A new mechanism for landslide initiation

It is not often that one reads a paper and finds a new and exotic landslide mechanism being suggested. I was somewhat surprised yesterday to find that in a paper just published in Geomorphology, Steve Evans and his co-authors have done just that. Although it requires further research, the mechanism is intriguing and undoubtedly has some very interesting implications for glacial hazard management as well.

The origin of the theory is the extraordinary Kolka Glacier landslide of 22nd September 2002. There is quite a nice basic description of this event at here. In a nutshell, this was a catastrophic collapse of the Kolka Glacier in the Genaldon Valley, which is located in the Caucasus Mountains in the Republic of North Ossetia, part of the Russian Federation. A catastrophic debris avalanche swept down the valley (Fig. 1) at velocities of up to 65 m/sec. The flow travelled a total of about 19 km, transitioning to a debris flow when it hit a narrowing of the valley. In all, about 125 people were killed.

Figure 1: This pair of ASTER images, from NASA, taken before and after the collapse, shows the vast extent of the disaster. Debris and ice filled the Genaldon Valley from the Kolka Glacier Cirque to the Gates of Karmadon—a distance of about 18 kilometers . (Images by Robert Simmon)

The controversy about this event comes in trying to determine what initiated this massive landslide. Two theories have been postulated:

1. The simple one is that the flow started because a large rockfall detached from the mountain behind and fell onto the glacier. This instantaneously loaded the ice, inducing massive pore pressure increases through the mechanism known as undrained loading. In consequence the resistance to movement rapidly decreased, and the flow started. This is the model proposed by Huggel et al. (2005).
2. The alternative, rather minority view, is that the event was a conventional glacier surge that rapidly developed into a full failure (see Kotlyakov et al. 2004). Whilst glacial surges do occur, the rates are usually much lower than was observed here.
   

Figure 2: NASA image, hosted by Wikimedia, showing the source area of the glacial surge

This paper suggests a third model. Most importantly, they propose that there is no evidence that the flow was initiated by a large rockfall event. The authors have looked at satellite imagery immediately before the event, which suggests that there were smaller scale failures occurring almost continually in the summer months of 2002. This is supported by observations by mountaineers at the site. Perhaps most importantly, they have obtained a Landsat ETM+ image that was collected on 20th September 2002, 8.5 hours before the failure event. Comparison with post event imagery suggests no major differences (Fig. 3), suggesting that a large failure did not occur.

Fig. 3: This is Fig. 4 in Evans et al. 2009, captioned: Fig. 4. (A). Landsat ETM+ satellite image obtained 20 September 2002, 11:31 am (local time); Kolka Glacier (1) is covered by new snow, with a very fresh and large (0.17 km²) debris trail (2). Also note exposed bed (3) of the former hanging glacier that entirely collapsed between 19 August and 20 September 2002, a pronounced shadow (4) indicating a 50-m-high margin of a northward glacier surface rise, and another shadow (5) of a high ice cliff where Kolka Glacier has already started to deform 8.5 h before the catastrophic detachment at about 20:05 h local time. (B) QuickBird image taken on 25 September 2002 (^©2007 Google™, 2008 DigitalGlobe). Note that there is very little difference in morphology of the mountain slope above the Kolka glacier (arrow), compared to the image of 20 September 2002 in (A).

However, it is clear that the glacier itself shows signs of extensive deformation in the pre-failure image. Most notably, they observe that a 50 m high ice cliff had develped on the glacier (at point 4 on Fig. 3A), suggesting that extensive movement was occurring. Evans et al. (2009) therefore suggest that the Kolka Glacier started to deform in response to loading from ice and debris. This disrupted the internal drainage of the glacier, triggering the development of excess water pressures at the base of the ice body, which in turn triggered a catastrophic decrease in effective stress and thus an almost complete loss of frictional resistance at the base of the glacier. As a result the glacier detached from its bed (85-175 m below the surface). As the upper part of the ice slide it loaded the lower portion of the glacier, which also began to move through undrained loading.

The theory is supported by seismic data, which do not show a signal that could be interpreted as the impact of a large rockfall (these are usually picked up by seismic stations).

Analysis
So how likely is this? Well, technically it is undoubtedly feasible. However, two things should be noted. First, the authors have not provided any numerical analysis to support the hypothesis that very high basal fluid pressures can be generated in this way. It would be interesting to try to model this to see if it can happen in reality, although this modelling is far from easy. Second, the analysis is based upon post-event reconstructions and the use of pre-event satellite imagery whose resolution is really not good enough. It is still possible that a rockfall occurred that can't be resolved on the imagery - vertical images with comparatively low resolutions are not good at observing vertical rock faces. It could be that a hybrid model is appropriate - i.e. that the massive glacial deformations observed created the conditions that allowed a much smaller rockfall to initiate failure.

If the authors are right, and they may well be, then this will change the way that we view hazards associated with surging glaciers. Given the changing dynamics of glaciers associated with anthropogenic climate change (global warming) (see here for example), this will become increasingly important.

Main Reference
Stephen G. Evans, Olga V. Tutubalina, Valery N. Drobyshev, Sergey S. Chernomorets, Scott McDougall, Dmitry A. Petrakov, Oldrich Hungr (2009). Catastrophic detachment and high-velocity long-runout flow of Kolka Glacier, Caucasus Mountains, Russia in 2002 Geomorphology, 105 (3-4), 314-321 DOI: 10.1016/j.geomorph.2008.10.008

Other references
C. Huggel, S. Zgraggen-Oswald, W. Haeberli, A. Kaab, A. Polkvoi, I. Galushkin and S.G. Evans, 2005. The 2002 rock/ice avalanche at Kolka/Karmadon, Russian Caucasus: assessment of extraordinary avalanche formation and mobility, and application of QuickBird satellite imagery, Natural Hazards and Earth System Sciences 5 (2005), pp. 173–187. This paper is avilable online for free here.

V.M. Kotlyakov, O.V. Rototaeva and G.A. Nosenko, 2004. The September 2002 Kolka Glacier catastrophe in North Ossetia, Russian Federation: evidence and analysis, Mountain Research and Development 24, pp. 78–83.