitudinal profiles, which have been shown useful to quantitatively constrain fluvial erosional response to tectonic deformation, climatic oscillations, or changes to sediment flux. Exposure dating of strath terraces, pothole surfaces, or otherwise "fluvially" polished surfaces flanking actively incising bedrock rivers may yield erroneous rates of river downcutting if they were partly formed and thus shielded by ice.