The Frank Landslide, 29th April 1903

As today is the 105th anniversary of the Frank Landslide in Canada, it seems appropriate to revisit this most interesting event. The town of Frank is located at about 49 degrees 35 minutes North and 114 degrees 24 minutes West at an elevation of about 1300 m in Alberta, Canada. The town on the back of coal deposits at the foot of Turtle Mountain to the west of the town (Figure 1). These were exploited by coal mines at the foot of the hill.

Figure 1: Google Earth image of the Frank landslide, Turtle Mountain, Alberta, Canada.

The landslide is fairly easy to see on Figure 1. The first signs of problems came in the weeks leading up to the collapse, when the coal tunnels started to show signs of movement. The final collapse occurred at 4:10 am when a block of about 650 m in height, 900 m in width and with a maximum thickness of about 150 m thick broke off the hillside and thundered into the valley below. The failure, which had an estimated volume of 30 million cubic metres, apparently took about 100 seconds, killing 76 people. Most of the victims were killed in their beds and their bodies could not be recovered as they were buried to a depth of about 30 m. Figure 2, from Natural Resources Canada, captures the scale of the event quite well. The landslide deposit extended over about 3 sq km, blocking the river and flooding about 2 km of the railway line. Seventeen coal miners were buried in the coal mine by the landslide but were able to dig themselves out over a 14 hour period.

Figure 2: Natural Resources Canada photograph of the Frank landslide, Turtle Mountain, Alberta, Canada.

The Frank landslide remains of great interest for a number of reasons:

1. There has been considerable controversy over the cause of the landslide. Initially the blame was pinned squarely on the mining activities, but more recently it has been increasingly accepted that in fact the geological structure, which was unfavourable for stability, was probably the primary cause. The mining probably did not help, and other factors such as rainfall may have also had a role.
2. There has also been a great deal of speculation over the the rate at which the landslide debris spread itself over such a large area. This very rapid runout behaviour is a somewhat enigmatic process in many large rockslides. Theories have abounded over the role of air entrainment, reduced basal friction, etc. The jury is still out - this remains one of the great questions in landslide science.
3. Parts of the mountain are probably still potentially unstable. In 1911 part of the town was relocated over fears about the stability of the slope. Today, real time monitoring is undertaken by the Alberta Geological Survey using a wide range of sensors. The aim is not to prevent a failure event but to provide a warning that it might occur. This is fine for a progressive failure or for one triggered by rainfall, but it offers little if there should be a large earthquake. Nonetheless, this is important work that is undertaken to a world leading standard. The work is described exceptionally well here
   

Updated: 25th April: The Kaiapit Landslide, Papua New Guinea, 1988

Update: Mark Drechsler has provided some images of the Kaiapit landslide, which are now integrated into the text below. Thanks to Mark for this - it is much appreciated. All images copyright of Mark Drechsler. In addition, I have placed a copy of the Drechsler et al (1989) paper on the International Landslide Centre website here. It provides a fascinating level of detail about the slide. I recommend this paper over and above the alternative.

Click on the images for a full-sized view - it is well worth it!

Original text: Thanks for Mark Drechsler of Parsons Brinkerhoff Australia and the University of Adelaide for reminding me of the Kaiapit Landslide, which is surely one of the largest landslides of the last two decades (can anyone think of any larger?). Even though its impact was not enormous (74 deaths in three villages) the scale of this landslide is staggering. The landslide was described in a paper by Dreschler et al. (1989) and one by Peart (1991), from which these details are taken.

The slope failed at 10:43 am on 6th September 1988 when approximately 1.8 cubic kilometres of mountian collapsed, leaving a scar 1600 m high. The debris ran out over a distance of 6.5 km, covering an area of over 11 square kilometres. The estimated velocity is up to 180 km/hour. The slide inundated three villages and formed four valley-blockage lakes.

What is particularly interesting about this slide is the lack of an obvious trigger. Peart (1991) reported that the time of initiation was not the rainy season (indeed it was towards the end of the dry season) and that the period in question was not particularly wet. No seismic activity was recorded. The area has not been glaciated, so over-steepening and loss of support cannot be a cause. Clearly the area is affected by high levels of uplift, which may have allowed the formation of over-steepened slopes. These are probably a causal factor, but do not explain the triggering.

Peart (1991) sagely suggested that there may have been a time-related factor in the initiation of the landslide. I am convinced that time-dependent behaviour is important for large-scale landslide triggering, which means that the time of failure might lag significantly behind the trigger event. This is not always the case, but quite often very large landslides do not have an obvious trigger event. Other examples include the Mount Cook failure in New Zealand and the Leyte landslide in the Philippines.

Sadly the Google Earth imagery of the slide site is covered by clouds. Comments are welcome as ever.

Reference
Drechsler, M. Ripper, I., Rooke, E. and Warren, E. 1989. The Kaiapit Landslide, Papua New Guinea. In: Engineering Geology in Tropical Terrains, Universiti Kebaangsan Malaysia.
Peart, M. 1991. The Kaiapit landslide: events and mechanisms. Quarterly Journal of Engineering Geology, 24, 399-411.

Hat-tip: Mark Drechsler from Parsons Brinckerhoff and University of Adelaide.

Cayton Bay Landslide

It is not often that I get to write about a landslide that is (practically) in my own backyard, but this is one of those rare occasions. In fact for the last fortnight or so this landslide has been generating quite a lot of local news - see for example a report in the Daily Telegraph.

The landslide in question is in Cayton Bay in North Yorkshire. Here, a set of cottages have been built close to the edge of a coastal cliff. Over this winter the cliff has suffered a set of landslides that have caused a series of reasonable large slips that have allowed the cliff to retreat.

This Google Earth image shows the situation quite well.

Google Earth image of the Cayton Bay site

The threatened cottages are in the north-west corner of the images. The landslide is the heavily wooded area between the cottages and the sea. Unfortunately as the image from the Daily Telegraph shows, since this image has been collected the cliff has retreated rapidly due to reactivation of the landslide - estimates are that the cliff top has moved back 7 m this winter - and now at least some of the houses are seriously threatened. Reports indicate that at least two have had to be demolished to date. Given that insurance rarely covers landslide hazards, this must be heart-breaking for the residents.

Inevitably, questions are being asked as to what is going on at this site. Inevitably there is a great deal of speculation, including suggestions that the site has been affected by the building of a new bypass or that the construction of extensions to the bungalows has triggered failure. In my experience such causes are unlikely. The cause is probably rather more local. It is clear that the the coastal slope here is clearly a part of a large, rotational landslide complex which has shown activity before. This complex has been documented and indeed mapped well-before the most recent failures - the BBC for example notes that observations of instability were made by highway engineers in the late 1960's. Importantly, in May 2004 Jon Carey, Paul Fish and Roger Moore from Halcrow gave a presentation at a meeting of the Yorkshire Geological Society entitled "LANDSLIDE GEOMORPHOLOGY OF CAYTON BAY, NORTH YORKSHIRE". The abstract of the paper says:
This paper describes the geomorphology of a large coastal landslide complex at Cayton Bay, North Yorkshire. The area inland of the landslide is occupied by a strategically important road and a number of properties, and knowledge of current landslide behaviour and possible future scenarios is therefore important for future planning and risk. Instability at the site is associated with a series of faults which bring argillaceous Upper Jurassic rocks to sea-level. These soft rocks are overlain by more resistant sandstones. The sequence is capped by a thick and variable series of glacial sediments, that comprise tills with inter-bedded sand and gravel lenses, deposited during the Dimlington Stadial of the Late Devensian. In connection with the development of a future coastal strategy for Cayton Bay, detailed geomorphological field mapping was conducted which identified two major landslide systems. These include a periodically active mudslide complex at Cayton Cliff, recognised by a series of shallow scarps and benches with occasional back-tilted blocks, and an area of dormant deep-seated landslides at Tenants’ Cliffs, that includes a series of graben and horst structures. The origins of the landslides are unclear, but probably involved a variety of processes that led to a reduction in material shear strength or increases in pore water pressures. The timing of original failure may relate to deglaciation following the Dimlington Stadial, or periods of wet climate in the Holocene. Since sea-levels were not higher than present in the Holocene along this stretch of the coast, coastal erosion is not thought to have been a factor. The causes of the contemporary instability are likely to be due to the combined effects of coastal erosion and of groundwater, both of which are predicted to increase in future years due to the impacts of climate change. The implications include increasing risks to coastal assets and a need to manage and mitigate such risks.

This team from Halcrow are pretty competent, so I would trust their interpretation. Their view that activity of the landslide is probably associated with high pore pressures is probably correct, and it is notable that the weather has been cool and wet for the last 18 months or so in this area. The key question that needs resolving is where the water is coming from - is it just that pore pressures are higher than usual (this is very possible) or could there be some other source, such as leaking pipes or changed drainage? If the cause is just naturally high ground water then the long term implications for the cottages is potentially serious. The site is very large, which means that the cost of stabilisation are very high. The land is a Site of Special Scientific Interest, which means that it is protected, and it is owned by the National Trust, who are keen to protect the environment.

EGU Day 4

I spent the morning in the session on the Characterization, monitoring and early warning related to large landslides. I was struck during the presentations by the degree to which the technologies for monitoring landslides have improved over the last decade or so. For example, Casagli and his colleagues gave a very polished presentation on the application of their ground-based radar LISA to the monitoring of the Ruinon landslide near to Bormio in Italy. This system, which costs about 100,000 Euros per annum to buy and operate, can produce a map of deformation with a 1 mm movement resolution at ranges of up to 1.8 km. In this case it is being used to monitor the movement of this somewhat hazardous slope failure to great effect. On the other hand, David Toll (a colleague from Engineering at Durham) presented the results of a new set of high precision tensiometers that are capable of measuring suction forces down to -2 MPa, a significant step forward for tropical soils for example. Thoeny and his collegaues introduced us to the use of dynamic fluid electric conductivity logging, which looks at changes in the electric consuductivity of saline water introduced into a deep boreholed drilled through the landslide. This technique was able to determine the locations at which water is entering and leaving the borehole, which gives a good indication of the location of, for example, the shear surface. This looks like a complex technique, but one that shows a great deal of promise once fully developed. The problem would seem to be the rather cumbersome and specialised processes involved in saturating the hole with saline water. I will watch this development with interest. Glimsdal et al. presented the results of tsunami modelling for the same slide as Thoeny, the Aknes landslide in Norway. The danger is that a large failure would trigger a tsunami that would rapidly inundate local villages. The paper introduced the initial results of modelling a potential tsunami. This is clearly work in progress, but looks interesting.

To me the key issue was highlighted by the paper of Ronchetti and colleagues, who attempted to analyse the data obtained from monitoring of a very large earthflow on Mount Modino near to Modena in Italy. They attempted to use the so-called Saito approach, which I have worked on extensively, to see whether prediction of the failure of the landslide was possible. The paper itself was interesting and there was some sign that movement could be used in this way. However, the data proved to be very noisy, such that identifying trends in real time would be an awesome challenge. This issue that this highlights is that we desperately need to develop better ways to analyse the data that wed are producing. At this meeting it has been clear that analysis is lagging well behind data production. This is surely the next challenge. I have been working with three first class PhD students - Angel Ng, Jon Carey and Chris Massey, all of whom are making really substantial progress in this area for landslide movement at least. It is clear that this is a fertile area for development and that considerable more work is needed.

In the afternoon, I attended the sessions on Rockfalls and large catastrophic landslides. This was a bizarre session in that the quality of the science presented was exceptionally variable. I won't go into detail regarding the weaker end of the spectrum. At the strong end, I was taken by a couple of presentations in particular. Pichler and his colleagues presented a very comprehensive and detailed analysis of the threat to hydrocarbon pipelines associated with rockfalls. This included dropping large boulders 20 m from a crane onto a gravel pit to see what happened, wich looked great fun at least! The result of the subsequent analysis was a clear demonstration that a 3 m thick layer of gravel placed over the pipeline is enoough to protect it against a 10 tonne boulder falling from 100 m, which would seem to be a rather valuable thing to know! Meanwhile Sausgruber presented a great paper on a massive (1.3 cubic km) ancient kink band landslide in Austria. This slide has in the past moved about 60 m, but the computer model of the slides showed that development of flexural toppling has mobilised shear strength between the beds, which has effectively stabilised the slide. Waldmann and his colleagues reported attempts to model the potential generation of a landslide induced tsunami in Nordfjord in Norway, based upon the development of failure scenarios plus an analysis of two earlier events, one in 1905 (which killed 61 people) and one in 1936, which killed 74 people. Historic photos suggest that the earlier events had a run up height of up to 33 m, illustrating the threat posed. Again, this is work in progress, but it is a very nice piece of research. Finally, Agliardi and his colleagues have untaken the massive task of mapping deep seated gravitation slope deformations (these are massive creeping slope failures that occur in high mountain areas). The study is notable because:
1. It has made use of Google Earth as a primary data source, which shows waht a fantastic resource this has now becaome;
2. The task is epic - it took the individual concerned 2 months to do the mapping alone, which must have been somewhat tiresome;
3. The statistical analysis of the results show very nicely that these types of failure represent a part of the spectrum of mass movement types, and are not a seperate class of landslide in their own right.
Again, there is probably more to do here, but the study is really valuable and needs to be continued.

All-in-all a most interesting day. Once again I was struck by the tremendous range of material presented and by the very high quality of the science in general.

EGU day 2

Day 2 of EGU had less of interest to me than Day 1. I started out in the Historical Landslides session, in which my paper was the first. Most of the other five presentations were excellent. Notable amongst these was a paper by Jan Klimes and his colleagues on the landslide threat to Macchu Micchu. In recent years there have been some fairly lurid headlines about the threat to Machu Picchu as a result of landslide activity. An example is this from the BBC in 2001:

Machu Picchu 'in danger of collapse
Geologists have warned that the ancient Inca fortress of Machu Picchu in Peru is in danger of being destroyed by a landslide...Now, geologists from the Disaster Prevention Research Institute at Kyoto University have warned that the site may be in danger of collapsing.They have found that the land is sliding down at a rate of about one centimetre (0.4 inches) a month. Scientists say this is quite fast and is a precursor to a major landslide.

Pretty scary stuff, especially the part that says that this is a prfecursor to a major landslide. Klimes made a very measured presentation in which he first showed that this is an area with a large number of neotectonic features - which means that care is needed in the interpretation of features in the landscape. He then presented the results of high quality monitoring of the movement of the deep-seated landslides at Machu Picchu. The data clearly showed that the level of movement was orders of magnitude than that suggested above - in fact no more than 2 mm per year (and often less than that). He concluded that these are probably surface movements and that there is no evidence of movement of deep-seated landslides as has been suggested before. It should be noted that the teams monitoring has used more than one tried and tested technique over a prolonged period.

The team deserve great credit for taking on this task and for showing that the threat at Machu Picchu is massively over-stated. One can only hope that the other teams that have been dedicating so much resource to this issue now focus their attention on the real problem in Peru, which is the multiple active landslides that regularly kill and injure the poor population. Addressing these slides could actually make a difference.

In another presentation, Cees Van Westen from ITC showed the sort of thing that is needed. He
presented the work of his team in trying to assess landslide hazard in Guantanamo Province of Cuba. The approach used was to assess the runout of a (very) large landslide from 1963, and to use the parameters derived in a model to assess the likely runout, and thus the risk, associated with other landslides from the same ridge. The study was very impressive and detailed. The approach is not problem-free, but represents one of the most credible attempts to do this sort of thing that I have seen.

Later on I attended talks in the session focussed on the use laser scanning and DEM analysis for the evaluation of slopes. Laser-generated DEMs are probably the biggest single advance ion landslide studies in the last decade. A few years ago when my colleagues Nick Rosser and Mike Lim presented the results of their work using this technique there was a sort of stunned amazement. Now almost everyone uses it. However, I was slightly dispirited by the simplicity of the analyses being generated. In most cases the output seemed to be little more than slope maps and displacement records. The beauty of laser scanning is that we know that the structural and movement data that it can generate tells is a huge amount about what is happening inside the slope itself. It is in many ways like an X-ray of a broken bone, and a similar level of analysis and interpretation is needed. These techniques tell us about mechanisms and processes, and there is an urgent need to look at the data properly.

Tomorrow I won't be at the conference as I have to attend an editorial board meeting, so my next post will be on Thursday.

EGU day 1

Over the next few days I will try to write up some comments on issues that arise at the landslides sessions at the European Geosciences Union (EGU) meeting in Vienna. EGU is a massive meeting (c. 10,000 earth scientists), and the natural hazards section is one of the largest. The landslide field is the biggest component of natural hazards, so there is usually something for everyone. In total there are about 240 talks in the landslides sessions, so I clearly won’t comment on them all! I will instead focus on the matters that arise that interest me in a purely selfish way.

The first session on Monday was on remote sensing techniques and slope failures. Unfortunately I missed it as I was trying to track down my luggage, which once again had been lost by British Airways. Thanks to them once again - there are times when you make me so proud to be British. Oh, and by the way, if Terminal 5 is the answer to Heathrow's problems then I do have to ask what on earth the question was...

I did make it to the second session, which was on a combination of the role of vegetation on slope stability and on landslides and climate change. All six talks were of a very high standard – clear and concise, and covering a good range of material. I am intrigued by the paucity of papers on landslides and climate change, given that this is the topic of the moment. One paper that did address this issue head on, a presentation by Remaitre and colleagues looked at the impact of climate change on shallow and deep landslides in the French Alps. The results were refreshing as they indicated that projected climate changes will reduce the occurrence of slope movements due to a drop of 1-3 m in the groundwater level. Far too rarely do we hear of positive impacts of climate change, but here they are clear.

The previous paper, but Guthrie and colleagues, looked at landslide occurrence on Vancouver Island in Canada as a result of climate change. Whilst this was an interesting approach, the analysis was on the whole a bit over-simplified I think. However, the paper did very clearly highlight the ways in which logging triggers slope failures, demonstrating that clear cutting increases the occurrence of landslides by an order of magnitude or more in this steep, wet environment. The other presentations sought to try to understand why and how vegetation improves slope stability. All were highly competent, but it is clear that more work is needed given the spatial and temporal complexities involved.

All of which seems to highlight the fact that climate change is important, but land use change is much more so in the context of landslides. We should therefore be deeply alarmed by the current situation, in which concerns about climate change are driving land use changes to allow the production of biofuels. For example, felling of forest for palm oil plantations in tropical areas would seem to be deeply unwise. In addition, we are currently seeing massive increases in food and fuel prices, which is also likely to drive increased rates of forest loss.

The first session in the afternoon focussed on landslides triggered by earthquakes. A common theme emerged in terms of the role of topographic amplification in landslide triggering (this is the way in which earthquake waves interact with slopes to cause higher levels of ground shaking, which in turn triggers slope failures). Lee and colleagues presented an investigation of slope failures triggered by the 1999 Chi-Chi earthquake in Taiwan, showing that position on the slope (i.e. proximity to the crest) is the key issue in terms of triggering failure. Meunier et al presented a rather more conceptual analysis that agreed with this, although the topographic role was less strong than some maintain. The latter seems to be let down by the use of the full length of a landslide feature – presumably including the deposit - when really only the source zone should be used. This inevitably biases the results towards the foot of the slope. The observation that slope failures might be more likely on slopes orientated away from the epicentre is interesting, but is not supported by all earthquake events. My own observations from Kashmir do not support their suggestion that landslides cluster around the epicentre either, so I think rather more work is needed here to make this analysis convincing.

Two other presentations in the session focussed upon landslides triggered by MW=7.9 the 15^th August 2007 earthquake in Peru. Wartman and his colleagues focussed primarily on an impressive lateral spread that has in places moved by in the order of 20 m. The surface area of this failure is several square kilometres. Joseph Wartman has pictures of some of the landslides triggered by the earthquake linked from his web site, including a narrated slide show. Hermanns and his colleagues reported on landslides triggered by the same earthquake, but raised a different but very interesting issue. This part of the Andes is littered with large, ancient landslides and rock avalanches, which occur on both the coastal bluffs and in the high mountains. Hermanns noted that this earthquake actually triggered surprisingly few large landslides – and indeed no rock avalanches. What sized event is therefore needed to trigger the landslides that observed – is does this require a different type of event (e.g. a crustal earthquake). There are no answers to this at the moment, but it is a question that needs to be investigated urgently. Dating the ancient landslides, which should be possible for the rock avalanches at least using cosmogenic techniques, would be a good start.

The final session of the day was the Union meeting on forecasting natural hazards. Fausto Guzzetti gave a masterclass on assessing landslide hazard and risk, provoking a lively round of questions. The following speaker, who was talking on a non-landslide topic, impressed me rather less, so I called it a day.

I have the rather unlucky task of starting the first session tomorrow at 8:30 am – I am not expecting a big crowd!

March 2008 landslide map

The map below shows the distribution of fatal landslides for March 2008.

The statistics are:
Number of fatal landslides: 17
Number of fatalities: 77

Thus February was well below the average for 2003-2007, which is 129 fatalities per annum. The distribution is definitely atypical, with more than usual for this time of year in S. America but less in Asia. This would seem to reflect the current La Nina conditions, which have also led to a globally cool spell (as an aside this has got the global warming denier community amusingly over-excited). La Nina conditions do still prevail, but the event is now weakening. Thus, we might sell see below average numbers of landslides for April (but above April for S. America again perhaps).

March 2008 fatal landslide locations (click for a larger version)

2008 fatal landslide locations up to the end of March (click for a larger version)