17 November 2014
The Markagunt gravity slide
In the current edition of the journal Geology, a paper by David Hacker, Robert Biek and Peter Rowley (Hacker et al. 2014) describes the Markagunt gravity slide in southwest Utah. This is a very exciting piece of work as it identifies for the first time a truly gigantic landslide. Whilst I like to avoid superlatives, the scale of the Markagunt landslide is remarkable:
- 90 km long
- 1700 – 2000 cubic kilometres in volume
- Surface area of over 3400 square kilometres
- Up to 200 m thick.
This is part of the area that it covers:
The landslide deposit is located in the Marysvale volcanic field in Southwest Utah in the USA. This is not the first time that these landslide deposits have been identified, but previous studies have suggested that they were formed in multiple landslide events, and have termed the deposit the Markagunt volcanic breccia. The change in Hacker et al. (2014) is that the deposit is now recognised as having originated in a single landslide that occurred about 22 million years ago. The paper demonstrates that the landslide consists of a large sheet of volcanic rock broken up by faults. The authors divide the landslide deposit into three distinct sections:
- A 58 x 42 km bedding plane segment;
- A 1-2 km wide ramp segment;
- And a 32 km long land surface segment.
The question of course is how such an enormous landslide can form. The base of the Markagunt gravity slide consists of of a clear shear surface with brecciated (i.e. intensely shattered) rocks. However movement has occurred on a shear surface that has an inclination of only a small number of degrees. The authors suggest that as the Maryvale volcanic field developed, it uplifted the Turshar Mountains, generating a slope in the surrounding rocks. At the base of what was to become the landslide is a very weak volcanic deposits known as the Brian Head formation, allowing sliding to develop.
There is only one other known landslide on this scale – the infamous and equally enormous Heart Mountain Gravity Slide. These two deposits are now the largest subaerial landslides on Earth. Both are of course in the USA; I wonder how many more there are around the world that have yet to be identified?
David B. Hacker, Robert F. Biek and Peter D. Rowley 2014. Catastrophic emplacement of the gigantic Markagunt gravity slide, southwest Utah (USA): Implications for hazards associated with sector collapse of volcanic fields. Geology 42, 943-946. DOI: 10.1130/G35896.1
12 November 2014
Over in Norway, the Mannen landslide continues to creep slowly – about 2 mm per day now, as the graph below shows:
- The risk level has been downgraded red to yellow, but this still requires continuous evaluation of monitoring data.;
- The monitoring system has been upgraded to provide better radar systems, the installation of a geophone (that might generate some very interesting scientific data), a LIDAR (laser) monitoring system and enhanced use of weather stations to determine the water inflow to the landslide;
- The railway has reopened to freight traffic only;
- Those evacuated from the site remain out of their houses;
- A decision has been taken (wisely in my view) not to use water-bombing due to the uncertainties
The latest update report from Aknes is also online (PDF again in Norwegian). The most interesting aspect of this report is a map showing the deformation across the hillside, with the monitoring positions marked as well. The map shows deformation over the period 6th to 29th October; I’d think this is slope radar data:
The very high level of deformation at the top of the slope is clearly visible, as are the smaller deformations downslope. This map helps gain an understanding of just what a large block makes up this active landslide. Given the onset of winter conditions the monitoring of the landslide and forecasting its future behaviour are becoming increasingly difficult. whilst the next obvious danger point will be the spring snow melt and thaw season, a period of warmer weather might also pose risks. Of course there is also a chance that the ongoing secondary creep will tip the landslide into a tertiary creep phase, leading to failure, so it is not possible to make any assumptions.
11 November 2014
The Donghekou landslide
One of the largest landslides triggered in the remarkable 2008 Wenchuan earthquake in China occurred at Donghekou in Qingchuan County. This was a large slide – it has a volume of about 100 million cubic metres and it traveled over a distance of about 2 km. Four villages were buried, resulting in about 780 deaths. This landslide has been examined in detail in a paper, Wang et al. (2014) just published in Engineering Geology. This pair of images, whose provenance is somewhat unclear but which was reproduced in another paper on this landslide (Zhou et al. 2013) provides a before and after view of the landslide site:
The paper is interesting in a number of ways. First, it notes that the slope was known to be unstable and indeed that during heavy rainfall events the local population was frequently evacuated. Of course this was not possible when the earthquake struck. Second, the paper notes that the landslide had a very long runout distance over a near horizontal surface, which suggests very high rates of movement. Using experimental data, the authors conclude that the landslide had such a long runout because of liquefaction of the valley fill deposits during the earthquake, which provided a very low friction surface. The landslide itself was retrogressive in nature, which is unsurprising for such a large slide.
Perhaps the most interesting aspect of the landslide though is that about six months after the earthquake fumaroles appeared on the landslide mass, generating high temperature gas and a sulfurous odour. These fumaroles are still active, although less so now than in the early phases. Unsurprisingly there has been considerable speculation as to what these fumaroles represent. The authors both sampled the gases emanating from the fumaroles and measured the temperatures a metre into the vent. They found that the ground temperatures were in the range of 50 to 60 degrees Centigrade and that the gases were primarily carbon dioxide and methane, with fluid emissions being rich in potassium, sodium, magnesium and some other trace elements, From this they concluded that the cause of the fumaroles was oxidation of carbonaceous siliceous shale that had been exposed to the air and oxygen-rich water as a result of the movement of the landslide.
This is of course not without precedent – indeed back in 2008 I wrote a blog post about the strange phenomenon of burning landslides, using an example from Dorset in the UK.
Zhou,J-W., Cui, P. and Yang, Y.G. (2013). Dynamic process analysis for the initiation and movement of the Donghekou landslide-debris flow triggered by the Wenchuan earthquake. Journal of Asian Earth Sciences. 76, 70-84. DOI: 10.1016/j.jseaes.2013.08.007
Wang,G., Huang, R., Lourenço, S.D.N. and Kamai, T. 2014. . A large landslide triggered by the 2008 Wenchuan (M8.0) earthquake in Donghekou area: Phenomena and mechanisms. Engineering Geology. DOI: 10.1016/j.enggeo.2014.07.013
7 November 2014
1. The Mannen landslide
Unfortunately the Mannen landslide in Norway continues to creep without failing. TV2 is heroically maintaining its webcam, and the residents in the valley below remain temporarily homeless. The Mannen Direkte webpage has a new movement graph with data from 1st to 6th November:
The dramatic slowdown in the rate of movement appears to be associated with the arrival of much colder weather. This suggests to me that the landslide is still in a secondary creep phase, meaning that there is deep uncertainty about when the landslide might start to fail.
A fatal landslide in Switzerland
Meanwhile, in Switzerland a landslide on Wednesday night struck a house in Bombinasco in the south of the country. Judging by the image below, the landslide was not large, but unfortunately it destroyed a house, killing a 31 year old woman and her three year old daughter:
3. A landslide in Quebec derails a freight train, leaving a worker missing
Over in Canada a train carrying iron ore was derailed yesterday close to the Moisie River, north of Sept-Iles in Quebec. The landslide appears to have been a joint-controlled rockslide, which forced the locomotives and a number of railcars into the river:
Sadly the driver is unaccounted for. The train had three locomotives, of which only one is visible in the above image. The image below suggests that at least one of the locomotives is submerged:
6 November 2014
Earthquakes and landslides in Peru
Peru is a country that has an abundance of landslides, resulting from a combination of high rates of tectonic uplift (which creates mountains that can be eroded), frequent earthquakes and regular intense rainfall events (often linked to El Nino episodes). It has been known for a while that the rate of landsliding in Peru has changed dramatically in the past, and it has been widely hypothesised that this might have been linked to periods of more intense precipitation associated with phases in which El Nino events were larger and/or more frequent. Clearly this is interesting not just from a scientific perspective but also because it might provide insights into the role if future climate change in generating landslide hazards.
In a paper just published in Nature Geoscience, McPhilips et al. (2014) have explored this hypothesis by examining cobbles located in the Quebrada Veladera river channel and an associated fill terrace. The have used 10Be concentrations in individual cobbles to determine erosion rates for the catchment. The modern cobbles in the river channel give an indication of the current erosion rate, which is occurring when the climate is very arid (less than 200 mm rainfall per year on average, whereas the terrace is dated from 16,000 years BP, when the climate was sufficiently wet to maintain a 60,000 square kilometre lake on the altiplano.
The results are quite surprising. In the words of the abstract:
The distribution of 10Be concentrations in terrace cobbles produced during the relatively wet climate before about 16,000 years ago is indistinguishable from the distribution in river channel cobbles produced during the drier climate of the past few thousand years. This suggests that the amount of erosion from landslides has not changed in response to climatic changes.
In other words, it does not appear to be the case that during this phase of higher rainfall in Peru the occurrence of landslides increased. This is counter-intuitive in many ways, and it is not at all clear to me as to why this should be the case. The authors hypothesise that the area might be so arid that even wet phases do not generate sufficiently high pore pressures to trigger landslides, but this does not seem to agree with what our modern day obserations . If the paper is right, then an alternative trigger must be responsible for the landslides that are recorded in the sediments in the river channels. McPhilips et al. (2014) suggest that these might have been triggered by large earthquakes. It has been hypothesised previously that in very dry environments with high uplift rates earthquakes must do most of the work to trigger landslides, whereas in equivalent wet climates rainfall may be the primary factor. This is one of the first studies to provide field data in support of this idea.
The implication of course is that a large earthquake in this area would be devastating in terms of landslides, and the 2010 earthquake event may support this idea. Given that the return period for great earthquakes in Peru is about 100 years, this is a sobering thought.
4 November 2014
The Mount Mannen rockslide
The Mount Mannen rockslide in Norway is continuing to behave in an unpredictable manner, no doubt to the intense frustration of all involved. The tv2 website (in Norwegian, though Google Translate does a good job), which has the live webcam, has produced a series of graphs showing the movement of the landslide. This one, complete up to yesterday, shows the cumulative displacement of the landslide over the last month:
The lines show different parts of the monitored slope of the Mount Mannen rockslide. Points 1 and 2 in the upper part of the active slide, points 3 and 4 are in the lower section of the active block. So, it appears that the upper part of the landslide is still moving quickly (about 15 mm per day) and has now displaced about 24 cm in a month, whilst the lower part has moved about 10 cm. The difference between the two is not unusual or surprising for this type of landslide. The behaviour on about 30th October is interesting as the landslide appears to have slowed dramatically in the upper portion and actually stopped briefly lower down, before starting to move again. We see similar behaviour in lab tests that simulate creep movements, although we cannot fully explain these episodes. Whilst the movement record is quite noisy (which is unsurprising for an area that receives regular snowfall), the medium term trend is still an accelerated creep movement pattern, and my view would be that the slightly longer-term rate of creep is still increasing, with variations caused by changes in the environmental forcing (i.e. temperature and rainfall).
In other words the authorities are correct to maintain the evacuation. I remain skeptical that much can be done to speed up this natural process. Although it is frustrating, nature probably needs to be allowed to take its course.
31 October 2014
The Mannen rockslide – an update
The Mannen rockslide in Norway continues to confound predictions of its imminent demise, and as of this morning it remains intact. A press conference was held yesterday morning in which it was reported that the rate of movement had declined from over 4 cm per day to about 1.5 cm / day. Some rockfall activity continues to occur, but the major collapse is still some indeterminate time away. The live web cams continue to broadcast from the Mannen rockslide site, although the public comments express an increasing level of frustration amongst those taking an interest.
.The scientific effort for the Mannen rockslide is being led by Lars Harald Blikra of NGU. He is vastly experienced in rockslope monitoring – there is no-one better – and his team are using appropriate techniques to monitor the slide. However, the processes of detachment of large rock blocks are extremely complex and inadequately understood, so forecasting (and predicting) a collapse event is very challenging. The recent snowfall will not have helped the efforts to monitor the landslide, given that radar is a key technique being deployed. This graph, from Nyheter, shows the movement of the landslide over a 24 hour period earlier this week:
I suspect that the emergency management people are now in a very challenging situation. The continued movement of the slope will be edging it closer to failure, but it is not clear when this might occur. They will be anxious to reopen the railway line and to lift the evacuation order, but very cautious about the ongoing risks from the landslide. I am sure that they are hoping that it will just fail. I suspect that they will come under increasing pressure to try to fail the slope artificially – water bombers are being mentioned – but it will be very difficult to get enough water in the right place for this to be effective. Explosives could also be used, but emplacing the charges would carry a very high level of risk. Thus, this may continue to be a waiting game for the time being. I have no doubt that Lars and his colleagues are under great stress.
The currently active block is part of a much larger landslide at Mannen. This image from Nyheter shows the multiple fractures at the top of the slope:
This whole slope is moving, albeit at a much lower rate than the active block. Whilst there is no suggestion that the main slide will fail, the active deformation will have probably created a much more fragmented rock mass below the active slide. The failure of the active block will almost certainly cause more detachment of material downslope. That the downslope material is likely to be fragmented, and thus more easily entrained, explains the estimates that the final mobile volume may be in the order of 2 million cubic metres.
29 October 2014
Mount Mannen is located in Romsdal in the northernmost part of western Norway; Alesund is the nearest large city. Over the last few days much of Norway has been watching with fascination to see if a large rockslide will occur on the flanks of the mountain. There is sufficient interest in this event that there are live webcams broadcasting the events on the slope. The potential failure is in the order of 120,000 cubic metres, so the final collapse, should it occur, will be quite spectacular.
The issue is a large, actively deforming block high on the mountainside. This is being monitored in detail, and in recent days the block has started to accelerate, with further very heavy rainfall was due overnight (although no failure appears to have occurred). The VG.no website showed this graphic with apparent displacement data:
Whilst in one of the videos on the above webpage a plot was shown with data extending over a longer time frame (apologies for the poor quality of this screenshot)::
This montage of photos in a VG.no article rather nicely shows the block that is causing concern – and in particular the second from right images shows the displacement across the crown of the landslide:
There is also a very nice helicopter video of the slope, which shows why it is causing such concern.
There are very few articles in English on this landslide, but this one from yesterday provides an overview of the current situation. Note that the article was speculating that the failure would occur yesterday though, which did not occur. A more recent article describes quite well the profound difficulties of trying predict reliably these final collapse events, especially where the rockmass is highly fractured and disrupted. There is little doubt that this block will ultimately fail, but whether it will be in the current movement episode is hard to tell. Unfortunately of course if movement of the block slows down then the authorities will have major problems in managing the long term hazard. This is an event that is worth watching. Fortunately, it appears to be a nice clear day on Mount Mannen today, so the webcams have an excellent view of the slope.
23 October 2014
Riverbank collapse is a type of landslide that has probably been under-researched in recent years. Whilst such events cause few fatalities in the global scheme, the impact on infrastructure and property can be serious. A fascinating case study lies in the Mekong River – a problem that is so serious that in 2009 the Prime Minister of Cambodia warned about the threats of the “seasonal collapse of riverbanks”. Earlier this year a woman and two young children were killed by riverbank collapse in Kandal Province in Cambodia. This image, from the same article, is a nice illustration of the problem:
Over in Australia there is a very nice project in progress, in which I have some involvement, looking at riverbank collapse along the Murray River, where collapsing riverbanks were a major problem when the river level was very low in the drought of a few years ago. There is a very nice e-poster by the research team online explaining the provisional results of this work – NB this is a pdf.
Whilst the mechanisms of riverbank collapse might initially seem to be simple, they are in reality extremely complex, not least because the stress state is very dynamic due to the effects of water coming into the bank from the river, from the bank top and through the ground. In many rivers water levels rise and fall quickly, causing unusually rapid pore water pressure changes. And of course humans have an annoying habit of messing around with rivers, and riverbanks, too.
Anyway, there is an interesting video on Youtube, which I don’t think I’ve seen before, showing a riverbank collapse event. I am not sure where this happened.
Clearly the river is in flood, but the complexity of the failure, and its progressive nature, illustrates how complex the mechanics can be.
22 October 2014
The Bukit Beruntung landslide
Bukit Beruntung is a relatively modern residential development in Selangor, Malaysia. On Monday morning heavy rainfall triggered movement in a slope behind some of the apartments, leading to the evacuation of over 2000 residents. Two days on, over 500 people are still out of their homes, although movement of the slop[e has slowed down. This is not a small slope failure - this image, from the Star, shows the slope from the crown area:
Of note here if the large displacement at the crown of the slope and the deep red colour of the soil, which indicates that it is highly weathered – typical for a tropical setting. A likely explanation for this slope failure is revealed by images of the toe of the slope, which show a heavily deforming retaining wall:
The deformation in the asphalt in front of the wall is interesting. Whilst this might indicate that the failure is passing beneath the wall, I think it is more likely that the wall is being shunted forward by the landslide, causing this deformation. Whilst in all probability this slope can be easily fixed, it is not going to be a trivial task.