29 July 2014
Oso landslide analyses
The journal Science is reporting an interesting public argument about the mechanisms of the Oso landslide in Washington State, USA. This is between the authors of the GEER report (NB pdf), which was released last week, and Dr Richard Iverson of the USGS, and is in essence about the chronology of events that generated the very rapid debris flow event that was so costly. I wrote about the GEER report last week, and noted that they interpret the sequence as being a two-fold failure event, as indicated by the seismic data. The latter is not controversial – there is a nice (pdf) report from the Pacific Northwest Seismic Network that shows this double event very clearly:
In the GEER interpretation of the Oso landslide the lower slope failed first to generate the rapid flow event, followed by the second failure, which was in effect a retrogression of the landslide as a new block slid onto the main mass, and then stalled. This is consistent with the fact that the first event generated a much larger seismic signal; and the fact that the, second seismic signal appears to terminate quickly.
However, in the Science report, Dr Iverson disagrees with this interpretation of the Oso landslide, based on a more detailed analysis of the seismic signals The report notes that:
“But Richard Iverson, a landslide expert at USGS’s Cascades Volcano Observatory in Vancouver, Washington, says that a closer look at the seismic data tells a different story. “The USGS disagree[s] significantly with several aspects of the GEER report,” Iverson says. According to the unpublished analysis by scientists with the USGS and the University of Washington, all the important action appears to have been compressed into a rapid chain reaction in the first few minutes. The lower slope began to slide slowly, but just 50 seconds after that began, an upper part of the mountain broke loose and collapsed onto it. The violence of that impact could cause the soil to rapidly liquefy and shoot across the valley, much like a foot slamming down into a mud puddle, he said. Iverson and scientists working with him concluded the second incident recorded in the seismic data was an ‘extremely small event.’”
I have no experience of interpreting seismic signals from landslides, so struggle to comment either way on this interpretation. I think I can see the essence of this interpretation in the data from the more distant seismic stations:
To my untrained eye the initial seismic event shows a slow increase in energy release rather than an abrupt peak. Note also that the second event appears to be much smaller than the first. Thus, in this interpretation, the initial failure was, I think, a small, slow failure in the lower slope that destabilised the upper block of the Oso landslide, which slid onto the mass below. This then generated massive undrained loading, allowing the generation of very high pore water pressures, and hence the high mobility flow. The second seismic event is interpreted as being “extremely small”.
I think one can take issue with the latter statement – if it was extremely small it would not have generated a seismic signal, surely – but the point is that it is relatively small when compared with the first failure. The Science have clearly put this alternative interpretation to the GEER team, and report as follows:
Responding to Iverson’s critique, GEER co-leader Jeffrey Keaton, an engineer at AMEC Americas, a private engineering firm, says that observations in Oso helped convince them that much of the mountain collapsed a few minutes after the initial slide. Large chunks of relatively intact earth still covered with trees would probably have broken up if they had been part of the first slide, he said. And swaths of sand had flowed up onto the back end of the first slide, suggesting it happened after the first slide slowed to a halt.
The data that is available is insufficient to be able to make a definitive argument either way on this – the next key step will be publication of the UGSG analysis that supports their line of argument. In essence the issue is the chronology of events shown in this Seattle Time image of the landslide:
At the end of the day, this comes down to whether the most likely scenario is that the block that was formed from material marked B above failed (event 1), allowing block A to slide in (event 2); or whether block A failed, causing massive collapse of block B (event 1) followed by a small but at present unknown slide that generated event 2. I would guess that a very detailed analysis of the deformation at the front end of block A might help – if this block moved after Block B had failed then there should be some impressive compressional deformation features preserved in this area. Maybe the very high quality Google Earth imagery collected on 1st April would allow someone with the right skills to make an interpretation?
Personally I can see merits in both interpretations, so will watch with great interest as this is resolved.
26 July 2014
China Development Gateway has some interesting images of the aftermath of a landslide that occurred on 17th July at Gedi Village, located in Muhuang Township within Yinjiang Tu and Miao Autonomous County, in Guizhou Province, China. The landslide has reportedly damaged 152 buildings and caused the evacuation of 275 people:
Note the much more disrupted component of the landslide at the front on the left side. In the main slide, even the apparently intact buildings have been rendered uninhabitable:
My interpretation is that this is a deep-seated landslide because the lateral scarps are very well developed and show considerable vertical extent:
There can be very little prospect that this landslide could be stabilised at a reasonable cost, and sadly the village is already an almost complete loss. The causes of this landslide are not clear to me, but of course this is the rainy season in China.
Nagano debris flow
Meanwhile, an interesting video has appeared of the aftermath of the debris flow at Nagiso in Nagano Province of Japan, which killed a child on 8th July. The video should be visible below (if the embed works!) – if not you can watch it here.
The video starts with the CCTV footage of the front of the debris flow. Probably the most interesting aspect of this is the aerial view of the damaged area from the flow:
The area of devastation lies immediately downstream of a bend in the existing channel. Until that point the landslide had remained mostly within its banks, but at that point it left the channel and ran through the town. The terrible impacts on the buildings in this area are clear.
24 July 2014
A very large an interesting landslide occurred in Iceland on the night of 22nd/23rd July in the flanks of the Askja stratovolcano in Iceland. This is a very interesting event in a number of ways, not least because the volume appears to be large – estimates at present range from about 24 million cubic metres to about 60 million cubic metres.
Images are appearing of the landslide, which is impressive in its scale:
The landslide is estimate to be about a kilometre in width. There is a nice video taken from a flight over the landslide on the RUV website too. The landslide entered the lake at the foot of the slope, generating very large tsunami type waves. A credible witness, Ármann Höskuldsson, who was in the area with a group of students, estimates that the waves were 50 m high.
The cause of the landslide is not clear at present. Some of the articles suggest snow/ice melt, but there is no evidence to support the hypothesis. The slopes on the southern edge of the Askja massif are steep. This Google Earth image of what I think is the site suggests that there may have been previous large-scale landslides on this slope:
There is also evidence of large amounts of erosion, suggesting that the slope may have been steadily destabilising with time. It is entirely possible that this is a progressive landslide with no trigger event.
If the waves were 50 m high then the level of erosion around the lake should be extremely high given the weak materials. I have yet to see any good images of the scour around the lake. This must be a golden opportunity to understand better the generation of tsunami waves by rapid, large landslides.
23 July 2014
Oso landslide – the last set of remains
Two significant events occurred yesterday with respect to the Oso landslide in Washington State, First, it was announced that the last set of human remains were recovered from the site. In the tragic circumstances of the landslide it is an extraordinary achievement to have recovered all of the remains, given the size and mobility of the slide. The search and rescue teams at Oso deserve great credit for what they have achieved. In case you haven’t seen it, the leader of the rescue effort posted a comment on one of my earlier posts, providing insight into how this was achieved:
22 July 2014
Last Tuesday a landslide at Erzurum in Turkey destroyed an almost new, and extremely expensive, ski jumping facility. The ski jumps were constructed for the 2011 Winter Universiade, at a reported cost of 20 million Euros. The lower part of the Kiremitliktepe ski jumps collapsed. Three of the jumps have been completely destroyed (image from here), whilst the two larger jumps have been severely damaged:
Pleasingly, the collapse event was in part captured on video and is buried on Youtube:
Whilst this video provides a pretty good overview of the aftermath of the landslide
There is also an excellent gallery of images of the collapse in motion here.
This is a Google Earth perspective view of the site, taken in 2012:
Compare this with the image below of the slope, taken in 2009:
It is hard to read the topography from these images given the quality of the digital elevation model, but I would make a few initial observations. First, the topographic shape of the slope appears to have been modified – in particular, there may have been some excavation of the toe to create space for the landsliding and runout zones. The main part of the failure seems to follow the excavated and modified area for the smaller jumps on the right hand side. Second, there has been considerable additional material placed on the middle part of the slope for the main jumps (on the left side), which may have added weight to the unstable mass. And third, there is a new lake at the top of the slope, presumably to provide water for the snow-making machines.
An article in a Turkish newspaper yesterday also suggests that there were major construction defects at the site:
Prosecutors have initiated a fact-finding mission and launched an investigation into last week’s collapse. The team includes four construction and geology engineers from Atatürk University in Erzurum. The mission’s initial report has revealed some fatal problems in the construction of the towers that supported the ski jumps. The expert report shows how poorly the towers were built. An expert at the site in Erzurum found that the contractor used only one meter of steel piles in the towers’ foundations, Turkish media outlets reported on Sunday. The reports said that Sarıdağlar was required to sink 50-meter-long steel piles into the ground to support the towers. Taking a deadly risk, the contractor first poured concrete into the tower foundation then sank only a one-meter-long pile in it, a critical fault. Construction experts said, according to global standards, steel piles have to be at least 25 meters deep for towers this high. Even worse, the contractor incorrectly calculated the angle of the slopes on which the ski jumps were placed.
Thus, overall this is looking like this could be a major design and construction failure, although we’ll need to see what the official report states in due course. Given the above, the likelihood of being able to reinstate the facility without complete demolition and reconstruction look slim.
21 July 2014
Maoxian County landslide
On Thursday afternoon a rockslide occurred on a road in Maoxian County in Sichuan Province of China. Unfortunately, Xinhua reports that the landslide struck nine vehicles, including seven cars, killing 11 people and seriously injuring a further 19. Amazingly, the rockslide was captured on the dashboard camera of a vehicle close to the landslide. Indeed, this vehicle was itself struck by a rock. The footage is available on youtube. A word of warning – this does not make easy viewing:
There are several things that are worth noting here. The first is the extraordinary speed with which the blocks, some of which are very large, come down the slope. I think that one intuitively knows that this is the case, but the actual velocities are still shocking. These lumps of rock are very heavy, so the destructive power is terrible. The second is the absolute terror of those people caught in the rockslide. The rates of movement of these rocks are so high that dodging them is exceptionally difficult, especially as the trajectories that they follow are unpredictable. The man who went into the gutter adopted a good strategy for these large blocks, but of course if the whole slope had collapsed he would have been killed. And the third of course is the person who was struck by the bouncing rock. This screenshot captures both the victim (circled by the video maker) and the lump of rock (above the red car):
This was a large boulder that was travelling very fast despite the bounce that it took above the road; I suspect that the victim must have been seriously injured, although he or she does appear to be walking at least when helped to safety.
Late in the video the man in the gutter also had a lucky escape when he started to move just as another boulder struck the top of the retaining wall just along from him, as this screenshot shows:
The video was taken in an area in which comparatively few boulders were falling. The situation further along the road in the main rockfall zone must have been desperate.
15 July 2014
Over the weekend I travelled up to the Lofoten Islands, north of the Arctic Circle in Norway, to see the midnight sun. We were lucky to have quite wonderful weather, which provided the perfect opportunity to enjoy this most beautiful place:
The topography of the islands is exceptionally steep and rugged, which means that landslides and rockfalls are common. This is a very impressive ancient ridgeline rockslide- note the complete loss of a section of the ridge, the long runout of the deposit and the very large boulder sizes:
On a completely different scale, small-scale failures that have transitioned into minor debris flows are also common. This is a very nice example, with just a very small initial scar but a long track:
Sheet (face-parallel) jointing in the rock masses is also very common, which creates release surfaces for rockfalls. This is an example above the main road in which rockfall barriers have had to be installed:
The rockfalls and landslides inevitably cause some disruption. At the southwest end of the chain there is a huge project ongoing to build rockfall shelters along the road:
The Lofoten islands are one of the most beautiful, interesting places I have visited. I strongly recommend them for a weekend, or for a longer stay!
10 July 2014
Xinhua reports that the rescue operation for the Shawa mudslide in Fugong County, Yunnan Province in China has now been completed, with 17 people killed and one injured. Details of the slide are sketchy, but it does appear to have been triggered by monsoon rainfall. The reports about it are somewhat contradictory, with some appearing to emphasise that the landslide was “natural”, whilst others note that it destroyed a silicon mine. Several pictures of the landslide are now available, but they really only cover the lower portions of the slide. This is probably the best overview:
Xinhua has two albums of images (here and here), but as usual they are of the “interesting views of the heroic rescuers” genre rather than much of interest. This image appears to show the landslide deposit and the very destructive nature of the landslide:
Minzhu mudslide in Yunlong County
Xinhua also reports that there was another landslide at Minxhu, a village in Yunlong County, Yunnan province. Although much less well reported, this also appears to be been a very destructive event, with the toll likely to be 14 people, consisting of six confirmed fatalities and eight people missing.
9 July 2014
Shawa village mudslide
The landslide season across Asia has got off to an unusually slow start, with the number of events being notably lower than is usually the case in July. This may well be associated with the potential development of El Nino conditions, which sometimes suppresses both the SW monsoon across South Asia and the occurrence of tropical cyclones in the North-West Pacific. However, the signs are that the true landslide season may be about to start. The enormous Typhoon Neoguri will strike Japan over the next couple of days, bringing very heavy rainfall. Some landslides are inevitable, though the high rate of movement of the storm may prevent the very large rainfall totals that these events sometimes bring.
In China, reports today suggest that a large mudslide has occurred in Shawa village, which is located in Fugong County of Yunnan province. Xinhua has a brief report, which indicates that 17 people are missing. The caveat here of course that there must be some uncertainty about that total.
I haven’t been able to find Shawa village, but this is a Google Earth image of Fugong, the county town:
This is the sort of landscape in which landslides are an inevitable and frequent process.
8 July 2014
Komansu rock avalanche
The Komansu rock avalanche is described in a new paper by Robinson et al. (2014), which has recently been published online by the journal Landslides. The landslide is located in the Alai Valley of Kyrgyzstan, in the Trans Altai range of the Pamir Mountains. The landslide deposit is ancient – the slide has been dated at 5,000 to 11,000 years BP, which is the period after the retreat of the glaciers in this region.
The Komansu rock avalanche is both very large and quite hard to see. This is a perspective view Google Earth image of the landslide – I’ve annotated the key parts of the slide, based on the interpretation in Robinson et al. (2014):
The current landslide deposit has an estimated volume of 3 – 5 cubic kilometres (although a large part of the deposit has been eroded away), and covers an area of 64 square kilometres – i.e. this is a very large landslide indeed, and we’d expect such a large landslide in very steep, high terrain to have a long run-out. Empirical relationships from similar events suggest about 10 km. However, the Komansu rock avalanche has a run-out that is much longer than this – about 26 km – which asks real questions about the dynamics of the landslide. In other words, what was special about this landslide that allowed it to go so far? An obvious explanation is that the landslide volume was much larger than the modern deposit suggests, but given that the source area only has a volume of about 4 cubic kilometres, this then poses the question as to the origin of the additional material.
Robinson et al. (2014) have compared the hummocky topography of the landslide deposit with that of other large landslides from this region. This is a Google Earth perspective view of the deposit:
Whilst on the face of it the deposit might look a little strange, it is very similar to that found in many other large rock avalanches both in this area and more widely. From this perspective, the landslide is not unusual. The one mechanism that is known to generate usually long runouts is the presence of large amounts of snow and ice – i.e. that is a so-called rock-ice avalanche. However, such landslides tend to have a distinctive final morphology; Robinson et al. (2014) suggest that the Komansu rock avalanche deposit does not match this. An alternative, but similar possibility is that the landslide ran-out over a glacier within the valley, but again the morphology does not seem to match that observed from other known examples.
Robinson et al. (2014) suggest that the unusual mobility might be explained by the presence of a large amount of ice in the initial failure, and that the landslide then entrained a large volume of material from within the valley. So, the authors propose that the initial failure was about 4 cubic kilometres of rock together with about 500 million cubic metres of ice. This very large landslide then entrained about 4 cubic kilometres of sediment in the valley, plus additional glacial ice, to generate a flow that behaved in a manner that was similar to a volcanic debris flow, allowing the very long runout distance.
A final observation in the paper is that such landslides have substantial implications for hazard assessment in high mountain areas. These giant rock avalanches with long runout distances are clearly extremely destructive. As people increasingly populate the high mountains the likelihood of a mass fatality event, especially during a very large earthquake, increases.
Robinson, T.R., Davies, T.R.H., Reznichenko, N.V. and De Pascale, G.P. (2014). The extremely long-runout Komansu rock avalanche in the Trans Alai range, Pamir Mountains, southern Kyrgyzstan. Landslides, DOI: 10.1007/s10346-014-0492-y