19 October 2018
Landslide tsunamis from the Sulawesi earthquake
Our understanding of the Sulawesi earthquake continues to improve as more information becomes available. This has shown Twitter at its very best, with a range of people bringing different expertise, and techniques to bear on the problem of what has proven to be a very complex event. I would like to recommend in particular Austin Elliott, John O’Leary, Xiaohua Xu, Eric Fielding and (most impressive of all) Sotiris Valkaniotis, all of whom have been very proactive in interpreting and synthesising the data, and making the information available online.
New videos continue to appear on Youtube. One is particularly interesting, taken from a boat located just offshore when the tsunami struck:-
There is so much in this video that it is hard to know where to start. The video opens with a view roughly southwards (corrected from earlier version) up the bay, showing an enormous landslide generated displacement wave from the western side of the bay:-
The view then switches to the east side, where another landslide tsunami has been generated:-
Incidentally, as the camera pans from one to the other, a third wave, generated from the eastern side of the bay, can also be seen. Meanwhile, in the background the multiple dust plumes generated by landslides in the hills can be seen:-
On Twitter, Austin Elliott managed to tie down the location of the boat to -0.8303, 119.8133, and John O’Leary confirmed that there was a landslide at that point.
This video aligns nicely with the remarkable video taken by a pilot of an aircraft that had just left Palu airport when the earthquake struck. This video also shows the generation of landslide tsunamis by the Sulawesi earthquake:-
We have never had such a detailed and nuanced view of the way that coastal landslides generate local tsunamis during an earthquake.
15 October 2018
Sigi Biromaru: terrifying footage of lateral spreading during the Sulawesi earthquake
Sigi Biromaru is a suburb of Palu located on the eastern side of the valley. Whilst most of the focus in the two weeks since the Sulawesi earthquake has been on the flowslides, it is easy to forget that other slopes also went through significant lateral spreading during the mainshock. In an initial (amazing) analysis of optical imagery collected by Sentinel-2, Sotiris Valkaniotis of the Aristotle University of Thessaloniki has shown that there was extensive lateral spreading on the eastern side of the fault. To confirm this, a video has been posted on Youtube that shows the development of very significant lateral spreading during the earthquake. In this case it appears that the lateral spreading did not transition into the type of flowslide seen elsewhere in Palu, but the level of damage is very substantial. The original video is available on Youtube:-
Unfortunately the camera is rotated, which makes watching the video quite challenging. There is another version that is easier to watch, albeit with an irritating backing track:
The astonishing thing about this video is that the lateral spreading, as indicated by the large tension cracks in the road, develops as the video runs. These two screenshots, taken less than a minute apart, show one of the fissures opening:-
A few seconds later the fissure looked like this:-
Fortunately in this case full liquefaction, if that is what occurred to generate the flowslides, did not occur. The magnitude of the damage is extraordinary though, illustrating the magnitude of the challenges that lie ahead in getting this area back on its feet:-
In interesting element of this disaster will be understand why this slope did not go to the full, flowslide-generating failure seen at Balaroa and Petobo. It is very fortunate that this did not occur.
12 October 2018
Bududa in Uganda: landslides kill at least 40 people
Heavy rainfall on Thursday (11th October 2018) triggered landslides in Bududa District of Uganda, with reports suggesting that at least 40 people have been killed. Some reports suggest that the toll might be very much higher, with The Ugandan stating that up to 400 people might be missing. This includes an alarming report about the impact on a school:-
“Over 200 pupils of Shiteteyi Primary School are also missing after the mudslides swept part of their school located at Maludu trading center.”
These reports are unconfirmed as yet. The reports includes a photograph of one of the landslides (although this might be an archive image):-
Clearly at present the picture is far from clear; hopefully it will clarify in the next few hours. All the indications are that this will be a significant disaster.
In terms of fatalities, the Bududa area of Uganda might be the most disaster prone in Africa. I have previously posted about significant landslides there in 2011, with at least 50 fatalities; and 2012, with about 90 fatalities. There is a report on Wikipedia about a large failure in March 2010, with about 100 deaths.
The area affected by these landslides is the flank of Mount Elgon, an extinct shield volcano that reaches an elevation of over 3000 metres. The flanks of the volcano are deeply weathered to laterite, and the area is affected by heavy rainfall events, with an annual total of over 1,800 mm. Humans have heavily modified the landscape, with widespread deforestation in the last 25 years. This has involved increasing cultivation of the steeper slopes higher up the volcano. The underlying cause has been a shortage of land in more suitable areas due to population pressures. Land is now being farmed that, without natural vegetation, is inherently unstable.
The image above illustrates these problems perfectly.
Interesting, the Kenyan side of the volcano is not subject to the same level of landsliding. Kenya declared this area to be a national park in 1968, and this area was declared as a Biosphere Reserve by UNESCO in 2003.
10 October 2018
Two developing large landslides – Mannen (again) and Diamante
Diamante is, according to Wikipeda, “a city in the west of the province of Entre Ríos, Argentina, on the eastern shore of the Paraná River. It has about 20,000 inhabitants as per the 2001 census“. Over recent weeks a large tension crack has developed, defining a block some 130 m long, causing the evacuation of 60 families. Concern was raised about the incipient failure on Saturday, since when the incipient landslide has continued to develop:-
Mannen is creeping again
Meanwhile, in Norway our old friend the Mannen rockslope failure is moving rapidly once again. This large incipient rockslope failure has gone through a number of periods of accelerated creep, but yesterday the rate of movement increased to about 40 cm per day in the upper part of the slope. This has been associated with increased rockfall activity, which is generally an indication of increased deformation of the creeping rockmass as strain develops. Overnight the rate has increased a little more, reaching 2 cm per hour. As a consequence, the long suffering residents were evacuated once again, the fifth time since August, and the tenth time since 2014.
There is a live webcam showing the site in this newspaper report. Should failure occur it is estimated that the landslide will have a volume in the range of 120,000 to 180,000 m³. NVE has a webpage that is providing regular updates on the landslide, including screenshots of the monitoring data.
The challenges of forecasting failure
In both cases the problems with forecasting future behaviour are well-illustrated. Evacuating the local people is essential at each of the two sites, but the geologists will be very aware of the impact that this will be having on the community. The experts will be feeling immense pressure, but there are huge uncertainties when trying to extrapolate future behaviour of natural slopes. It is frustrating for all involved.
Thanks to Donaldo for highlighting the Diamante landslide
9 October 2018
Too close for comfort – a rockslope failure at Shongtong in northern India
In northern India major works are underway to construct a new hydroelectric facility at Shongtong Karcham in Kinnaur. This will be a 450 kW run of the river scheme generating an estimated 450 MW, reportedly enough to power 800,000 households. This is another scheme being constructed in an area with high landslide potential. There is an interesting article on Himdhara about the environmental impact of these major construction projects in the Kinnaur area, which includes a section on the impact of this scheme:-
“The 450 MW Shongtong-Karchham project which is now under construction had been impacting 5 villages. Its tunnel was being constructed in a landslide-prone area—no wonder then that fresh landslides are being reported from the area with alarming frequency. The people of one of the affected villages, Barang, had brought to a halt the construction activities for the project in 2013 to negotiate their demands. Last year the work on the project’s tunnel began in full swing and led to caving in of private lands belonging to four families.
“Similar was the case of the neighbouring Mebar village. In April 2015, a section of the national highway, where the tunnel construction was going on, was blocked as a mass of rocks just collapsed. This area called Laal dhank (rocky cliff/over hang) has the village Barang situated right above it. It took close to 48 hours to fully clear the road as vehicles and travelers lined up at both ends of the road.
“Perhaps the one striking case which managed to get some coverage over the last year is of the Urni landslide. “Urni is sitting precariously above the junction of the flushing tunnel, Head Race Tunnel and Adit tunneli of the newly operational 1200 MW Karchham Wangtoo project. In July 2014 the Urnidhank collapsed blocking the national highway,” said Ramanand Negi of the village.”
This image shows the magnitude of the problem:-
In the last few days a video was posted on Youtube, collected by a group of workers at a part of the project, showing a rockslope failure at Shongtong:-
The video captures the initial failure with unusual clarity:-
And the development of the rockfall:-
8 October 2018
The Palu landslides – the worst landslide disaster in five years?
It is becoming increasingly clear that the lass of life from the three Palu landslides, triggered by the earthquake in Sulawesi, is very high. In particular, the two landslides that struck main parts of the town, at Balaroa and Petobo, are thought to have exacted a terrible toll. A spokesman from Indonesia’s disaster agency, Sutopo Purwo Nugroho, was quoted thus on Sunday:-
“Based on reports from the (village) heads of Balaroa and Petobo, there are about 5,000 people who have not been found. Nevertheless, officials there are still trying to confirm this and are gathering data. It is not easy to obtain the exact number of those trapped by landslides, or liquefaction, or mud.”
Given the scale of the Palu landslides, and the very high number of houses involved, this level of loss is entirely possible. It is clear from the videos that the flow was highly mobile, which will have made escape from the crush of houses and debris extremely challenging, especially in a dense network of streets. The mobility of the debris will have meant that buildings will have filled with mud, even where there were voids that might have protected a person. It is therefore unsurprising that no survivors have now been recovered in a week. This AFP video shows the magnitude of the damage:-
The search for the missing will continue until Thursday; after that it is likely that the landslide sites will need to be classified as mass graves.
There is considerable confusion in the reporting as to what actually happened at Balaroa and Petobo. This is Petobo, as seen from the air:-
Whilst this is Balaroa:-
Various newspaper articles have described these events as being liquefaction, even nonsense such as:
“Petobo, a cluster of villages in Palu, was virtually wiped out by the powerful quake and wall of water that devastated Palu. Much of it was sucked whole into the ground as the vibrations from the quake turned soil to quicksand.”
In both cases these were landslides, of that there is no doubt. The mechanism of failure might have been liquefaction, but this is just a hypothesis. The houses were not sucked into the ground, they were buried and crushed by a flow of debris coming from upslope, albeit down a very low gradient slope. Whilst this might seem academic, it is important in the context of understanding future hazards.
To understand why these slopes flowed under the effects of earthquake shaking will require detailed investigations, in due course, using field data collection, lab testing and modelling. But seismically induced flowslides are not a new phenomenon by any means, and in previous events they have caused huge numbers of fatalities.
5 October 2018
Indonesian earthquake: loss maps from the Palu landslides, and large landslides to the south of the city
Indonesian earthquake: loss maps from the Palu landslides
A notable aspect of the Indonesian earthquake has been the rapid availability of large amounts of satellite data, greatly assisted by the effective activation of the International Charter Space and Major Disasters. The data is now being analysed by agencies around the world. LAPAN have been using the data to map the losses from the landslides in Palu. The results are fascinating and dreadful. This is their before and after comparison for the landslide at Balaroa:-
The number of houses destroyed by the landslide is very high – some reports suggest about 1,700 homes in the Balaroa area alone. Similar maps are being produced for the other landslides. Interestingly, Reuters have an article about the earthquake damage that includes some excellent graphics, including detail about the landslides. However, the images of the site show the full magnitude of the damage:-
The article quotes Sutopo Purwo Nugroho, spokesman for the National Disaster Mitigation Agency: “We don’t know how many victims could be buried there, it’s estimated hundreds”.
Indonesian earthquake: large landslides to the south of the city
Meanwhile, to the south of Palu there is an area of reasonably dense landslides, as shown in the Planet Labs imagery:
In this area the fault passes through hilly terrain; for the mostpart the landslides appear to be close to the fault trace. There are some large landslides in this area, although these are more conventional failures than was the case in Palu. This landslide for example appears to be a reasonably deep-seated slip that is about 800 m long:-
This landslides, and indeed most in this area, did not have the high mobility of the landslides in Palu, and so are probably not liquefaction-initiated flowslides.
Planet Team (2018). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://www.planet.com/explorer/
3 October 2018
High resolution imagery of the Palu landslides
DigitalGlobe have now released high-resolution imagery of two of the Palu landslides, triggered by the 2018 Sulawesi earthquake. This is available as a “before and after” slider on the CNN website (and elsewhere). The level of detail is remarkable, and we are now starting to get an idea of the mechanisms and motion of the landslides.
As I noted yesterday, the landslide, which moved from east towards the west, appears to be a planar flowslide of some type, with a very strong lobate form away from the source. Note that the landslide debris itself is an extraordinary combination of soil and debris, and houses and buildings, all compressed together. The total loss of homes is very high, but it is hard to know as to the likely level of human casualties. There are multiple tension cracks in the headscarp area, and this will be a very challenging site in the next heavy rainfall event.
meanwhile, the smaller landslide at Balaroa, the site closest to the city centre, is also terrible. On this occasion the landslide moved from west to east, and again a large, mostly evacuated, source area can be seen. There is a raft of mostly intact debris and vegetation downslope, and then an extraordinary compression zone of destroyed buildings. The slide is about 1 km long. The level of loss is once again high, but the number of fatalities is difficult to determine.
It is worth noting that for the survivors this type of event is worse than the more normal building collapse. In an earthquake-induced building collapse people may lose their property and most of their possessions, but at least sometimes still have their land. In the case of a landslide they typically lose property, possessions and the land that they inhabited or owned, as the image above shows. This is devastating in the long term.
InSAR data of the area affected by the earthquake is now becoming available, which shows the line of the fault. As expected, this data suggests that these big landslides are close to the fault trace. Unexpectedly, the data also suggest that the fault that ruptured extends into the bay north of Palu, but then steps onshore to run through the isthmus rather than offshore. This step may explain the tsunami, rather than it being a submarine landslide (the jury remains out on that one). Rupture appears to have been strongly directional, propagating south from the epicentre at the northern end of the earthquake zone. The upshot is that Palu suffered a bullseye hit from a large earthquake, with a strong directional component. It is no wonder that the level destruction is so high. In that context the number of human casualties currently reported seems surprisingly low to me.
2 October 2018
Giant landslides in Palu from the 2018 Sulawesi earthquake
As decent satellite imagery starts to become available, the landslides in the vicinity of the city of Palu, triggered by the 2018 Sulawesi earthquake, are becoming clear. The picture is fascinating and tragic, providing an explanation for some of the remarkable footage that we have seen. There is a key message here – although it is the tsunami that has attracted most of the attention, three very large landslides in Palu have caused very high levels of loss. These three landslides are remarkable.
So, starting from the south, let’s take a look at these three slides. The first appears to be in the vicinity of an area called Tinggede, on the outskirts of Palu. This is the Planet Labs image of the landslide:-
This is the landslide that was caught on the video with the transmission tower. This is a very large slide – I estimate that the headscarp is about 1 km across, and the slide is over 3 km in length. On the east side of the scarp is a canal, which has breached, but my reading of the situation is that the breach happened because of the slide, not vice-versa. This is a flow type slide (rather than a lateral spread), with many houses being involved. The level of destruction is almost complete, judging by the imagery. I find it hard to believe that there was not significant loss of life here.
If that is bad, the second landslide in the area of Palu is worse:-
This landslide is just to the south of the Palu airport runway. Again, it is very large. I estimate that the scarp width is about 1.5 km, and the length is over 2.5 km. There are two main components, and the image needs careful inspection. What captures the eye initially is the area of bare soil in the landslide source. But look to the west of this – you will see a lobe structure extending almost to the edge of the image, which is the landslide deposit. This consists of soil and debris from the landslide, but also huge numbers of buildings, compressed together and wrecked. There is a second lobe, albeit smaller, to the south. The level of destruction to vast number of properties is vast. Again, it is hard to believe that there is not very significant loss of life here.
The third landslide is smaller, but far from trivial. This landslide is close to the city centre, with movement from the west towards the east:-
This landslide in Palu is about 500 m wide and about 1km long. Again, the bare soil of the source draws attention, but the vast number of houses compressed into a debris zone to the east is also evident, extending to the edge of the image. The level of destruction is very high.
All three of these landslides in Palu appear to be shallow slides on comparatively low angle slopes. All are close to the fault trace. In this area there are few other large landslides (there may be many other ones that cannot be resolved on this imagery).
Further to the south there are huge numbers of smaller landslides. These will feature in a future post.
Planet Team (2018). Planet Application Program Interface: In Space for Life on Earth. San Francisco, CA. https://api.planet.com
1 October 2018
Lateral spreads from the 2018 Sulawesi earthquake
NB: there is an update on these landslides on my post of 2nd October. These landslides are more complex than initially appeared to be the case).
The impact of the 2018 Sulawesi earthquake, which struck the isthmus of the Minahassa Peninsula in Indonesia on Friday 28th September 2018, is still being determined. The most serious impact appears to have been a significant series of tsunamis; initial Planet Labs imagery shows very widespread destruction to the coastal margins, once again. This will be a tsunami that will need to be studied in detail – the strike-slip mechanism of the earthquake is not one that would normally generate such a wave, although the 3D topography of the bay through which the fault runs may have allowed substantial changes in local volume, allowing the wave to form. However, the most likely explanation would seem to be one or more large submarine landslides, perhaps not surprising in an area with very sediment production in the tropical climate (which can generate large, unstable piles of sediment in the shallow marine environment). We will find out more in due course, and can only speculate at present.
In terms of terrestrial landslides, I would not expect this to be a particularly significant event, even though it has happened in tropical, hilly environment. Two factors should mitigate against very large numbers of landslides. First, the faulting mechanism seems to be associated with smaller numbers of landslides in general, based upon previous studies. And second, much of the surface trace of the fault is in the sea. Strike-slip faults generate landslides in very close proximity to the fault trace, and comparatively few at even intermediate distances. In this case, the amount of significant terrestrial topography close to the fault seems very limited. There will be landslides, but probably not in the huge numbers that we have seen elsewhere.
Two videos have emerged though that appear to show lateral spreads occurring during, or immediately after, the 2018 Sulawesi earthquake. These are remarkable, and truly terrifying, recordings. They have been posted on Twitter, and have drawn some criticism for being too graphic. I am not quite sure why this is the case, they do not seem to be unreasonable to me, but I note the concerns. The first is somewhat shaky (understandably), but remarkable for showing large structures in motion, including a transmission tower:-
There is a version on Twitter, posted by @patton_cascadia, that corrects some of the rotations in the recording. It seems reasonably likely that this is a lateral spread type landslide, perhaps associated with a quick clay or dynamic liquefaction process. Unfortunately at present the location is unclear. The second is perhaps more clear in terms of process, and is equally dramatic:-
The absolute terror of those involved is obvious. As yet the location of the events, and indeed whether this is the same one or two different locations, is uncertain. This second video captures the aftermath of the earthquake well, and in particular the stunned disbelief of those involved in the 2018 Sulawesi earthquake.
NB: there is an update on these landslides on my post of 2nd October. These landslides are more complex than initially appeared to be the case).