30 April 2017
Highway 101, Vietnam and others: a selection of new landslide videos
In the last two weeks a range of landslide videos have appeared on Youtube, including examples from Highway 101, Vietnam, India and Turkey. This is a summary:
I think it’s fair to say that this one developed a little faster than the Caltrans workers had expected:
This one occurred in Turkey on 23rd April:
This property is built on an incredibly steep slope:
28 April 2017
The Oso landslide: a new paper in the material properties and failure mechanism
There continues to be considerable discussion in the literature about the mechanisms and processes that led to the dreadful Oso landslide of 22nd March 2014, which killed 43 people. The debate has raged on the sequence of events that led to the landslide, and on the processes that controlled both the initial failure and long runout. This issue is revisited in a paper just published in the the Journal of Geotechnical and Geoenvironmental Engineering (Stark et al. 2017). In my view this study provides clarification on the likely sequence of events, and in particular shows that some of the previous ideas do not fit with observations.
Stark et al. (2017) propose a two phase mechanism, in common with some others. However, they also propose that most of the damage was caused by a complex first phase of movement in which an initial slip on a compound (i.e. non-circular) surface impacted upon colluvium deposits further down the slope. These deposits underwent liquefaction, and flowed across the valley to create the huge damage, and loss of life. The colluvium behaved as a classic flowslide. The main elements of this are shown in my sketch below:
This was followed by a second failure, also shown in the schematic diagram, that results in the morphology that we see at the site today. This was compound shear surface landslide from the so-called Whitman Bench, consisting of unsaturated sands, glacial till and clays, which show a much higher level of friction, meaning that this slide was much less mobile than the first phase. The cause of this second landslide was debuttressing by the phase one slide. I have made a quick sketch of this sequence below:-
This explanation is, in my view, consistent with the available evidence, including the morphology of the site, the eye witness descriptions of what happened, the seismic data and the properties of the materials. Other explanations, such as those that propose an initial failure in the lower portion of the slope, do not seem to be consistent with the evidence.
In essence this is a more conventional explanation for the landslide than some of those that have gone before. It is reassuring that the slide behaviour fits with that observed from elsewhere.
Stark, T.D., Baghdady, A.K., Hungr, O. and Aaron, J. 2017. Case Study: Oso, Washington, Landslide of March 22, 2014—Material Properties and Failure Mechanism. Journal of Geotechnical and Geoenvironmental Engineering, 143 (5).
25 April 2017
The 2015 Gorkha earthquake in Nepal: two years on
Today is the second anniversary of the 2015 Gorkha earthquake in Nepal, which killed almost 9000 people and set back the development of the country significantly. Kunda Dixit has written a nice piece in the Nepali Times about the disaster. He concludes that:
And speaking of funds, Nepal has actually received less than a third of the $9.38 billion the NRA estimated it needed to rebuild homes, public buildings and infrastructure. Of the $4.1 billion pledged by donors in 2015, only $2.73 billion has actually materialised (most of it in loans). This is not even enough for the housing grant of Rs 300,000 per family, which itself is inadequate to rebuild. The National Reconstruction Authority has come up with a new affordable design, but there seems little interest.
The main takeaway on the second anniversary is: International help has fallen far short of pledges and is much less than the amount actually needed, a lot of it is not going through the National Reconstruction Authority, and (as in other spheres of development) the government has failed in coordination.
This is of course a depressing story; one can only imagine what would have happened had the earthquake fulfilled its potential in terms of damage and loss. Worst is yet to come in a future earthquake; we know not when. There is an urgent need to prepare.
Of course one of the most significant impacts of the earthquake was the landsliding, especially in the north of the area of the fault rupture zone. I have highlighted previously that the number of landslides was far lower than we had anticipated might be the case. This is investigated by a paper recently published in the Journal of Earth Science by Xu et al. (2016). This paper, which is open access, compares the landslide distributions of the 2008 Wenchuan earthquake and the 2015 Gorkha earthquake. Although similar in size, the landslide impact was vastly different. For example, Xu et al. (2016) note that the 2008 Wenchuan earthquake generated 25,580 large landslides over an area of 44,000 square kilometres, whereas the 2015 Gorkha earthquake generated 2,064 large landslides over an area 35,600 square kilometres. That is an order of magnitude difference. This is illustrated beautifully by the landslide inventory maps for the two events:
Most interestingly, Xu et al. (2016) try to understand the difference, and in particular why the 2015 Gorkha earthqake appears to have been relatively inefficient in generating landslides. Interestingly, they find that the key issue was not topography nor peak ground acceleration – in both cases the Wenchuan earthquake generated far more landslides than did the Gorkha earthquake for any given slope angle or peak ground acceleration value. Xu et al. (2016) note that:
The landslide numbers and density related to the Wenchuan Earthquake are much higher than those of the Gorkha Earthquake in any combinations of slope angle and PGA. Both slope steepness and PGA cannot explain the low number of Gorkha earthquake-triggered landslides.
They ascribe the difference to the angle of the fault that generated the earthquake. In the case of the 2015 Gorkha earthquake, the fault dips at about 10 degrees, whereas for the 2008 Wenchuan earthquake it dips at about 40 degrees. Xu et al. (2016) suggest that this is important because it causes a different level of deformation in the hanging wall block, and because it causes a different pattern of peak ground acceleration. In particular, as they put it:
Because the Wenchuan Earthquake occurred along a high-angle reverse fault, the area of the surface directly overlying the fault surface was relatively small, concentrating the surface energy and resulting in many large, intense landslides. In contrast, the Gorkha Earthquake occurred along a low-angle fault, so the overlying surface area was wide, the surface energy less concentrated, and fewer smaller landslides were triggered. In short, we suggest that that these two effects determine the capability of an earthquake to trigger landslides because the dip angle of the seismogenic fault controls the manner of earthquake energy release, and thus influences the deformation of the hanging wall bedrock.
I think this conclusion is of vital importance if we are to understand the potential distribution of landslides from future earthquakes. It will be fascinating to see how the landslide distribution maps out for the 2016 Kaikoura earthquake in this context.
Xu, C., Xu, X., Tian, Y. et al. 2016. Two comparable earthquakes produced greatly different coseismic landslides: The 2015 Gorkha, Nepal and 2008 Wenchuan, China events Journal of Earth Science 27: 1008-1015.
24 April 2017
The Wellington earthquake landslide problem
The Wellington earthquake landslide problem was the front page story in one of the local papers 10 days ago. The article is online. The crux of the piece is summarised in the first few sentences:
“Up to 10,000 Wellington homes could be damaged by landslides if a major earthquake strikes close to the city. But while landslides may pose the greatest threat to those living in the capital when the big one hits, it could be years before residents know which slopes across the city pose the highest risk.”
The article also highlights a study being undertaken by GNS Science to try to understand the risks associated with the Wellington earthquake landslide problem:
GNS engineering geologist Sally Dellow said earthquake-induced landslides were usually bigger than those caused by rain. A research project was already under way looking at modified slopes in the Wellington region, she said. Homeowners could expect slopes around them to be evaluated, but it would take extra funding of about $5 million for the work to be completed in four years. “GNS Science has been working on this problem for many years, but only recently has the work started to be funded to the level required to deliver the detail needed to give information that can be acted upon.”
A part of the issue here is that the Wellington Fault runs at the foot of a series of steep slopes very close to the city centre. The Leapfrog blog has a rather lovely image that captures this efature well:
As the image above shows, there are many houses built on the slopes around the city. I snapped a couple of pictures before I left on Saturday:
We know from experience elsewhere in New Zealand and further afield that, in general, slopes close to the surface traces of faults perform badly in earthquakes. The Kaikoira earthquake illustrates this quite well. This is a Google Earth image of the terrain close to the fault for example:
Sometimes after a large earthquake, there are large numbers of cracked and deformed slopes close to the fault that have not transitioned to full failure. These deformations are sufficiently large to damage houses very badly though. This is an image I took from a helicpter after the 2005 Pakistan earthquake:
It isn’t possible to say how the slopes will behave in an earthquake in Wellington, and no-one is suggesting that any particular property is at risk. But it is clear that the work that GNS Science and others are doing on the Wellington earthquake landslide problem is both timely and important.
22 April 2017
March for Science: Wellington, New Zealand
Today, my last day in New Zealand for this visit, I was honoured to be able to attend the Wellington March for Science. Its geographic position means that New Zealand has kicked off a global series of events – I think over 600 are planned. The Wellington March for Science was one of the three that happened this morning; Auckland follows this afternoon. The event was well-attended in glorious, fabulous sunshine – Wellington was at its very best. The crowd initially assembled at the Civic Square:
Before marching over to the National Museum at Te Papa:
And then five short speeches, all good natured and well received:
There were some brilliant placards:
Of course different people were there for different reasons 😉
But thank you to everyone who was there for a happy, well-organised, positive and uplifting day. I wish everyone who is marching today my best wishes. This is so very important.
19 April 2017
Losses from landslides in Colombia: the long term picture
News is coming in today of further landslides in Colombia, with at least 14 reported fatalities in Manizales, where it is reported that at least 50 landslides have been triggered by up to 156 mm of rainfall. Some reports suggest that a further 23 people may be missing. This comes on top of the recent Mocoa landslide, in which at least 300 people were killed.
Back in 2012 I posted about landslide losses in Colombia, noting that:-
If we are to reduce the impact of landslides across South America then Colombia needs to be a priority country.
In 2015 Sergio Sepulveda and I analysed landslides in Latin America in a paper that was published in Natural Hazards and Earth System Sciences. The paper is open access, so it is free to download. I posted a summary of the paper at the time. In that paper we also noted that landslides inflict a heavy toll in Colombia.
Of course in the almost five years since my 2012 post, Colombia has gone through considerable change, including the end of the conflict between FARC and the government. This is therefore a timely moment to take another look at landslide losses there.
The graph below shows the cumulative total number of landslides and landslide fatalities in Colombia, based on my long term landslide fatality database. Note that the data only records landslides that cause loss of life. This does not include the landslides in the last 24 hours in Mazinales.
Sadly, the data suggests that there is no sign that losses from landslides in Colombia are reducing, and it is clear that the cost of landslides in Colombia is high. The good news is that they do not seem to be getting worse either, although the recent losses are clearly the worst since I started to collect this data in 2004.
In my post in 2012 I noted that rainfall in Colombia is heavily affected by the ENSO cycle. Interestingly, the large El Nino of 2014-15 did not drive an increase in landslide losses. However, the rainfall events that the country is currently experiencing are being ascribed to a “coastal El Nino” event. This is not a term that I have heard before; it will be interesting to see if this scientifically valid, and if so if it might provide greater insight into the variable triggers of landslides in South America.
18 April 2017
The Almaluu-Bulak landslide in Kyrgyzstan: amazing new Google Earth imagery
Possibly the most astonishing landslide video of 2016 showed the Almaluu-Bulak landslide in Kyrgyzstan. I wrote about it at the time, and also managed to pin down the location. The video is worth seeing a second time:
This was the second, larger, event in the sequence, both of which occurred on 27th April 2016. One person was killed.
Last month, the Google Earth Blog wrote about this landslide. The prompt was new imagery in the tool that captures the before and after situation of this landslide, and the adjoining area. And wow, this is worth a look. The easiest way (assuming you have Google Earth) to view the images is via this KML file. They have even mapped some of the landslides, though a quick inspection shows that there are many, many more. Indeed this is a landscape that is dominated by landslide processes.
But, in terms of the Almaluu-Bulak landslide itself, the imagery is fantastic. This is the image collected in November 2016:-
It is a most beautiful example of a flowslide. The deposit at the toe is very similar to the deposits seen at the toe of some flume experiments when the flume toe is planar, unrestricted surface, although others will be able to better comment on this.
A really interesting aspect is the before and after comparison of the source area. This is an image after the event:
Note the cracks in the hillslope in the foreground, and behind the crest of the landslide scar. But look at this image, taken two months before the failure event, from the same perspective:
Note the multiple arrays of cracks in the source area. This was a landslide that was developing over a long period. Interestingly, these cracks do not correspond to the current back scar. But we know that two failure events occurred in close succession. I would hypothesise that these crack define to source of the first landslide, whilst the second occurred because of toe release by the first.
The large cracks behind the scar suggest that there is a risk of a third, possibly even larger, event at this site, but that can only be ascertained with a more detailed analysis that involves fieldwork.
17 April 2017
The Meethotamulla garbage landslide: an avoidable tragedy
The number of victims recovered from the Meethotamulla garbage dump landslide in Sri Lanka is now 28, with at least 30, and possibly as many as 100, people still missing. Rescue operations continue, although the reality is that it is unlikely that anyone will be recovered alive now. Meanwhile, controversy has developed over the blame for the tragedy. This Reuters report says:
The police said they were investigating whether the landslide was natural or man-made.
This is not in question. There is nothing natural about this landslide in any way. The question is not whether it was natural, but whether it was foreseeable (note not this is not the same as asking whether it was predictable, as I noted in a post in December). To me the answer is hinted at by this image, published in the Sri Lanka Daily Mirror in 2014, showing the dump at that time, but an investigation is needed to answer the question properly:
Compare this image with this one, published by LankaeNews, showing the aftermath of the failure:
There is a very interesting article on Groundviews website about the garbage dump, its impact on local residents, and of course the landslide. On the Meethotamulla garbage dump itself, it says of the local residents:
One thing they all had in common was their distaste for the nearby dump. It was clearly visible from most of the lanes – swarming with flies, with the occasional pig rooting through the garbage. Even if you couldn’t see it, you could certainly smell it. The people I interviewed said the stench was particularly unbearable when it was hot, forcing many to close their windows and doors to avoid the smell.
Worse, respiratory diseases and skin rashes were quite common thanks to the proximity of the dump.
But the article also quotes a 2010 piece in the Sunday Leader that is deeply worrying:
[CMC Commissioner Badrani Jayawardena] claimed that this site is the only one of its kind to be maintained. “No where in Sri Lanka is it maintained like this,” she said. “Everyday we put a layer of soil on top of the garbage to seal it. We do it step by step but within five days we get the area covered. The reason it looks like such a huge dump is because of the soil as well.”
Putting layers of soil on the garbage way well have been wise in terms of containing the odours, but may have been dangerous in terms of the stability of the dump. This sounds to me to be a recipe for creating planes of weakness, increasing the likelihood of failure. The investigation, if done properly, will need to look into this as well.
In the short term, in addition to completing the recovery operation, there is an urgent need to start to make the dump stable. The peak summer rainy season in Colombo is in May, with an average of almost 400 mm rainfall over the month.
16 April 2017
The Meethotamulla garbage dump disaster in Sri Lanka
At the time of writing, the loss of life from the landslide at the Meethotamulla garbage dump in Colombo, Sri Lanka is 26 people, with reports varying as to how many more victims have yet to be recovered. A press release from the Ministry of Defence in Sri Lanka, which is leading the response, indicates that 145 houses have been buried. This is of course the second major garbage dump landslide this year.
The most informative images that I have seen of Meethotamulla landslide were tweeted by Azzam Ameen (a BBC reporter) on Saturday. They were apparently collected by the Sri Lankan Air Force. This image shows an overview of the Meethotamulla site, with the failed side of the dump clearly shown by the freshly exposed garbage:-
According to Google Earth the dump is about 380 metres long. From the image it appears that the landslide is a pseudo-rotational slip that has formed a very steep back scar (further work is going to be needed to make this site safe before the rainy season begins). The slide appears to have engulfed houses along a large section of the toe. The interaction with the buildings is complex, as this image (from the same source) shows:
It appears that the debris has over-ridden some buildings, but bulldozed others to incorporate them into the landslide:
This disaster did not come out of the blue. This article, from August last year, highlights grave concerns about the stability of the Meethotamulla site. It includes this image, which clearly shows that mass movements were occurring in the garbage:
It is undeniable that this site was unsafe. The garbage mound is clearly too high and too steep, inviting a rotational failure. With houses so close to the toe of the slope the hazards were severe. I am no expert on garbage dump landslides, but it seems likely that the decay of the waste will cause it to weaken with time, increasing the risk. In addition, in a humid monsoonal climate, the pore pressures are also likely to increase with time (as is well-documented in embankments, for example). Thus, the stability of this slope was probably decreasing even without the addition of further garbage.
This is another case in which we know and understand the hazards, but fail to manage them. The results are once again tragic.
13 April 2017
Shimla landslide video
An excellent video has been posted on Youtube of a landslide that occurred at Shimla in Himachal Pradesh in India on 7th April. The landslide has occurred on a slope that appears to have been cut as part of a construction project that is in the foreground:
The apparent lack of slope support at this site seems bizarre given the materials from which this slope is formed:
I have to admire the coolness of the construction workers, who seem remarkably unconcerned about this developing slope failure. Amazingly, the main failure event was captured from a second angle as well:
It appears the the highway is the road that links Shimla and Chandigarh. The construction site may be the base station for a ropeway and a shopping mall.
A gentle landslide from Idaho
Meanwhile, the Weather Channel has posted a most elegant video of a smaller landslide that occurred above a road in Idaho:
I have rarely seen such a gentle, almost balletic, landslide. Amazingly, it did not even cause the light pole at the toe of the landslide to topple as it was gently moved across the road. The Idaho State Journal has an account of this landslide:
ITD crews were on scene with cameras rolling as a hillside broke free and covered a section of U.S. 95 south of Bonners Ferry on Friday, April 7. In the video, you can see tons of mud, rocks, and trees slide down, pushing two layers of concrete barriers off the road. Overall, 800 cubic yards of debris slid onto the road. Thanks to quick action of ITD crews on scene, no one was injured by the slide. ITD Land Surveyor Mathew Wilson, who took the video, was assessing the hillside stability. He heard popping and cracking and knew another slide was imminent. Thinking quickly, the flaggers on scene stopped traffic and just three minutes later the hillside gave way. ITD crews cleared the debris from the road Friday evening. Additionally, crews dug out an extra 600 cubic yards of dirt beside the road to create a channel for water to run.