29 January 2020
The role of weak pumice layers in landslides triggered by the M=6.7 6 September 2018 Hokkaido Iburi-Tobu Earthquake
The role of weak pumice layers in landslides triggered by the M=6.7 6 September 2018 Hokkaido Iburi-Tobu Earthquake
On 6 September 2018 the M=6.7 Hokkaido Iburi-Tobu earthquake triggered large numbers of landslides in Japan. A subsequent analysis (Yamagishi and Yamazaki 2018) suggested indicated that over 6,000 landslides were triggered, resulting in 36 fatalities. They suggested that the exceptionally high landslide initiation rate probably resulted from high susceptibility of the local geology to earthquake shaking.
In a new study, Li et al. (2020) have examined the role of weak pumice layers in the triggering of these landslides. The landslides occurred in an area of pyroclastic deposits from the various volcanoes located in the vicinity. Field mapping suggests that there is a particularly weak pumice layer, the so-called Ta-d strata, which dates from an eruption of Tarumae volcano about 8,700 to 10,000 years ago. The image below, from Li et al. (2020), shows the complex layering in these landslides, including the Ta-d pumice low down in the section.
Li et al. (2020) consistently found that the Ta-d pumice layer acting at the sliding layer in the landslides triggered by this earthquake. They have undertaken a series of laboratory analyses of this material, finding that it is extremely susceptible to liquefaction. They have concluded that a lower permeability palaeosol under the pumice layer enabled seepage parallel to the slope, which may have weakened this layer over time. When subjected to ground accelerations in excess of about 5 m/s2 the pumice layer underwent liquefaction, reducing shear strength to close to zero, allowing the slopes to fail.
This paper provides an elegant explanation for the extensive landsliding that occurred in the M=6.7 Hokkaido Iburi-Tobu earthquake. As we found in our work in New Zealand (Massey et al. 2013), thin layers of weak volcanic materials can often control the susceptibility of slopes to failure. This is probably the most extreme version of this effect that I have seen to date.
Li, R., Wang, F. & Zhang, S. 2020. Controlling role of Ta-d pumice on the coseismic landslides triggered by 2018 Hokkaido Eastern Iburi Earthquake. Landslides https://doi.org/10.1007/s10346-020-01349-y
Massey, C.I., Petley, David and McSaveney, M.J. (2013) Patterns of movement in reactivated landslides. Engineering Geology, 159, 1-19.
Yamagishi, H. & Yamazaki, F. 2018. Landslides by the 2018 Hokkaido Iburi-Tobu Earthquake on September 6. Landslides. https://doi.org/10.1007/s10346-018-1092-z.
21 January 2020
The 23 July 2019 Shuicheng County landslide: a first scientific report
On 23 July 2019 a large landslide occurred at Pingdi village in Shuicheng County, in Guizhou Province, China. The landslide, which occurred after a spell of heavy rain, killed an estimated 42 people after it destroyed 20 houses. Eleven people were rescued. I blogged about this landslide at the time, but now a first scientific account of this landslide (Li et al. 2020) has been published in the journal Landslides.
This was a large slide – Li et al. (2020) estimate that it had a total length of about 1,300 m, a vertical extent of about 120 m, a surface area of about 370,000 m² and a volume of about 2 million m³. The authors have generated this annotated orthophoto of the landslide site, which clearly shows the source area of the slip, the transportation and entrainment zone and the landslide deposit:-
Interestingly, Li et al. (2020) describe this as being a rotational landslide, with an initial thickness of about 20 m. This would not have been obvious to me. The authors note that a road passed through the source area, but they do not comment as to whether this might have been a factor in the failure. In the seven days prior to the failure the area experienced three heavy spells of rainfall, totaling about 154 mm. Whilst this is undoubtedly intense, it seems unlikely that this was particularly exceptional.
Finally, Li et al. (2020) note that this slope continues to represent a hazard:
“The field investigation suggests that this slope may fail again in the future under rainfall/earthquake events. Hence, to avoid any secondary disaster and protect people’s lives and properties, it is crucial to strengthen the monitoring and early warning systems of landslide by the local government.”
A key question remains though. That is, could it have been anticipated in advance that this slope, rather than any other in the area, was close to failure. And if so, how?
Li, H., Xu, Y., Zhou, J. et al. 2020. Preliminary analyses of a catastrophic landslide occurred on July 23, 2019, in Guizhou Province, China. Landslides. https://doi.org/10.1007/s10346-019-01334-0
20 January 2020
Brumadinho: the Expert Panl report on the failure of the Feijão tailings dam
In a few days time it will be the first anniversary of the failure of the Feijão tailings dam at Brumadinho in Brazil. Last month the official expert panel released its report on the disaster – there is a website dedicated to the findings, which includes the full report in English and all of the appendices.
As expected, the report finds that the failure occurred as a result of static liquefaction. The investigation has deduced that failure initiated close to the crest of the dam but very rapidly progressed through the entire structure, allowing a comparatively shallow failure to develop. This was then followed by a series of retrogressive failures that released the large volume of mine waste.
There is a particularly interesting aspect of the report that will cause deep concern for those responsible for such structures:-
“The failure is also unique in that it occurred with no apparent signs of distress prior to failure. High quality video from a drone flown over Dam I only seven days prior to the failure also showed no signs of distress. The dam was extensively monitored using a combination of survey monuments along the crest of the dam, inclinometers to measure internal deformations, ground-based radar to monitor surface deformations of the face of the dam, and piezometers to measure changes in internal water levels, among other instruments. None of these methods detected any significant deformations or changes prior to failure.”
That such a catastrophic failure can develop with no signs of distress, and no indication that stability was being compromised, is a great surprise. It implies that the failure was extremely brittle. The Expert Panel looked in detail at all of the monitoring data, and undertook a back analysis of historical InSAR data. They conclude that the dam was settling at rates of up to 30 mm per year (as illustrated schematically in the image below), but this was expected and could not be used to infer that stability was compromised.
In terms of the failure, the Expert Panel blames the upstream construction method deployed by Vale at Brumadinho. In particular, the Expert Panel is critical of the lack of effective drainage installed during the early phases of construction, which allowed very high pore water pressures to develop both during deposition (which meant that the tailings were loose) and then afterwards (which promoted failure). At the same time, the tailings contained a very high content of iron, which allowed them to become bonded, introducing the brittleness highlighted above. The upshot was that the dam was, in the words of the Expert Panel, “composed of mostly loose, saturated, heavy, and brittle tailings that had high shear stresses within the downstream slope, resulting in a marginally stable dam (i.e., close to failure in undrained conditions)”. In other words, Brumadinho was a disaster waiting to happen.
The final failure did not need a specific trigger – it was in essence progressive. Long term heavy rainfall had reduced suction forces whilst creep of the tailings caused strain to localise, which promoted creep rupture.
There are many lessons to learn from this failure, but absolutely critical will be the finding that failure could not have been anticipated through monitoring. This means that to understand the behaviour of the dam the local team needed a much more nuanced view of the properties of the tailings (and in particular their propensity to fail in a brittle manner) and the conditions within the dam. This is a critical lesson for the industry.
16 January 2020
Oktibbeha: another landslide induced threat to an earthen dam
In Mississippi, USA attempts are underway to lower the lake level behind the Oktibbeha County Lake dam (location 33.509, -88.944), which is described as being in imminent danger of collapse. Reports suggest that an inspection of the dam at 7 am on Tuesday indicated that a significant landslide had occurred on the downstream face. A further inspection four hours later found that substantial further deterioration had occurred. This image, from Starkville Daily News, shows the extent of the landslide:-
Attempts are now underway to lower the level of the lake using siphons too supplement the spillway, but further rainfall is forecast in the coming days. The local residents have been advised to evacuate, although at the moment this is not mandatory.
Ongoing problems with the Oktibbeha dam
This is not the first time that the Oktibbeha dam has faced problems. An inspection in 2016 rated the dam condition as “fair”, whilst a further report in February 2019 identified ongoing seepage problems with the structure. The Starkville Daily News reports that the ongoing failure has occurred at the same location as other slope problems, and a slope repair was also required in 2016 and in 2017. In September there were discussions about the need for the replacement of the levee, and State level funds were sought to renew the dam as “the waters of Oktibbeha County Lake Dam could pose danger to nearby residents if the levee is not replaced”. However, the costs of this work are around $8 million – as the image below shows, this is not a small structure:-
Once again, this incident highlights the fragility that is being created by the legacy of ageing infrastructure with inadequate resource being committed to upgrades. The impacts of a collapse, which will hopefully be avoided this time, would be substantial. Interesting, the Star Tribune report includes the following:-
There have been at least two other dam failures in the South after heavy rains within the past month.
Holmes Lake Dam in Hinds County, Mississippi, failed Jan. 2. Some vehicles were damaged, but no injuries were reported. A post-failure inspection indicated that faulty construction may have allowed internal erosion of the earthen dam, said Willie McKercher, chief of the Dam Safety Division at the Mississippi Department of Environmental Quality.
A pond swollen by heavy rains broke through a dam in Aiken County, South Carolina, on Dec. 23, damaging several vehicles but causing no injuries.
15 January 2020
Major landslides caused by agricultural irrigation in Peru
It is well-established that irrigation by farmers can be the cause of landslides, especially in areas with low rainfall totals. A year ago I wrote about evidence that irrigation in Washington State in the northwest of the US has been driving landlsliding on river bluffs. This month, a very nice study was published in Nature Geoscience (Lacroix et al. 2020) that uses a long archive of satellite data to examine landsliding on river cliffs in Peru. The study area, the Vitor and Siguas valleys in southern Peru, is hyper-arid, which has meant that historically agriculture has been limited to the flood plains of the incised river channels. However, in the second half of the 20th Century, agriculture was extended onto the desert plains using irrigation to sustain the crops. This Google Earth image, of the Rio Vitor, shows the deeply incised (c. 200 m deep) channel, the agriculture on the flood plain, and the irrigation-supported farming areas on the desert plateau-
The image also shows the extensive landsliding on the river bluffs below the areas under irrigation. Note that in two locations there are ponds at the foot of the rear scarp of the landslide.
Lacroix et al. (2020) have used a combination of declassified Hexagon spy satellite and SPOT 6/7 images to measure movement rates in the landslides. They found surprisingly high movement rates – in some cases total displacements of 400 m were recorded. In general, movement started about 20 years after irrigation was started, presumably representing the period over which pore water pressures developed. Thereafter movement was somewhat episodic, without an obvious driver from environmental drivers such as precipitation. Interestingly, the highest movement rates were recorded in the parts of the slope, presumably reflecting the locations in which pore water pressures are highest.
The development of irrigation driven landsliding on these river slopes can be clearly seen in the Google Earth imagery. The image below is the same area as that shown above, but this time the image is from 2017, 14 years later:-
Very clear development of the landslides – for example the one slightly left of centre – can be seen. The head scarp has eroded back into the farming area, whilst the toe of the landslide has moved forward, claiming further fields. It is interesting to see many more ponds.
Interestingly, the authors note that further large irrigation schemes are planned in this area. Lacroix et al. (2020) suggest that further development of landsliding can be expected as a result.
Reference and acknowledgement
Lacroix, P., Dehecq, A. & Taipe, E. 2020. Irrigation-triggered landslides in a Peruvian desert caused by modern intensive farming. Nature Geoscience 13, 56–60. doi:10.1038/s41561-019-0500-x
I came across this article through of a very nice description of it in Eos by Jane Palmer:
Palmer, J. 2020. Modern farming kick-starts large landslides in Peruvian deserts. Eos, 101, https://doi.org/10.1029/2020EO138556. Published on 14 January 2020.
14 January 2020
The Hope Slide in Canada: new images released by BC Highways
Early on 9 January 1965 a very large landslide occurred in the Nicolum Valley in the Cascade Mountain Range in British Columbia in Canada. Now termed the Hope Slide, the landslide involved the collapse of Johnson Peak, generating a 47 million cubic metre landslide that buried BC Highway 3, killing four motorists.
Last week, BC Highways released onto their blog an archive of images taken in the immediate aftermath of the landslide, together with a blog commentary. There is also a nice article on the CBC website about the image set. This is an overview of the landslide from the archive:-
The image below shows the opposite side of the valley. The landslide has clearly run up the valley wall, stripping the topsoil and the trees from the slope, demonstrating that this was a high velocity flow.
The image below shows the aftermath of the landslide, with the extensive debris deposit across BC Highway 3. Clearly this presented the highways agency with a substantial challenge in terms of reopening the road. Perhaps surprisingly, a temporary road was opened in just three weeks.
There is a good Wikipedia page about the Hope Slide, including a section on the causes and the trigger of the failure. The underlying causes were pre-existing tectonic structures (faults and shear zones), which chronically weakened the slope, ultimately providing a detachment surface. The trigger is less clear – whilst two small earthquakes were recorded in the area, it is unlikely that these were sufficient to initiate collapse. It seems likely therefore that this was a progressive failure.
13 January 2020
Burundi: Planet Labs and Sentinel images of the 4 December 2019 landslide disaster
On 4 December 2019 intense rainfall triggered a significant, if under-reported, landslide disaster in Burundi in Africa. Whilst there were problems in many regions of the country, the most serious impacts occurred in the area around Nyempundu, in Mugina commune in Cibitoke. If you want to take a look, this is around -2.62, 29.09.
According to the Wikipedia page about the disaster, at least 38 people were killed by landslides in this event, and there was substantial damage to infrastructure such as bridges and roads, as well as loss of houses. The response has inevitably been hampered by the remote location.
This is not an area in which it is particularly easy to get good satellite imagery due to cloud and water vapour. However, Planet Labs captured this PlanetScope image on 7 December 2019 showing the landslides in the most heavily affected area:-
The imagery appears to show multiple shallow landslides with some evidence of debris mobilisation through the channel system in the form of debris flows. In some sub-catchments much of the topsoil appears to have been removed.
These are the same landslides, shown in a Sentinel-2 L1C image, via Sentinel Hub. The image is dated 14 December 2019:-
Although it is lower resolution, the advantage of Sentinel is that it is easy to generate an NDVI image, in which the contrast between vegetation (which appears red) and bare soil (which appears tan) is clear. This is an NDVI image from the above Sentinel image, which highlights the landslides:-
It is clear that in this area of Burundi a substantial landslide disaster has occurred. Of course the damage to the hillsides and sediment in the channel will cause long term problems for the local population.
Planet Team (2020). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://www.planet.com/
9 January 2020
An international workshop on “Assessment and Mitigation of landslides in the Himalaya”
The organisers describe the event as follows:
The purpose of this workshop is sharing scientific knowledge and experiences through case studies and best practices by the experts of different institutes including the ministries, international agencies and landslide researchers. This would add value to the present understanding of different approaches used in landslide studies and policy planning & practices, and enable to prepare a roadmap for addressing risks and vulnerability in the region. It would also help in developing network of institutions and experts around the thematic areas of the workshop. The two-day deliberation will include plenary lecture, invited talks, special talks on the landslides in Uttarakhand and Sikkim Himalaya, and presentation of contributory papers (oral and poster) on the following themes:
- Landslide Hazard, Vulnerability and Risk Evaluation
- Landslide Mechanism and Modelling
- Climates and Landslides
- Current Practices for Landslide Studies
- Prediction of Landslides and Early Warning System
Registration and call for Abstracts
Abstracts (title, name of the authors, affiliation, presenting author), not exceeding 500 words are invited, and may be submitted by email to the Convenor. It is also intended to publish the proceedings of the seminar as a special issue in the Journal of Himalayan Geology.
- Workshop dates : March 13-14, 2020
- Last date for submission of abstract : January 31, 2020
- Last Date for registration : January 31, 2020
8 January 2020
Bududa, Uganda: Planet Labs images of the landslides in December that killed more than 50 people
On 3 December 2019 heavy rainfall triggered a number of landslides in the Bududa area of Uganda. In my dataset I have recorded 53 fatalities from these events, although there is some uncertainty (the Red Cross for example has recorded 51 deaths). It appears that around 50 houses were destroyed and around 1,000 people were directly impacted. This area of Uganda is one of the most landslide-affected parts of Africa – I have reported multiple fatalities in this region on several previous occasions, including in 2011, 2012, and 2018.
The Red Cross reports that the villages affected by the landslides in December were Shikhururwe, Namasa, Naposhi and Namwau in Bushika Sub County and Bushibekye in Bunabutiti Sub County. I have taken a look at the Planet Labs imagery for this area before and after the landslides. The landslides are clearly visible.
There is a good image of the area on 17 November 2019, which shows the hilly, mostly deforested landscape with large numbers of houses scattered across the slopes. No very recent landslides are evident:-
This is the same area on 15 December 2019, in the aftermath of the landslides:-
Multiple landslides are evident on this images. There is a cluster of slides in the northeast, on forested slopes, and multiple smaller slides in the centre of the image, some of which appear to have become channelised debris flows with a long runout. The most important landslides appear to be the two in the lower portion of the image, left of centre, on a conical hill. The image below is from 1st January 2020 – I include it here as it provides more detail of the landslides:-
The large slide in the south west of the image appears to be the most significant in human terms. I can count at least 15 houses that have been destroyed by this landslide; it is likely that there were substantial numbers of fatalities at this site.
Planet Team (2020). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://www.planet.com/
7 January 2020
Trimingham: a large landslide in complex glacial deposits
Yesterday morning (6 January 2020) a large landslide occurred on the sea cliffs at Trimingham in north Norfolk, on the east coast of the UK. ITV News has some spectacular images of the aftermath of the landslide, and a nice drone video of the site too. This is one of the images:-
An alternative view shows that the landslide had quite high mobility across the beach:-
Substantial landslides on the east coast of the UK are not unusual – note the multiple landslides of various types shown on the images above for example. This is an area of exceptionally high rates of coastal erosion – indeed the British Geological Survey has an ongoing project in this area estimating the rates of loss of land. The data suggests that over the period between 1966 and 1985 this section of coast lost an average of 1.5 to 2.5 metres per year.
The geology of the cliffs at Trimingham is particularly interesting, and accounts for the style of landsliding that occurs at this site. The cliffs are formed from very large blocks of chalk sitting in series of glacial and periglacial sediments. This has been interpreted as a large push moraine created by the Middle Pleistocene ice sheet. It is draped by large amounts of outwash sand and gravel.
In a short field guide, available online, Lee et al. (2011) describe how these glacial and periglacial deposits control the landsliding on this coast:-
Landslide development along the coastal traverse is strongly controlled by glacitectonic structure. Thrusts (i.e. synclines) offer natural failure planes for landsliding as well as acting as planes that enable groundwater to infiltrate and migrate around the sequence. Thrusting also alters the geometry of discontinuities such as bedding and jointing ‐ at Trimingham, subvertically aligned beds of diamicton are highly prone to failure, generating earth flows and falls. The movement of sediment blocks of different permeability relative to each other during thrusting (and landsliding) leads to the development of highly localised and confined groundwater conditions which often accentuates landslide hazards. Synclines within the sequence also offer topographic lows within which groundwater seepage is focussed causing failure along bedding planes and the generation of large deep‐seated rotational and translational slides.
The UK has suffered an exceptionally wet few months, so coastal landsliding is not surprising. As rainfall intensities increase as the effects of global heating continue to accumulate, it is likely that we will see many more landslides in these highly vulnerable geological deposits.
Lee, J.R.; Pennington, C.V.L.; Hobbs, P.R.N.. 2011. Trimingham : structural architecture of the Cromer Ridge Push Moraine complex and controls for landslide geohazards. In: Phillips, E.; Lee, J.R.; Evans, H.M., (eds.) Glacitectonics : field guide. Quaternary Research Association, 218-227. (QRA field guides).