20 April 2018
Further information about the Castell de Mur landslide in Spain
Over the last couple of days more information has emerged about the Castell de Mur landslide in Spain, which killed two people earlier this week. My friends at UPC, Jordi Corominas and Marcel Hurlimann and colleagues have provided two drone images of the landslide, which I post here with their permission. This image shows the rear scarp of the landslide:
Note that the face is weathered, indicating that detachment has occurred via a pre-existing surface. Lower down there are clear signs of fresh fractures. The UPC interpretation, with which I agree, is that this is a toppling failure primarily of an existing jointset, with fracture of rock generating the final landslide event. The image below shows the resultant landslide deposit:-
There are several really interesting features here. In the images the remains of the former village can be seen, including ruined houses and the tiny chapel, which was destroyed in the landslide. It is also clear that the Castell de Mur landslide occurred in an area of undercutting and rockfall events.
The final failure was recorded on seismic instruments. ICTJA have posted the seismic records online:
The translation of their text states:-
In this image you can observe the seismic wave caused by the detachment of land that occurred in the area of Castell de Mur, in the Pallars Kussà and which could be recorded by different stations of the seismic network of catalonia of catalonia . It can be established, from these records, that the landslide occurred at 14:33 p.m. on 16 April.
The image, made from the data processing made by our researcher Jordi Diaz Cusi, shows the vibration of the floor in the vertical direction recorded by the stations located near Balaguer, Pont de Suert, Sort and Organyà, at Between 25 and 40 km from the place where the event occurred.
“the signal is weak and can only be identified after properly processing the data”, explains Jordi Diaz.
Although it is not easy to attribute a magnitude value to this type of signals, from the ICGC they indicate that it would be the equivalent of a very small earthquake, of local magnitude 0.8. Thus, the vibrations produced by the detachment could only be perceived near the place where it occurred.
18 April 2018
Castell de Mur: a major, fatal rockslide in Spain on Monday
On Monday a major rockslide occurred at Castell de Mur, in Catalonia, Spain. Estimates suggest that this rockslope failure had a mass of about 50,000 tonnes, and a volume of about 20,000 cubic metres. As the image below shows, the landslide came off of a steep scarp and swept across and buried a highway. Unfortunately there was a car on the road at the time, and the two occupants were killed. This image, tweeted by the local fire and rescue service, provides a good overview of the landslide:-
Meanwhile, this image, posted by Naciodigital, shows the impact of the landslide on the road:-
Carlos Jiméne, an engineering geologist, pointed that this is an area with a history of landslides. He noted that near to this rockslide there is an abandoned small village called Les Esplugues de Mur. This village had to be abandoned around 70 years ago because of effects of repeated rockslides, which even destroyed some houses. The only remaining building was the church, but this has been destroyed in the rockslide on Monday. It appears that the landslide on Monday followed a period of heavy rainfall.
It is worth noting the challenges faced by the emergency service in an accident like this. Accessing the site is very challenging, and of course there is always the risk of further slope collapses. Moving the boulders, when they are this size, is a great challenge, and of course a rockslide such as this is very traumatic to the victims.
Many thanks to Carlos Jiméne, Antoni López-Arenas i Cama and Oscar van der Velde, all of whom very helpfully and kindly alerted me to this rockslide. Your help is really appreciated.
17 April 2018
Landslides in Nepal: the need for a national agency?
In a report in Kantipur, various Nepali and Japanese experts are calling for the setting up of a national agency to take responsibility for the management of landslides:-
“Following the adverse impacts of landslide on diverse areas like infrastructure, arable land, human settlements and lives, disaster experts from Nepal and Japan have jointly proposed a Landslide Disaster Management Centre. Disaster experts have drawn the attention of state bodies in a recent five-point declaration on how the government should be prioritising its efforts for saving property and lives to landslides.”
The report points out that over the last year landslides have been the leading cause of disaster induced fatalities, accounting for 29% of the total. I recorded a total 83 fatalities in 2017 – interestingly the government statistics recorded 70 deaths. I suspect that the difference is that my data includes both rockfalls and landslides associated with construction activities. This is the cumulative total number of landslide fatalities in 2017 for Nepal:-
The bunching of the landslides in the early part of the second half of the year is, as usual, clear. This is the effect of the summer monsoon, as I have noted previously. This period is just around the corner, and more losses from landslides are sadly inevitable.
It is hard to argue with the need to manage landslides better in Nepal. The high losses from landslides are partly the result of a landscape-climate system that makes mass movements inevitable, but they are also largely due to poor land use, planning and construction practices. Setting up a national agency is a good first step, but it can only work if it is given the resources and the tools to effect change. That requires good people (Nepal is blessed with good levels of expertise in landslides in the public and private sectors), investment and political will. I am unconvinced that the latter is there. Until that is the case then this level of loss will continue.
12 April 2018
Candelaria mine: a large rockslide in October 2017
The Candelaria mine is a large open-cast copper pit located in the Atacama region of northern Chile, and run by the Lundin Mining Corporation. Back in October 2017 this mine suffered a major rockslide that inhibited operations. In December 2017 the CEO of the company apologised to investors for the low level of communication around the rockslide. The share value of the parent company lost 16% of their value when it was revealed that production would be cut by 20%, although that has partially (not completely) recovered since.
The Globe and Mail article about this event provides some detail about the landslide:
“Phil Brumit, president and managing director of the Candelaria operations, said Lundin came upon a previously unseen and unmapped structure that led to a “wedge failure” at the mine site. Technicians saw movement in the affected area about 5 days before the slide happened. All personnel were then evacuated out of the area. During the slide, between six and seven hundred thousand tonnes of waste material slid into the pit floor. Now the company must devote some of its energy to clearing out the significant amounts of waste material and also ensuring the stability of that section of the mine.”
The landslide happened on 31st October 2017. This is a Planet Labs composite image from October 2017 showing the Candelaria mine. Note the bulging slope at the centre of the image on the eastern slope, extending from the pit bottom about half way up the slope:-
This is the same site, collected again by Planet Labs in November 2017:-
The significant wedge-shaped failure is clearly evident, as is the damage to the roads and the benches. However, once again this is a case in which the failure event was anticipated. The mining industry is extremely effective at monitoring slopes and detecting potential failures, ahead of the rest of the sector. There is a need to try to translate this expertise more widely. This need was tragically illustrated in Japan this week, when a large rockslope failure struck a number of houses, killing six people:-
NHK has an article that explores the cause of this landslide:-
Geological experts say Wednesday’s deadly landslide in southwestern Japan was likely caused by weathering of the slope’s bedrock…At the land ministry’s request, a team of researchers from a Kyushu University graduate school inspected the site later in the day. The researchers found that the bedrock on the upper-most part of the slope has turned into whitish sand, due to weathering. The team says on top of a relatively solid layer of andesite is a softer layer of ignimbrite near the top of the slope. The upper layer is a pyroclastic flow deposit caused by a volcanic eruption between 900,000 and 1,000,000 years ago. The team says the slope collapsed because the weathering of the bedrock extended more than 10 meters below the surface.
Such a landslide would almost certainly show precursory deformation that could be detected with the technology used in the mining industry. The challenge for us all now is to work out how this might be practicable given the numbers of natural slopes that might be at risk.
Planet Team (2017). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://api.planet.com
10 April 2018
Karimabad, Pakistan: a dramatic rock and ice avalanche caught on video
On Monday a very large rock and ice avalanche struck near to Karimabad in Hunza, northern Pakistan. This event appears to have involved a large failure on the Ultar Glacier, which is close to the famous Baltit Fort. This is the Google Earth perspective view of the Ultar Glacier, with the settlement of Karimabad in the foreground:
The landslide was caught on a dramatic video, posted on Youtube, taken from the Baltit Fort:
News reports suggest that five people were caught in the rock and ice avalanche. Of these, two have been rescued, but the prospects for the other three do not look to be good.
The Pamir Times has a good news report about this event, which includes a terrific gallery of images of the aftermath of the event, taken by Ali Hurmat. This image appears to show the rock and ice avalanche deposit, possibly with the glacier in the background:-
Whilst this image appears to show the track of the rock and ice avalanche, with some of the deposit. This appears to be a mixture of boulders with a dust and ice mixture:-
This event appears to be very similar to the giant rock and ice avalanche that struck the military camp at Gayari in Siachen, Pakistan, six years ago. It is only a few kilometres downstream from the very large Attabad landslide, which blocked the Hunza river in 2010. These sorts of events often start as a rockslope failure onto the glacier, which then entrains debris to become a rock and ice avalanche. It will be interesting to see if this is the case here. Satellite data is not yet available, but hopefully the skies will clear in the next few days.
9 April 2018
Santa Cruz island: rockfalls during the M=5.3 earthquake in California on Thursday
On Thursday 5th April California was struck by a shallow (9.9 km depth) M=5.3 earthquake. The epicentre of this event was located off the coast, 29 km to the south of Santa Cruz island, and almost due west of Los Angeles:
This earthquake was not sufficiently large to cause substantial damage, although it was felt over a large area. Probably the most interesting impact was that it triggered rockfalls on the coast of Santa Cruz island. These rockfalls were caught on camera from boats located off shore and posted online by both the Ventura County Fire Department and National Park Service. But Frank Horowitz, a geophysicist at Cornell University, kindly obtained permission for me to post the best of the images here:
The image shows extensive dust generation as cliff instability is triggered. Note that falls are occurring on a number of locations. It is also interesting to note that the apparently displaced block to the right of centre does not appear to be unstable.
The LA Times has an account of these rockfalls from an eyewitness:-
“There were a bunch of rocks falling everywhere around us,” one visitor told KCBS-TV. “We were pretty close to the edge of the cliff and I thought it was going to split off,” another said.
Many thanks to Frank Horowitz at Cornell University for gaining access to the photograph above.
8 April 2018
Fatal landslides in 2017
I have now been able to complete the database of fatal landslides or 2017 (an increasingly challenging task given the commitments of my day job). This is the 15th full year in which I have compiled this dataset, which I formally wrote up in the journal Geology a few years ago. I have written about this work many times before, and Melanie and I have a paper with the most recent analysis of the data under open peer review (please do take a look and leave a review if you have time).
In total in 2017 I recorded 453 landslides worldwide that caused loss of life, excluding landslides caused by earthquakes. In total these caused 4,164 deaths. The graph below shows the annual total number of fatal landslides since 2003:-
So, 2017 was the third worst year in this 15 year dataset, and 2016 and 2017 have reversed the slight downward trend seen in the period from 2010 to 2015. The data continue to show a statistically-significant upward trend over the full period of the study, but no significant trend since about 2009. I remain unclear as to whether this is because I became better at capturing the data, or whether there is a genuine increasing trend. The dataset needs to be much longer to be able to deduce a meaningful trend that exclude the possible effects of improvement in methodology. It is likely though that the number of events is not decreasing, despite our increasing knowledge of landslides and their mitigation. This is quite depressing.
The graph below shows the number of fatal landslides recorded each month in 2017, again excluding the coseismic events:-
As usual the very strong seasonality in the data is clear, with the highest totals falling in the Northern Hemisphere summer – i.e. June to September. This is the effect of the Asian monsoon. It is interesting to note that more events occurred in August than in July – it is generally the opposite way around.
The graph below shows the cumulative total number of fatal landslides and landslide-induced fatalities through the year:-
Note the large steps in the number of people killed. This reflect the contribution of a small number of larger events. The largest step was caused by the terrible Regent landslide in Sierra Leone on 14th August (1309 deaths). Other significant events include the Mocoa landslide in Colombia on 1st April (406 deaths), the somewhat enigmatic Ituri landslide in the Democratic Republic of Congo on 16th August (240 deaths); the Koshe garbage landslide in Ethiopia on 11th March (125 deaths); and the Xinmo landslide in Sichuan, China on 14th June (83 deaths). Rainstorms that caused multiple fatal landslide events occurred in Sri Lanka in May and Bangladesh in June.
A number of earthquakes probably caused landslide fatalities, but I have not been able to extract these from the available datasets as yet.
4 April 2018
Two of the more interesting landslides from the Papua New Guinea earthquake
Slowly improved satellite imagery is becoming available showing the landslides triggered by the Papua New Guinea earthquake. This event seems to have been quite efficient at generating large landslides, some of which appear to be quite interesting. I thought it would be useful to highlight two of these, although there are many more.
A very large avalanche type failure, with a displaced but intact block
On 20th March, Planet Labs was able to image beautifully a very interesting landslide located at -6.14, 142.91 :-
This slide is about 3 km long and the track is over 500 m wide (the headscarp area is about 1 km wide). This landslide is located southeast of Komo Station. The main part of the landslide appears to be a flow type failure, presumably consisting if a fragmented block from the hear scarp, with some debris entrainment along the track. It appears to me that there is another very large displaced block stalled in the headscarp area (note the raft of displaced but intact trees). The rear scarp if this is block is orientated roughly east – west, with considerably more displacement towards the east end. It is likely that this block will progressively degrade over the tears ahead, although a more rapid failure cannot be eliminated. I suspect that the watercourse downstream from this landslide is going to suffer a large input of sediment in the coming years.
A large, complex flow type failure
Further to the west is a very complex area of extensive landslides, in the area of -6.02, 142.62:-
This image appears to show very extensive landsliding around the drainage system. In the centre of the image is a much more complex and interesting landslide, with dimensions of >2 km by >1 km. Downstream from this slide there appears to be a very wide swathe of damage around the river, extending for up to about 4 km down to the main channel. It is not clear to me as to whether this was a part of the landslide itself or erosion from an outbreak flood when a landslide dam breached.
Planet Team (2017). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://api.planet.com
3 April 2018
Landslides triggered by the 14 November 2016 Mw 7.8 Kaikoura Earthquake, New Zealand
The 14 November 2016 Mw 7.8 Kaikoura Earthquake in New Zealand was one of the most significant seismic events in terms of landslide triggering in recent years. I have written about this earthquake, and its landslides, on a number of occasions. In the months since the earthquake a large team has been involved in constructing an inventory of landslides from this event, and in analysing it. A first description of this inventory, and analysis of the landslides associated with the earthquake has now been published (Massey et al. 2018) in the Bulletin of the Seismological Society of America. I have to declare an interest – I am a co-author on this paper.
Massey et al. (2018) demonstrate that the earthquake generated over 10,000 mappable landslides (the inventory described was under development when we wrote the paper, so this total may well increase) over an area of about 10,000 km². Neither the number of landslides nor the area over which they occurred are surprisingly high; indeed if anything the number is low when compared with other events in New Zealand and with global landslide inventories. The largest landslide, on the Hapuku River, had a volume of about 20 million m³ and ran out over a distance of about 2.7 km.
The map below shows the distribution of the landslides triggered by the Mw 7.8 Kaikoura Earthquake, as mapped by Massey et al. (2018). The fault ruptures that generated the earthquake are also shown; note that this was a highly complex seismic event that involved the near-simultaneous rupture of multiple faults and fault segments. The resultant landslide distribution is also highly complex.
I think there are two really interesting aspects of the landslides mapped by Massey et al. (2018). First, the density of landslides triggered by the Mw 7.8 Kaikoura Earthquake was highest close to the fault – indeed our model of the landslide distribution suggests that the distance to the fault rupture was a better predictor of landslides than either the modeled peak ground acceleration or peak ground velocity. It is not clear as to whether this means that the distance function is capturing a key parameter in cosieismic landslide initiation (if so, what is it?) or that the models of peak ground acceleration and peak ground velocity are not capturing reality well. Indeed Massey et al. (2018) found that the landslide density within 200 m of a mapped surface fault rupture was, on average, three times that of locations at a distance of 2500 m.
And second, Massey et al. (2018) found that the point density of landslides for the coastal slopes was much higher than that for inland slopes in similar materials. This may well indicate that more active erosion has meant that the coastal slopes are more out of equilibrium with the environment when compared with the inland slopes.
C. Massey, D. Townsend, E. Rathje, K. E. Allstadt, B. Lukovic, Y. Kaneko, B. Bradley, J. Wartman, R. W. Jibson, D. N. Petley, N. Horspool, I. Hamling, J. Carey, S. Cox, J. Davidson, S. Dellow, J. W. Godt, C. Holden, K. Jones, A. Kaiser, M. Little, B. Lyndsell, S. McColl, R. Morgenstern, F. K. Rengers, D. Rhoades, B. Rosser, D. Strong, C. Singeisen, M. Villeneuve. 2018. Landslides Triggered by the 14 November 2016 Mw 7.8 Kaikoura Earthquake, New Zealand. Bulletin of the Seismological Society of America doi: https://doi.org/10.1785/0120170305
27 March 2018
You can help compile the NASA landslide catalogue
For the last few years Dalia Kirschbaum and colleagues have been compiling a NASA landslide catalogue, with a focus on rainfall-induced landslides, to help with their work on landslide climatology. In a move that we should all welcome, this dataset has now been placed online and can be accessed via a web-based GIS application. The is an incredibly helpful and powerful tool, both for understanding the nature and distribution of the hazard and for teaching. There are a range of different ways to plot the data, allowing the user to tailor the analysis to his or her needs.
As an illustration, this is the global distribution of landslides in the dataset, plotted in terms of fatalities (i.e. the size of the dot represents the number of deaths in each event). The smallest dots are events in which no fatalities occurred:-
The map shows the extraordinary role of Asia in terms of landslide impacts, of course. But as expected, it also highlights the role of reporting in the nature of these types of databases. Thus, for example, the USA has a very large number of recorded events in the NASA landslide catalogue, with few recorded fatalities. The same applies to the UK and New Zealand. In comparison, Africa as a continent has few recorded landslides, but the majority appear to be associated with loss of life. This is unlikely to be the case in reality; the difference is merely the quality of reporting (especially the availability of online media and issues associated with languages). This is the reason that my own work has focused only on fatal landslides – it is more likely that landslides that cause loss of life will be reported reliably, and thus the dataset is probably more homogeneous (although very far from perfect). Of course collecting data on fatal landslide introduces other biases (the location of people, vulnerability, etc), so no dataset can be said to be superior to any other.
It is unlikely that these biases can be overcome completely for a hazard as complex as a landslide, but providing additional means to collate data is very helpful. So, it is now possible to provide information to add to the NASA landslide catalogue. This can be undertaken using the Landslide Reporter tool. It is hoped that this will greatly improve the quality of the dataset, especially in those areas in which data is quite sparse at present.
As an aside, Melanie Froude and I have a paper in review on our work on global landslide impacts. The journal is Natural Hazards and Earth System Sciences, which has an open review process, so anyone can download and comment upon the paper. We would be delighted to hear your views. We think that the data shows that the impact of human induced landslides is increasing.