This blog provides a commentary on landslide events occurring worldwide, including the landslides themselves, latest research, and conferences and meetings. The blog is written on a personal basis by Dave Petley, who is the Wilson Professor of Hazard and Risk in the Department of Geography at Durham University in the United Kingdom.

This blog is a personal project that does not seek to represent Durham University.

Friday, 30 April 2010

New landslide video from Brazil

Live Leak has an impressive video of a large debris flow striking and destroying a house, reportedly in Blumenau, Santa Catarina, Brazil :

Thursday, 29 April 2010

Latest update on the Attabad landslide in Hunza

The lake level at the Attabad landslide continues to rise at about 55 cm per day, reflecting the increased level of inflow from snowmelt.  NDMA and Focus now conduct joint surveys of the freeboard and have agreed a broadly common figure - currently about 23 m.  At current rates this continues to indicate an overtopping date of early June (full details on the monitoring site):

Seepage remains a concern, although at the moment the rate of increase remains linear with time over recent days:

Meanwhile, the length of the lake is now extending quite rapidly as the valley gradient at the head of the lake is low, as this Pamir Times image shows:

I remain deeply concerned about the potential for failure through seepage or another landslide, and of what will happen at the point of overtopping.

Monday, 26 April 2010

The mechanism of the Highway 3 landslide in Taiwan

That bastion of "quality" journalism, the Daily Mail, has an article about the Highway 3 landslide that includes a set of high quality images of the site:

The article correctly identifies the failure as being a dip-slope slide:

"The hill had a dip slop [sic - should be slope] on the side nearest the motorway. The other side of dip slopes are steep and irregular, while the slope itself makes it easier for rocks to slide down, experts said.  The Ministry of Transportation said it as [sic] investigating up to 20 similar dip slopes near major roads in Taiwan."

A dipslope means that the beds in the slope were inclined sub-parallel to the slope surface, creating a plane of weakness on which sliding can occur, like this (from here):

The sliding surface can be clearly seen in the second image, though it was less steeply inclined than in the cartoon above.  The inclination of the beds parallel to this surface on the displaced block is very clear.  Excavation of the debris will have to be done very carefully indeed to avoid the upslope material slipping down onto the workers.

The lack of a trigger implies a progressive failure - but I am surprised in that case that signs of distress were not observed prior to failure. 

Finally, it is interesting that the Ministry of Transportation says it is investigating "up to 20 similar dip slopes near major roads in Taiwan".  Having spent a great deal of time in Taiwan I would be surprised if there were just 20 such slopes, but maybe these are just the very large ones by major highways.  The 18th August 1997 Lincoln Mansions landslide in Taipei, which killed 28 people, was also a dipslope failure, and the 2009 Shiaolin disaster was a wedge failure with a dipslope component.

The government really needs to get a grip of the management of slopes in Taiwan.

Images of the Highway 3 landslide in Taiwan

I have still not managed to track down some really good images of the Highway 3 landslide in Taiwan, but for now here are two pictures from the China Post:

More later I hope.

Sunday, 25 April 2010

Aerial video film of the Highway landslide in Taiwan

Youtube now has a helicopter video of the Highway landslide in Taiwan:

It is an extraordinary landslide, apparently being a deep translational slide that has displaced a great raft of material fronted by the cutslope shown in the previous post.  I will be interested to see some decent images in the morning!

The location of the Taiwan highway landslide

I have I think identified the location of the Taiwan highway landslide today.  Based upon this image (source):

The location appears to be here:

It appears that the failure has occurred on the cutslope shown in the centre of the image, with the margin of the slide being close to the bridge (which can be seen in the first image on top of the debris).  Thus, it appears to be a large cut slope failure.

This is certainly not the first time a cut slope has failed on a major highway, as these two examples show.  First, the Pigeon Gorge slide in North Carolina last year:

And second the 2003 Bukit Lanjan landslide in Malaysia:

However this slide in Taiwan is unusually large, with an interesting mechanism.

A huge landslide on Freeway No.3 in Taiwan

Taiwan today suffered an extraordinary landslide in Highway number 3, which links Taipei with Keelung.   The landslide appears to have completely buried the road for a distance of about 300 m.  The depth of burial looks to be more than 10 m.  There are reports that there were a number of cars on the road at the time - if so, the chances of survival are slim.

At the moment the available pictures of this event are poor - no doubt much better ones will emerge tomorrow as this will be a very big story in Taiwan.  The best I have been able to find so far are these picture from TVBS:

There is also a youtube video of a news report from the slide site, which you should be able to view below:

Compare the video above with this image of the site before failure to get an idea of just how large this slide is:

Interestingly, there was no recorded rainfall or earthquakes at the time of the collapse.  The mechanism and nature of the failure will be very interesting.

I have been arguing for some time that Taiwan needs to start managing its slopes better.  Will this finally see the authorities take some action?

Hat tip to Chingying Tsou of Kyoto University for pointing this one out and providing the links.

Friday, 23 April 2010

On morbidity and mortality in landslide disasters

Landslides kill thousands of people each year across the world but, strangely, there are very few studies of the causes of the injuries and deaths that people suffer when they are affected by a landslide. This contrasts with avalanches, which has a long track publication history of causes of mortality.  This is important in the context of treatment of victims - in particular, where a rescue is ongoing, the medical practitioners need to be able to prepare for the likely state of those who might be recovered.

In this context, a fascinating article, Sanchez et al. (2009), was published in the journal Disasters late last year in which a group of medical and well-being experts examined the causes of morbidity and mortality associated with a landslide disaster that occurred in Micronesia in 2002.  The event was a series of rainfall induced debris flows that occurred on the islands of Chuuk on 2nd July 2002 as a result of the passage of a tropical storm named Chata'an.  The storm reportedly triggered 265 landslides, 12 of which caused a total of 43 deaths and 48 injuries that required hospital treatment.  There is an nice pdf of a presentation by Herman Semes Jr on the background to this disaster here, from which the following images are taken:

The research team interviewed 52 survivors of the disaster, plus a number of eye-witnesses and also reviewed the death certificates of those who lost their lives.  The landslides themselves were slumps that transitioned into debris flows, the largest of which had a volume of 1.5 million cubic metres.

There are some really interesting results of the study, which I will summarise below:

Cause of death
Even though the slides were rapid debris flows, 39 out of the 43 deaths occurred through asphyxiation, with only four resulting from trauma.  Even for those who died outside of buildings, the majority of deaths occurred through asphyxiation.

Types of injuries
For those 31 victims who were injured by the landslides and required hospital treatment, and for whom information on their injuries was available, 16 suffered lacerations, nine contusions/abrasions, four concussions and only two fractures.

Gender of victims
About 58% of those killed by landslides were female, even though women represent 39% of the general population. 

The age of victims
The fatalities were recorded on individuals aged between two months and 84 years, with a median age of 14 years old.  The comparison group of survivors had a median age of 29 years, suggesting that landslides preferentially killed the young. 

Location of victims
Very surprisingly, the study found that "being inside a house or building during the landslide was associated with a higher risk of mortality", although the difference in risk is not statistically significant.

This study is really interesting - we desperately need more research of this type!  The observation that most of the fatalities occurred as a result of asphyxiation is consistent with avalanche research, but given the greater potential for the materials in landslides to cause trauma, it is a slightly surprising observation.  It is also notable that the landslides appear to have preferentially killed the young - probably a reflection of ability to escape an oncoming event.  Finally, the observation that structures offered no obvious protection is also interesting, but may reflect both the poor quality of buildings in a less developed country, and the fact that a structure may prevent an individual from seeing / hearing an approaching landslide, and may inhibit escape once it occurs.

Sanchez C, Lee TS, Young S, Batts D, Benjamin J, & Malilay J (2009). Risk factors for mortality during the 2002 landslides in Chuuk, Federated States of Micronesia. Disasters, 33 (4), 705-20 PMID: 19459918

Monday, 19 April 2010

Some reflections on the Eyjafjallajoekull ash cloud

Over the weekend I have been watching and reading the media response to the aviation ban resulting from the Eyjafjallajoekull ash cloud, and reading some of the online discussions.  I would like to make four observations from a natural hazards perspective:

A. This event is the result of a "perfect storm".
I have been surprised by the lack of reflection by the media, including by some scientists (see below), that the current event really is the result of an extraordinary set of circumstances.  To generate the current crisis, three events had to occur simultaneously.
  1. The volcano had to suffer an eruption that puts large amounts of ash high into the atmosphere.  Many eruptions do not do this, which is why we rarely see these types of crises even though eruptions occur regularly around the world.  In fact this volcano has been erupting for a month, but the initial events did not inject ash into the high atmosphere. 
  2. The jetstream needed to be located over Iceland.  If the jetstream had been to the south then the ash would not have been brought to Europe.  
  3. The atmospheric conditions over Europe had to be extraordinarily stable.  Those of us who live here know that what we are seeing in northern Europe is not typical (though it is not all that rare either).  Usually we see a conveyor belt of low pressure systems from the southwest.  At the moment we have stable high pressure and no wind, which means the ash is lingering.

If any of these three conditions had been different then this event would not have occurred.  It is the rare (and quite unlikely) confluence of these three conditions that have led to where we are.

2. Some pronouncements by scientists really have not helped. 
The media are now in a pickle as they have to keep the story on the front page, but finding new angles is a challenge.  For this reason, alarming pronouncements by scientists are a goldmine.  A number of prominent scientists (no names) have been warning that the eruption could go on for months, or even years, and that Hekla might cause an even bigger eruption.  For goodness sake, please stop!  There are no indications that Hekla is about to erupt; the linkage between Eyjafjallajoekull and Katla are not clear; if it did erupt, a similar problem is not inevitable if the jetstream or the European weather conditions are different; and it is not helping to raise concerns about another eruption.  In a similar manner, suggesting that the volcano might cause us problems for months or years is not helping either.  Yes, an eruption may continue, but the chances that it will continue to inject ash into the high atmosphere are not high, and most of the time the ash wouldn't be brought over Europe.

Unfortunately, doom-laden predictions undermine the credibility of science and scientists when the situation calms down, as it inevitably will.  This is not helpful.

3. Environmental science / natural hazards research is grossly underfunded
The third lesson is that these events can have a big impact on our lives and our economy, but our investment in research into them is pitiful.  The limited capability that we have in Europe to collect good samples from the upper atmosphere is all to obvious, for example.  A comparatively small investment in the science would reap large rewards at a time like this, and would provide financial benefits many times greater than the costs.  If there is one thing that should come out of this it is a recognition that we have to invest in hazards research.

4. The general understanding of risk is very poor
It has been extraordinary to see pronouncements that the successful operation of the test flights over the weekend means that the flight ban is an over-reaction.  The fact that a small number of flights can successfully operate does not mean that the skies are safe.  It may be that the safety of, say, only 1 flight in 10,000 is put at risk by the ash, caused by very isolated pockets od denser / more abrasive material.  As there are 28,000 flights per day in Europe, this would not be an acceptable risk.  Unfortunately, due to the lack of research we do not know what true probabilities, but the Finnish F-18 engines suggest that they are far from trivial.

Sunday, 18 April 2010

Hazard management in Taiwan

The Taipei Times has a very interesting editorial (17th April) that reflects upon hazard management in Taiwan.  It returns to the edtorial that I wrote for the paper last summer, based upon a posting that I put on this blog.  The need to take action to improvement the management of hazards in Taiwan is undeniable.  As we approach a new typhoon season the legacy of the continued inaction could be all too apparent again.

Worrying signs on the landslide at Attabad in Hunza

The water level at the landslide dam at Attabad in Pakistan continues to rise (see the data on my Hunza landslide monitoring site), with the reported freeboard being less than 30 metres.  However, I have to report a very much more worrying development.  Dr Michele Comi and Dr Chiara Calligaris from Ev-K2-CNR are just returning from a visit to the site, and have sent me the picture below (reproduced with permission):

This sinkhole has apparently opened up on the upstream side of the dam.  The most likely cause of a sinkhole is that erosion is occurring in the core of the dam, although this could be localised.  However, given that seepage is now accelerating again, the signs are worrying.

Saturday, 17 April 2010

Extraordinary video of a Jokulhlaup in Iceland

A jokulhlaup is a sudden release of water from beneath a glacier.  One of the key triggers for Jokulhlaup is the eruption of a volcanic beneath an icecap.  It shouldn't be a surprise to hear that jokulhlaups have been triggered by the Eyjafjallajokull eruption that is causing such chaos across Europe (guess who was supposed to go to Hong Kong on Thursday...), and increasingly beyond.

Now, a jokulhlaup is not technically a landslide, but as these flood carry vast amounts of debris they are at the hyper-concentrated flow / debris flow end of the flood spectrum.  On that basis I thought it reasonable to show this fantastic video of a jokulhlaup cascading off the margins of the Eyjafjallajokull volcano on 14th April 2010:


Friday, 16 April 2010

Latest update on the Hunza (Attabad) landslide in Pakistan

I have today updated my monitoring site for the Hunza (Attabad) landslide.  However, there are a number of key developments that are worth noting:

1. The rate of seepage has started to increase once again
The graph of seepage against time shows that seepage is increasing rapidly at the moment, albeit at rates that are still only a fraction of the inflow rate:

The key factor here is the development of a new seepage point on the dam face (seepage point 4), which is now producing about 85 litres per second of water.  The location of the four seepage points are shown below (from Focus):

Seepage points 2 and 3 cause me little concern given their location.  Points 1 and 4 are clearly the ones that probably represent seepage from the lake.

2. The date of overtopping remains unclear
The outgoing NDMA chair recently estimated that overtopping could occur on 17th April (i.e. tomorrow).  This is highly unlikely.  On the monitoring site I try to estimate the date of overtopping based on the reported rate of reduction of the freeboard.  My current best estimate is mid-June.  However, the lake length has not really been increasing recently as the upstream section is currently steep (see image below from the Pamir Times:

Once this steep section is overcome the rate of rise may decline.  However, this may be offset by an increase in inflow as snowmelt develops.

3. Something strange is reported at Shishkat
Shishkat is near the head of the lake, probably best known for the location of the KKH bridge that has now been drowned (image from the Pamir Times):

Local people at Shiskat are reporting "underground blast-like sounds" in this area.  Of course there is little information about what these "blasts" might be, or indeed the reliability of the reports.  However, one possibility would be the development of slope instability on the banks of the lake.  Given that one failure mechanism for the dam is a slide into the lake and a wave that then overtops the dam, this needs to be checked out with some urgency.

Sheepfall - a hitherto unknown mass movement hazard

I spied this sign in my local paper, the Weardale Gazette, last night.  Inspired!

Wednesday, 14 April 2010

First news of the Qinghai Earthquake in China

This morning's Mw=6.9 earthquake in China is now reported to have killed at least 300 people, with the toll likely to rise over the next few hours.  Although the earthquake occurred in an area with a low population total and density, vulnerability of both buildings and the landscape to shaking is likely to be high, especially given the reported shallow nature (depth = 10 km) of the event.

The Google Earth imagery of the area shows a landscape that is sparsely populated with substantial mountains such that landslides may well be a serious issue.

The landscape appears to vary, with rolling hills in the south (image from here):

To the north the landscape is undoubtedly steeper and more rugged, and thus landslide-prone: (image from here)

The town in the foreground above is Jiegu (sometimes spelt Gyegu). Early reports suggest that up to 80% of the buildings in the town have been destroyed.  Images of the town before the earthquake suggest that it is likely to have been very vulnerable both to ground shaking and to landslides:

Given the remoteness of this area, and the likelihood of landslides on the roads, the delivery of aid is going to be a major challenge.  My earlier comments on the reporting of large earthquakes in mountain areas applies here once again (insert China for Haiti):

1. Everything stops at night. At the time of writing it is still night time on Haiti. In the aftermath of an earthquake electricity and power supplies are wiped out, so for the night time period it appears that the disaster is not as bad as is feared. As the sun comes up so the reports on the true picture start to emerge, and the fatality statistics start to increase rapidly. This increase will continue for several days at least, but may ultimately exceed the final toll;

2. The initial focus is often wrong. In the immediate aftermath of a disaster the initial focus of the media reports is often on the biggest city. This is rarely where the biggest impact has occurred, but it is most accessible so will be the focus of the reports.

3. No news is very, very bad news. The biggest impacts are often in rural areas with the highest levels of shaking. These areas had poor communications to start with, but when an earthquake strikes the roads become blocked, power is lost and there is no telephone service. Therefore, no news comes out for some time after the quake. The picture is actually the opposite of the obvious. If news starts to emerge quickly from those areas with the highest shaking then the picture is not as bad as we feared - at least some communications are open - although it may still be quite grim. If there is almost no news at all from the rural areas for a day or two, then the picture is probably very bad indeed, with almost all of the communications wiped out.

4. The media focus will quickly change to the foreign rescue teams. However, although these efforts are valuable, their overall impact is very small. The real work is actually done by local people - most rescues are made by untrained people in the first 24 hours - this should really be the focus.

Tuesday, 13 April 2010

Junk natural hazards science of the month - earthquake risks to the World Cup in South Africa

Although this is meant to be a blog on landslides, occasionally things appear in other areas of natural hazards that are so irritatingly ridiculous that I just can't let them go.  And so in the last few days a story has been doing the rounds about earthquake risk to the forthcoming football World Cup in South Africa.  For example,  The Daily Star newspaper in the UK (not necessarily worth reading on a normal day) is carrying the story:

There are some extraordinary quotes, from the source of the story, Dr Chris Hartnady:

"A major earthquake disaster in the region is inevitable because wide areas of southern Africa are affected by the slow, southward spread of the East African rift system...It is not a question of if, but when. The consequences would be so expensive in terms of mortality and economic cost that the risk of being ill-prepared is unacceptably high.”

So lets take a look at the seismic record for South Africa.  First, here is the USGS seismic hazard map for Africa:

Whilst the seismic hazard for South Africa is not zero, it is comparatively low. For comparison, here is the similar map for Europe:

You will see that South Africa's seismic hazard is similar to, and in many cases lower than, that of the UK.  Hardly a major source of concern.  Here is a map of earthquakes since 1990, again from the USGS:

Compare the number of recorded events in the Mediterranean area (top of the image) with that of South Africa to get an idea of the level of hazard in the latter.

Of course an earthquake can in theory happen anywhere and of course if the community is unprepared then the impacts of said event can be devastating.  However, the level of seismic hazard in South Africa is not high, and the likelihood of a large earthquake there during the four weeks of the World Cup is vanishingly small. 

Of greater concern is how any team can hope to cope with Argentina's maestro Lionel Messi!

Toyota Yaris vs rock...

The Mirror newspaper in the UK yesterday carried a dramatc illustration of the effect that a boulder can have on a car.  This car was parked by the side of the road in Hayfield in Derbyshire on Sunday when an eight tonne boulder slipped off a lorry.  Fortunately there was no-one in the car:

 Time to buy a new bottle of T-cut?  Picture from the Mirror

Monday, 12 April 2010

The location of the Italian train crash landslide

Thanks to David Bressan of the Cryology and Co blog for help in identifying the location of the landslide in Italy today that derailed a train, killing at least nine passengers.  The location is (48.632N, 10.882E) is shown on this Google Earth image:

 The landslide was small - only 400 cubic metres - but it doesn't take much to derail a train if it happens at the wrong time and in the wrong place, as this Il Giornale image of the slide shows:

Images of the Italian railway landslide today

A landslide (frana in Italian) occurred this morning at Merano, near to Bolzano in northern Italy, striking a train that was derailed.  Reports suggest that 11 9 people were killed and a further 30 were injured.  This is a Google Earth image of Merano with a railway line marked in black, but I am not sure where the accident occurred or even if it is in this image.

Il Giornale has some images of the accident site:

From the look of these images, this appears to be a shallow failure from a slope above a retaining wall.  Early suggestions are that the landslide was triggered by a leaking irrigation pipe, which saturated the soil, triggering failure.  Clearly that is just conjecture at this point.

This type of landslide-induced rail accident occurs occasionally.  I am reminded of this one in China in July last year.

Friday, 9 April 2010

The location of the Morro do Bumba landslide in Niteroi, Brazil

Thanks to a video posted on the BBC website, I have managed to locate the Niteroi landslide, which is a garbage dump slide that is believed to have killed over 200 people.  The slide is at the following coordinates:  22.893° S, -43.085°W

On a Google Earth image this area looks like this:

 A zoomed in view looks like this (the crown of the landslide is at the bottom of the image):

Thus, a perspective view of the landslide looks like this (looking from the toe towards the crown):

It is clear that a large number of houses have been destroyed.  Google Earth measurements suggest that the slide is about 300 metres long and 50-70 metres wide.    It is interesting that most of the crown appears to be located at a road (see the screenshot from the video below) - which might suggest that drainage was an issue in the causation of the failure:

The description of the slide as an old garbage dump is certainly correct judging by this image from O Globo:

The likelihood of surviving burial in a garbage landslide is very low given the density of the material, the release of methane gas and in some cases the release of heat from the decaying material - see my earlier post here on the hazards posed by garbage dump landslides.  Needless to say the recovery and clean up operation will be both very difficult and very unpleasant for those involved.

Latest Update on the Attabad landslide

Whilst my Hunza Landslide Monitoring website is the main location for the monitoring data of the landslide at Attabad, I will continue to put updates here every few days.  The current situation is that the lake level continues to rise at 30 to 40 cm per day.  This is perhaps slightly lower than expected as temperatures in this region remain unseasonably cold, meaning that snowmelt (and thus inflow) has been delayed.  Based upon the current rate of rise of the lake level and downcutting of the channel, the water should reach the spillway in two to three months.  This may shorten if the rate of inflow increases.

This graph shows the water level through time and the level of the spillway, based upon NDMA and Focus data:

Meanwhile, this one shows the rate of seepage at the three points plus the overall rate.  It is clear that the seepage rate has stabilised over the last few days:

Thursday, 8 April 2010

Stop Press: Over 200 people buried in last night's landslide in Niteroi, Brazil?

AP is reporting that over 200 people have been buried in last night's landslide in Niteroi, Brazil:

"A top Rio de Janeiro state security official says at least 200 people have been buried in the latest landslide to hit the area. Civil Defense subsecretary Pedro Machado tells Globo television that the slide plowed into a slum in the city of Niteroi. Machado said Thursday that anyone there at the time was likely killed instantly."

O Globo's report also confirms that the landslide did occur on an old garbage dump, which was abandoned in 1986.

If over 50 houses were struck by this slide in the middle of the night a death toll of this order is certainly possible.

Another major landslide in the Rio de Janeiro area of Brazil

 O Globo in Brazil is reporting that another large landslide (in Portuguese "deslizamento de terra" for those looking the original website) struck the Niteroi area overnight, burying up to 50 houses, following the multiple slides yesterday.  The slide (shown below in an image from O Globo) occurred at Viçoso Jardim, in the Niteroi suburb of Cubango.  At least five people have been killed, with an unknown number of people missing.  The site is reportedly an old garbage dump that has subsequently been occupied by illegal dwellings.

Overall the toll from this multiple landslide disaster is very high.  To date the confirmed number of fatalities is 150, with 135 people injured and 40 or so still missing.  The Civil Defense agency is reporting 806 rainfall induced events, most of them landslides of various types. The reader-led O Globo map of the distribution of the events is a fantastic resource.  Hopefully it will be visible here:

Visualizar O mapa da devastação no Rio em um mapa maior

The landslides vary in size and nature considerably, but some are quite large, as this AP image shows:

Wednesday, 7 April 2010

Widespread landslides in Rio de Janeiro and Niteroi in Brazil

Over the last few days the city of Rio de Janeiro, and its twin city Niteroi just across the water, have been affected by extremely intense rainfall - a reported 288 mm in 24 hours.  The upshot has been a series of flash floods and landslides that have killed over 100 people so far.  More heavy rain is expected today.

O Globo has produced an interactive map of the Rio and Niteroi regions in which readers are invited to add information about their area.  This gives a first order estimate of the impacts of the rainfall:

Alagamento = flooding;
Lixo, lama e avores = garbage, mud and trees
Deslizamento = landslide
Postos de deocoes = Donation stations

It appears that many of the slides have affected the shanty towns constructed on the hillsides around the cities.  Over the next few days I will try to collate a list of the major fatality inducing landslides, but for now
O Globo has these two images of landslides that have affected residential areas:

Strangely, it is often a slightly abstract image that captures the horror of the event.  O Globo also has this exceptionally powerful photograph:

Finally, it is interesting to note that this rainfall event appears not to have been captured by the TRMM landslide warning system:

24 hours of rainfall (the yellow circles indicate those areas considered to be at risk of landslides):

72 hours of rainfall (the yellow circles indicate those areas considered to be at risk of landslides):

More later.