3 October 2014
Prehistoric rockfalls in Christchurch, New Zealand
One of the most interesting aspects of the Canterbury (Christchurch) earthquake sequence of 2010-2011 was the rockfalls that the shaking triggered in the Port Hills area on the edge of the city. I’ve covered those impressive and highly destructive events on a number of occasions. The great challenge after the earthquake was assessing the future rockfall hazard; a task that was undertaken with impressive rigour by GNS Science.
In assessing future rockfall hazard, a key element is evaluating the likely occurrence of the various trigger events. In Christchurch there were three key processes under consideration – a repeat of a local earthquake like the Canterbury sequence; a distant but larger earthquake, which in South Island is most likely to be a rupture on the Alpine Fault; or a large rainfall event. In each case much work has been undertake to evaluate the return periods of these trigger events, but for the Alpine Fault earthquake a key question was the degree to which a large earthquake would induce collapse events. This is challenging – we would expect that a distant very large earthquake would induce less intense shaking but for a longer duration when compared with a local event, but to what extent is this likely to trigger a large-scale rockfall?
Dating prehistoric rockfalls
In a paper that is in press for Geology, but is available for download from the author’s website, Benjamin Mackey and Mark Quigley from the University of Canterbury have tried to address this problem by dating prehistoric rockfalls at a site in Rapaki in the Port Hills. This site was one of the locations in which rockfalls caused eyebrow-raising damage, in particular when a boulder went straight through a house, out the other side and on down the hill. Interestingly, the slope was also littered with other boulders from previous rockfalls. Mackey and Quigley identified boulders from previous rockfalls that were of a similar size to those that detached in 2011 and undertook a dating exercise on them. The approach used is new and novel dating technique that gives a duration over which a rock surface has been exposed to cosmic rays. The reasoning is that a boulder is shielded from cosmic rays until the rockfall event. From this point onwards the top surface is bombarded; the dating system allows this length of time to be determined. Whilst that might sound trivial, in reality the technique is very challenging and experimental.
The results are really interesting. In essence the technique suggests that the last big rockfall event at this site occurred about 7,000 years ago, with the possibility of an earlier one at about 13,000 years ago. Earlier rockfall events than this are likely to have occurred, but these boulders are now buried in the slope. The big, distant earthquakes happen much more frequently than this – for example the Alpine Fault generates a large earthquake about every 300-400 years and the Porters Pass Fault generates one every 1500-2000 years. Thie implication is that none of these events in the last 7,000 years has generated wide scale rockfall activity at Rapaki.
So, it seems likely that the prehistoric rockfalls at Rapaki were generated by local earthquakes similar to the Canterbury earthquake sequence. And of course the data might indicate that these events only occur every 6,000 to 7,000 years on average, which is reassuring. Note however that a single site gives limited information about the seismic hazard as a whole – Mackey and Quigley point out that the fault generating a Mw=6.0 earthquake would have to be within 10 km and a Mw=7.0 within 20 km of the site to generate sufficient shaking to trigger rockfalls at Rapaki. Other faults in this area but beyond these distances would not generate enough shaking to generate rockfalls at Rapaki. So the return period of earthquakes of this size across the Canterbury area may be shorter than this if there are other active faults.
There is a news report about this research here. Mark Quigley should be congratulated or engaging with the comments at the foot of the article and for trying to explain his research. Some of the other comments are deeply frustrating.
Mackey, B., and Quigley, M., (2015 in press) Strong proximal earthquakes revealed by cosmogenic 3He dating of prehistoric rockfalls, Christchurch, New Zealand, Geology (PDF here)
28 September 2014
A landslide in Rize, Turkey
I know very little about this landslide, which apparently occurred in Rize, Turkey on 25th September:
The size of the blocks that slipped in quite impressive. It may be significant that the slide occurred in the section of road for which there is no masonry wall supporting the cut slope.
24 September 2014
A couple of weeks ago, and with great fanfare, it was announced that the diversion roads around the Sunkoshi landslide were now open and the traffic was flowing again on the Arniko Highway, linking Kathmandu with Tibet. The timing was important as this is a key route for the flow of goods into Nepal, and there was real anxiety about the effects of having large numbers of container lorries stuck to the north of the border. From the very start I was concerned that these roads would be very vulnerable to landslides in heavy rainfall. In the last few days news about the flow of goods has gone very quiet, coinciding with a period of heavy rainfall. The obvious implications were that the roads were blocked once more, but there was little hard evidence.
In the last few days geotechnical engineer Giorgio Maderni travelled along the highway. He has very kindly sent the following images to me (note that he retains copyright). The first shows the old Arniko Highway upstream of the dam. The level of the water prior to the breach is clear, as is the damage that the road has suffered:
Further downstream the level of damage to the road and infrastructure, and the slow progress in putting it right, is clear:
The size of the breach of the landslide dam is impressive:
The very poor state of the track across the landslide debris is clear from this image. In the distance a long queue of trucks waiting to cross the landslide can be seen.
And this image shows the state of the road in more detail. It will be a difficult task to get this to a state in which it is possible to see two-way movement of large numbers of trucks:
At present there is no news about the alternative roads, but given that the trucks are queued up the assumption must be that these are also in a poor state,
23 September 2014
Joshua Tree mudslide
Last week a mudslide, triggered by heavy rainfall, swept through Joshua Tree in California. This video shows an event on the same day at Joshua Tree, though I’m not sure if it is the same event:
The mudslide swept through the small town, and killed one person who, according to the reports, attempted to drive across the channel before becoming stuck. The mudslide struck his car, killing him. In the town, the mudslide struck the house of the Scarcy family, sweeping through the building. Breitbart has an image of the aftermath, showing how the mudslide demolished the rear wall of the building:
In the bedroom at the back of the house, one year old Tristan Scarcy was asleep when the mudslide struck the house. By the time his parents were able to get into the room he had been washed away. The likelihood of anyone, let alone an infant, surviving being caught up in a mudslide is very low, Tristan was found alive about 60 m downstream of the house. He had some cuts and bruises, and some debris in his lungs, but he’ll make a full recovery.
19 September 2014
Moving to the University of East Anglia
Regular readers will have noted my lack of blog posts in recent weeks, for which I apologise. As you may be aware I have been changing jobs – I gave up the Wilson Chair in Hazard and Risk at Durham University last Friday, and on Monday took up the position of Pro-Vice-Chancellor (Research and Enterprise) at the University of East Anglia. The last week has been a whirl of introductory meetings, all of which have been great fun. This hasn’t left much time for the blog, but as I now feel established I will be returning to previous levels of activity.
So to get things underway again, two new landslide videos have appeared on Youtube in the last week:
A lucky escape in Costa Rica
In the last few days very heavy rainfall in Costa Rica has triggered extensive landsliding. This video shows a very lucky escape for a motorcyclist:
It is of course only a small landslide, but illustrates nicely the dangers posed by even apparently superficial soil slides.
Ongoing landslides in northern India
Meanwhile northern India and Pakistan have been affected by heavy rainfall that has caused extensive damage (though has received very little coverage). This landslide in Ramban District appears to be causing considerable disruption to the road:
8 September 2014
The Sunkoshi landslide dam breach event
On Saturday night, apparently without warning, the landslide dam on the Sunkoshi River in Nepal abruptly breached, allowing the lake to mostly drain. This event appears to have occurred without warning, such that one of the Nepal Army excavators was washed away. The breach event appears to have been natural – a response to heavy rainfall – rather than a result of the attempts by the Nepal Army to widen the channel. Whilst some damage has been reported downstream, the flood does not appear to have been exceptionally large, so no casualties have been reported. Republica has a very nice set of photographs of the breached lake and dam:
Whilst Kapil Dhital (@bewitchkapil) tweeted this image of the lake this morning:
So what caused the breach? This is the gauging data for the Bharabise station (upstream of the dam) and the Pachuwaghat station (downstream). Note that the two sets of data use different y-axes. The flood peak generated by the breach event is very clear at Pachuwarghat, although this is not a very high peak. The Bharabise station probably responded to rainfall – there is a dramatic increase in discharge a few hours before the flood wave reached Pachuwarghat. Thus, it appears that the breach was initiated by this sudden increase in river flow in response to heavy rainfall:
This is clearer when the data for the few hours before the breach event are plotted:
In many ways this is the best possible outcome. The release of most of the water means that the hazard has in effect been removed, but without causing a huge amount of damage downstream. Unfortunately the huge amount of resource and effort that has gone into building the two new roads has been wasted. Once again I would reiterate my opinion that a focus on deepening the channel, rather than widening it and building new roads, would have been a much more effective way to manage this hazard. That the dam breached in a rainfall-induced event suggests that a targeted effort to deepen the channel could have resolved the crisis much earlier.
24 August 2014
Sunkoshi landslide – draining the lake
The Nepal Army is continuing in its attempts to drain the Sunkoshi landslide-dammed lake. Whilst the threat of a breach event has now subsided (though it has not gone completely of course), the major emphasis is on re-establishing the trade link to Tibet. There can be little doubt that considerable effort has been expended, but the results are shown in these two photographs of the lake, tweeted by Kapil Dhital. This is the lake on 6th August:
And this is the image he tweeted yesterday – 17 days later:
One really does not need to be a trained photographic interpretation expert to see that the lake level has not fallen, despite the best efforts of the Nepal Army. I have done a quick search on the Republica and other Nepal websites for news stories about the lake level behind the Sunkoshi landslide dam:
On 8th August, Republica reported that “The water level in the Sunkoshi River in Sindhupalchowk that increased last Saturday after a massive landslide blocked the river, has started to increase even after the Nepal Army blasted outlets…The water flow in the river was recorded at 180 cusec on the day of the landslide, but now it has reached 300 cusec, Singh said quoting the army technicians.”
On 9th August: “The inflow of water in the artificial lake was 205 cusec while the outflow was 210 cusec, and the water level has decreased by seven centimeters by Saturday, according to a statement issued by NA´s Directorate of Public Relations. The army plans to release accumulated water in the artificial lake, and resume the obstructed Kodari highway at the earliest, the statement added.”
On 10th August: “The water level in the artificial lake formed after the recent landslide occurred at Jure, Sindhupalchowk blocking the Sunkoshi River has increased. The Nepal Army (NA) team carried out three blasts targeting the blocked sites today to ease water flow of artificial lake, said Deputy Superintendent of Police Bharat Bahadur Bohora.”
On 11th August: “If we managed to reduce water level by 10 meters, we can bring the submerged road into operation within another three days,” Sitaula [secretary of the Ministry of Physical Infrastructure and Transport] said. “As the dam was filled in 11 hours, we can drain it out by doubling water flow downstream.”
On 13th August: “Nepal Army on Wednesday carried out eight blasts to resume the regular course of the Sunkoshi River that was blocked following a massive landslide at Jure of Ramche VDC-5 in Sindhupalchowk district on August 2…As per the measurement carried out in the afternoon the internal water current in the artificial lake is 255 cubic sec while the external current stands at 397 cubic sec.”
On 17th August: “The water level in artificial lake formed after massive landslide at Jure blocked Sunkoshi River, is gradually decreasing with increase in water outflow from the reservoir after Nepal Army (NA) constructed another channel to drain out dammed water…According to NA Brigadier General Ashwin Kumar Thapa, the water level has decreased by 75 centimeters after excavators were used to increase water outflow. “The water level in the dam is decreasing at the rate of two centimeters per hour,” he said. He further informed that water inflow in the artificial lake is 180-200 cumec, while outflow of water from the reservoir is 350 cumec “The current water inflow rate is normal rate for Sunkoshi River during Monsoon,” he said. During three-days of torrential downpour both water inflow and outflow were high with outflow and inflow rate standing at 330 cusec and 530 cusec respectively.”
On 19th August a different message emerges: “Although 900,000 kg/meter water is exerting pressure on the landslide dam [note - this statistic cannot be correct for a lake less than 50 m deep], the huge rock material is strong enough to withstand the pressure, the NA said organizing a press conference at its headquarters in Kathmandu. “We cannot open large outlets for the pooled water as that may cause more flooding downstream,” Geologist from the Tribuvan University Ranjan Kumar Dahal said. Also, it is very costly to completely remove the landslide debris, he said. He said it might cost Rs 450 million to remove the whole debris, which is almost impossible. The water level above the landslide dam has so far decreased just by a meter. NA also clarified that it can´t carry any strong blast to make the water exit from the dam.”
On 20th August, eKantipur reported that: “The Nepal Army team, entrusted to drain the landslide-dammed lake in the Sunkoshi river, plans to bring down the water level by 15 metres in the next 30 days, said army officials…“In the last few days, we’ve been able to reduce the level of water by a metre,” said Niranjan Kumar Shrestha, assistant spokesperson for the Army. The Army plans to widen the smaller of the channels currently draining the lake to bring down the water volume.”
On 23rd August the Himalayan Times reported that: “In a bid to ease outflow of water…Nepal Army carried out two controlled blasts on Saturday. NA Directorate of Public Relations said the average inflow in the artificial lake as of this afternoon is 187 cusec whilst the average outflow is 280 cusec”
On 24th August the Himalayan Times reported that: “Issuing a statement on Monday, NA’s Directorate of Public Relations said the Army’s efforts to remove the obstruction…are also ongoing. According the the statement, while the inflow of water in the river was measured at 193 cusec this afternoon, the outflow was recorded at 303 cusec”
In recent days the media has repeatedly reported that the outflow is greater (often much greater) that the inflow. If so the water level should be declining quickly, which is clearly not the case. Thus, the figures must be wrong or misleading, so I am bemused as to why they are repeatedly reported in this way.
The sad reality is that all of the immense effort that has gone into building the two channels has had no impact on the net lake level. The reasons are probably demonstrated in this image from the Nepal Army:
Widening the channel is of course no bad thing, but to achieve a drawdown of the lake level behind the Sunkoshi landslide dam the channel bed needs to be deepened. The lake level will not change in aggregate without deepening the channel.
Finally, Kapil Dhital tweeted this image of the private road being constructed by local entrepreneurs:
Whilst the frustration that has driven the local people to do this is understandable, the environmental consequences of this road are clear. And, of course, this road will be very prone to landslides, especially as the lake will have elevated the water level in the slopes. I would not want to be on this road is heavy rainfall.
23 August 2014
Dariali Gorge landslide number 2
In May this year a very large landslide occurred at Devdoraki in Georgia, blocking the strategically-important Dariali Gorge. This transpired to be a large rockslope failure from directly below the glacier, which turned into a high speed, long runout flow. I was lucky enough to be able to feature some images of the landslide.
On 20th August, another landslide occurred at the same site, once again blocking the Dariali Gorge. This landslide, which is reported to have originated at the glacier, is reported to have been larger. The trigger was reportedly heavy rainfall. The landslide, and the resulting damage, are sufficiently serious to have needed a visit to the site by the President of Georgia. Two people lost their lives in the landslide, both workers from the nearby HEP project.
There are some images on the internet of the landslide. The best that I have found are on this Georgian news website. At times it is not entirely clear what they are showing, but they are interesting nonetheless. I suspect that these two images show the landslide deposit itself – it appears that the landslide blocked the river, which has then breached the dam to create an outburst flood(?):
An interesting aspect of this is the landslide deposit marks on the valley walls, which suggest that a very substantial volume of material has been eroded away. It is not clear to me whether this is from the May or the August landslide events.
This image seems to show downstream damage from the outburst flood:
In many ways this image is the most interesting. It is possible that it shows the channel through which the landslide travelled to reach the Dariali Valley, although this is speculation at best:
It would be fascinating to see a set of images of the source and track of this landslide – to have two very mobile, long runout landslides in the same valley in a short period is both unusual and intriguing. Does anyone know if such a set of images has been posted online?
21 August 2014
On Wednesday, a suburb of Hiroshima in Japan was struck by a series of catastrophic rainfall-induced landslides, which resulted from a period of prolonged and exceptionally heavy rainfall. These landslides are reported to have killed 39 people, with as many as a further 43 people reported to be missing. Sadly, some of the deaths appear to be rescue staff who were overcome by subsequent slides.
From a technical perspective, the best images I have found of these landslides are on the IBT website, which includes this overview of the site:
These landslides appear to be channelised debris/mud flows. It appears that they have started as small failures along watercourses that have entrained large volumes of material to become highly destructive flows. It appears that they have mostly consisted of very fine-grained materials, which is consistent with the bedrock geology, which is reportedly weathered granite.
Unfortunately, there is further heavy rainfall forecast for this area, and at the time of writing Robert Speta tweeted the following:
Line of storms approaching Hiroshima. Rescue efforts have been suspended. pic.twitter.com/6nHPe9UADN
— Robert Speta (@robertspeta) August 21, 2014
The image shows a line of storms approaching the area:
Whilst it seems unlikely that this rain will cause as much damage in Hiroshima itself, landslide and flood warnings are in force across the region and there have been many evacuations.
18 August 2014
ICIMOD has released a report about the Sunkoshi landslide, which has been reported in The Himalayan Times. The most interesting aspect may be some statistics on the landslide, which appeared in a tweet by Kanak Mani Dixit, which said:
ICIMOD collation re Sunkosi landslide: dam length 409m, width 106m, height 55m, volume/material: 6mill m3, lake length 3km, volume 8mill m3.
— Kanak Mani Dixit (@KanakManiDixit) August 18, 2014
These statistics are in line with those generated elsewhere, but are the first to provide a complete set. The main thrust of the ICIMOD report (which I’ve not seen) seems to be that more investigation of the Sunkoshi landslide itself, and more detailed landslide hazard mapping in general, are required. It is hard to disagree with these suggestions.
Meanwhile, the Nepal Army continue to try to drain the lake, but the degree to which they are being successful is very unclear. There is a profound deficit of reliable information about the state of play at the dam site – over the last two weeks there has been a number of reports that the lake level was falling, all of which have proven to be incorrect. The latest reports, on Sunday, were that:
The water level in artificial lake formed after massive landslide at Jure blocked Sunkoshi River, is gradually decreasing with increase in water outflow from the reservoir after Nepal Army (NA) constructed another channel to drain out dammed water. Decline in water level could be clearly witnessed on Sunday as clouds separated revealing blue sky after three days of torrential downpour came to an end.
According to NA Brigadier General Ashwin Kumar Thapa, the water level has decreased by 75 centimeters after excavators were used to increase water outflow. “The water level in the dam is decreasing at the rate of two centimeters per hour,” he said. He further informed that water inflow in the artificial lake is 180-200 cusec, while outflow of water from the reservoir is 350 cusec. “The current water inflow rate is normal rate for Sunkoshi River during Monsoon,” he said.
During three-days of torrential downpour both water inflow and outflow were high with outflow and inflow rate standing at 330 causes and 530 causes respectively.
But what does this mean? The figures above suggest that during the heavy rainfall the inflow exceeded the outflow, which means that the lake level must have risen. Then, when the rain stopped the inflow reduced to a level below outflow, so the lake level must have fallen. This is quite normal dam hydraulics. A key question is whether the net (i.e. low flow) level of the lake has fallen. The data being provided are not adequate to judge this.
The key to reducing the level of the lake has to be to lower the bed of the channel. Widening the channel, or creating a new one, will make the discharge of storm flow more efficient – i.e. the lake level won’t rise so much in the next heavy rainfall event. But to meaningfully draw down the lake the channel bed has to be lowered. Now this is of course risky – in particular if the bed of the stream is being armoured by large boulders, then these have to be blasted. But, this could be the case that beneath the boulders lie fine-grained deposits that might erode quickly once exposed. So this sort of operation is very sensitive and challenging, and requires high levels of expertise. And this is why I still contend that the Nepal Army should be seeking assistance from groups that have managed these problems before, most notably the Chinese. They might note the speed and efficiency of the operation to drain the valley blocking landslide in Yunnan Province, despite the very challenging access problems.
If the Nepal Army wants to drain the lake then the engineers have to deepen the channel. The challenge is to do this is a safe way. Why not get assistance from the Chinese team whose raison d’etre is to manage valley blocking landslides? After all, China wants this important trade link open too.
Finally, an additional dimension is the rapidly escalating pressure to reopen the road to Tibet. The Nepal Engineers Association is pressing for concerted efforts to drain the lake and construct an alternative road to bypass the Sunkoshi landslide. In a separate development, a group of local entrepreneurs have started to construct a new alignment with a bulldozer:
Whilst I can understand the frustration that is driving this, the track shown above is both environmentally catastrophic and a recipe for further landslides. It seems unlikely that it is a sustainable solution.