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.

Monday, 30 June 2008

International Symposium of Landslides - Wenchuan (Sichuan) earthquake landslides

This week is the 10th International Symposium on Landslides and Engineered Slopes, which is a once every four years get together. This time round it is being held in Xi'an in China, which is particularly appropriate given the recent events a few hundred kilometres to the southwest. The organisers managed to timetable a session of two keynotes and six other presentations on landslides associated withe the earthquake. The session was pretty interesting and highlighted both the amazing job that the Chinese authorities and people did in the response and recovery stage after the earthquake, not least with respect to the valley blocking landslides, and the incredible amount of information that has been collected already. For example, 15,000 landslides have been mapped, hundreds of unstable slopes have been identified and in some cases hazardd assessments have been conducted. One speaker mentioned that 700 geologists have been working on landslides in the earthquake zone - if so then this is a remarkable response that few if any other countries could match.

I don't intend to cover all aspects of the presentations, not least because there was some inevitable and understandable duplication and contradictions between them. Highlights for me included:
1. The first speaker was the conference convenor, Prof. Chen, who presented a paper on behalf of Ning Liu, who was the engineer in charge of the successful draining of Tangjiashan. The presentation made the achievement of mitigating this slide all the more notable. Interestingly, the first part of the talk was on another valley blocking landslide, Yigong, which was successfully breached by the same team in June 2000. This landslide was much bigger than Tangjiashan - 300 million cunic metres, creating a lake of about 1.5 cubic kilometres. The same team, called the hydro-police, successfully drained that lake too, although the peak discharge was c.140,000
cumecs! Thus, the team that tackled Tangjiashan had experience of undertaking similar tasks. The approach used was exactly the same - i.e. to build a small channel and then to allow the water to slowly excavate it to allow the lake to drain away. They were confident that it would be OK as there were rock fragments and boulders in the landslide mass - and they were quite right of course.

He also stated that all of the other landslide lakes have now been resolved too. If so this is an unprecedented achievement for which the authorities and individuals involved deserve high praise.

2. Yin Yuping of the China Geological Survey gave a more general overview of the landslide problems, suggesting that the key issue from a slope stability perspective was probably the very high peak ground accelerations (although unfortunately these were estimated rather than empirical data being produced). He talked at length about Beichuan. The landslide that struck the old town (near the apex of the bend) killed 1600 people, partly through burial and partly as a result of a ferocious air blast that flattened every building bar one that it struck. The other was the slide that hit the New Town and buried the Middle School. 350 children were killed. Sadly it appears that this was a slope in a known state of instability - and indeed it seems that 16 piles had been installed on the slope on 2006 to improve safety.

He suggested that 15,000 landslides had been mapped to date (this looks a little conservative to me?) and that 9,000 unstable slopes had been identified. All of the landslide dams occurred close to the fault surface trace, but one large landslide occurred >300 km from the fault!

3. Qiang Xiu presented a study of the Donghekou landslide, which had a volume of 12 million cubic metres. It inundated three villages, killing about 300 people. The pictures of the landslide were very impressive indeed, although I think it is fair to say that more work is needed to understand the landslide mechanisms.

Tomorrow includes a series of keynote lectures, a round table discussion (in which I am involved) and a series of parallel sessions, all wrapped up with a cultural evening. I will try post the highlights (excluding the cultural bit) tomorrow.

Friday, 27 June 2008

TRMM and landslide monitoring

One of the most admirable science projects at NASA at the current time is an attempt to create a global landslide and flood warning system using the Tropical Rainfall Monitoring Mission satellite data. TRMM using a precipitation radar system to provide near real time data on the intensity of rainfall in the tropical and subtropical regions. NASA then make these data available on the TRMM website.

The usefulness of these data can be illustrated using the last seven days of rainfall in Asia. During this time the monsoon has strengthened - we are just moving into the real wet season in south and SE. Asia. Furthermore, the first really destructive tropical cyclone of the year has just tracked across the Philippines and then made a more-or-less direct hit on Hong Kong, before dissipating in southern China. As well as causing the sinking of a passenger ferry, this typhoon also deposited large volumes of rain that have been responsible for extensive landslides and floods. Using TRMM it is possible to obtain an estimate of the volume of rainfall that has fallen in Asia over the last week (Fig. 1). This does come with a caveat, which is that the spatial distribution of rainfall is very variable at the local scale, which this instrument does not resolve. Nonetheless it does provide a really interesting indication of the rainfall distribution.

Figure 1: TRMM image of the rainfall distribution for the seven days to 27th June 2008 at 06:00 UTC. Note the high rainfall totals for the Philippines and S. China, generated by the passage of Typhoon Fengshan (click on the image for a better view).

The locations of high rainfall appear to correlate with areas of flooding - for example, in this time floods were reported in the Central Philippines, South China around the Pearl River delta and in parts of N. India. Unfortunately, the correlation with fatal landslides at least is far poorer. Figure 2 shows the locations of fatal landslides that I have recorded in the same time period. A caveat is that perhaps not all the landslides in China have yet been reported, so the map in Figure 2 might change a little. However, it is clear that a simple rainfall total algorithm is not enough to provide a really good indication of the distribution of fatal landslides at least.

Figure 2: Locations of reported fatal landslides in the 7 day period shown in Figure 1.

The team at NASA are working hard on their landslide algorithm and are doing a terrific job. This is one of the most interesting landslide projects underway at the moment. I am hoping that they will present some interesting progress in this work at the International Symposium on Landslides and Engineered Slopes in China next week.

Tuesday, 24 June 2008

Some useful landslide and hazard links and blogs

I thought it would be interesting to provide some details of three blogs / sites that I have come across recently, and which I have found useful:

Save the hills: I have posted about this site before, but as we come into the monsoon rainfall season in north India it is worth highlighting again. Save the hills is a community blog focused on drawing attention to the appalling losses from landslides in Darjeeling. It is a terrific campaigning site that also contains some really interesting reports of landslides as and when they occur.

MCN Disaster Newsletter: This is a weekly newsletter that aims to raise awareness of disasters in Nepal, which is a country with more than its fair share of such problems. It contains a really interesting mix of hazard reports and details of capacity building activities in Nepal. This is an organisation that is doing a great job but which desperately needs support to maintain its activities.

GDACS: This is a really interesting idea. GDACS provides real time information about natural disasters occurring around the world. Users can create an account that means that the site will email or text bulletins as and when events occur. At the moment it is focused on tropical cyclones and earthquakes. The site is the product of support from the European Commission.

Sunday, 22 June 2008

Garbage dump landslides

News today of a garbage dump landslide in Guatemala City on Friday that is believed to have killed at least 50 people (Fig. 1) has set me thinking of the hazards of these types of mass movements . The slide in Guatemala appears to be fairly typical of these events - a large number of very poor people who were scavenging for recoverable items in a dump that was known to be dangerous were buried by a sudden slide.

Fig. 1: Rescuers recovering the body of a victim of the 20th June 2008 garbage dump landslide in Guatemala City. Image from AP.

A quick scan of the fatal landslide database for large (>10 deaths) events of this type showed that I have recorded at least seven large-scale garbage landslides since 2000 (Table 1), responsible for a recorded 571 deaths. Of course this is probably very much the tip of the iceberg as:
1. I have only looked at the large events; and
2. It is very likely that most garbage landslide events go unreported, given the locations and the people involved.

Table 1: Large-scale (>10 fatalities) garbage landslides recorded in the fatal landslide database. This is undoubtedly just the tip of the iceberg. Note the death toll for Guatemala City is provisional at this stage.

Garbage landslides are particularly horrible events. For most people being buried alive raises an inbuilt fear, but the thought of being buried by garbage is particularly unpleasant. In addition, garbage often releases toxic gas, which means that victims may die from poisoning whilst awaiting rescue. Garbage also generates heat, which means that the victims may be die from hyperthermia. In extreme cases the garbage may ignite.

So why do garbage landslides occur so frequently? Sadly, a substantial number of people in less developed countries make their living from garbage scavenging. Many of these people live amongst the rubbish as well. The numbers are staggering - for example, the Stung Meanchey municipal rubbish dump on the outskirts of Phnom Penh has 3,000 scavengers, 1,000 of whom are believed to be children. Life in such an environment is hard - garbage dumps are by definition unhealthy places that are also physically dangerous. Ironically of course these people are performing a valuable public service by recycling metals, plastics, etc, from which the rest of us benefit. The garbage is typically dumped by truck with little thought for the safety of the scavengers to form steep slopes (sometimes up to 70 degrees). Garbage sites are often on slopes anyway as they occupy land that is not ueful for development (Fig. 2).

Fig. 2: BBC News image of the Payatas garbage dump in the Philippines, illustrating the steep slopes often associated with these locations. Note the houses at the bottom of the slope - it is unsurprising that fatalities are frequent in such locations. It was in this dump that nearly 300 people were killed by a garbage slide in 2000.

The garbage pickers themselves then unknowingly destabilise the piles by removing material from the base, whilst heavy rain often triggers landslides as well, as happened in most of the cases above. Most of the slides are small, often killing just a few people. It is likely that such slides are very common, but the majority undoubtedly go unreported.

A quick search of Web of Knowledge showed not a single paper on garbage landslides. I guess this type of research is not very attractive, but if ever there was a landslide issue that needs attention then this is it! A study to try to determine the actual human cost of garbage slides would be a good first step, backed up by an analytical study of the mechanisms of garbage slides so that the people can be better protected. I am sure that the losses associated with garbage slides could be reduced without taking away these people's livelihoods.

Saturday, 21 June 2008

Earthquake triggered landslide at Aratozawa Dam in Kurihara, Miyagi prefecture, Japan

Many thanks to reader Ripendra for alerting me to the following, and indeed for doing nearly all the work for this post. The 14th June 2008 earthquake in Japan is known to have triggered a number of landslides, as per earlier posts, but the most notable appears to have occurred at Aratozawa Dam in Miyagi prefecture.

The Aratozawa Dam (Figure 1) is a rockfill dam used for flood control, irrigation and hydro-electric power generation. It is 74.4 m high and 413.7 m long, impounding up to 14,130,000 cubic metres of water.

Figure 1: Miyagi Prefectural Government image of the Aratozawa Dam before the earthquake and landslide.

In the earthquake, a landslide occurred in the catchment of this reservoir. Professor Sassa of the International Consortium on Landslides has provided a brief description of this failure here:

"A huge landslide akin to a massive hole being punched in the side of a mountain was triggered by the Iwate-Miyagi Inland Earthquake at an upstream section of the Aratozawa Dam in Kurihara, Miyagi Prefecture, causing widespread damage and making roads impassable. Kyoji Sassa, head of the International Consortium on Landslides, said the volume of collapsed earth near the dam is believed to be more than 10 million cubic meters. Sassa believes that this is comparable to the largest landslides that occurred as a result of the 2004 Niigata Prefecture Chuetsu Earthquake, which also struck in a mountainous region, and to the volume that fell in single landslides after last month's cataclysmic earthquake in China's Sichuan Province.

"It may have been a 'deep landslide,' in which a large amount of earth completely collapses from a point deep in the mountain slope," Sassa said of the landslide set off by Saturday's quake. In such deep landslides, tremors that seem to thrust up from below cause the pressure of underground water to rise, making the ground above seem to float before abruptly collapsing.

Hiroshi Fukuoka, an associate professor at Kyoto University's Disaster Prevention Research Institute's Research Center on Landslides, said there was a complicated relationship between the landslide and the dam downstream of it. "Water is stored in the dam, and it seems that the underground water level rose and caused a high-speed landslide. Earth on the outside of the slope fell into the artificial lake behind the dam, bringing the danger of flooding," Fukuoka said. "It is necessary to establish whether such a risk exists when building a dam."

In many ways this description does not really do justice to the landslide that actually occurred, which is rather impressive. Figure 2, from Getty Images, provides a rather more helpful illustration of what occurred here:

Figure 2: Getty images picture of the Aratozawa Dam landslide.

Tuesday, 17 June 2008

Recent landslide round-up

Just a quick post to highlight some of the more notable recent fatal landslides as we move into the Asian rainy season:

07th June 2008: Castle Peak, Hong Kong: A man and woman were killed in Hong Kong on Saturday when severe rainstorms caused a landslide that crushed the hillside hut they were living in (Reuters image below):

07th June 2008: Wangkha, Bhutan: Fourteen Indian labourers working on a hydropower project in Bhutan were killed when a boulder fell on their truck. (Very pleasing graphic from Kuensel Online below to illustrate the event:)

11th June 2008: Mogok, Myanmar: Torrential rains caused a landslide that killed 30 miners in Mogok, military-ruled Myanmar's "Valley of Rubies" gem zone.

14th June 2008: A Mw=6.8 earthquake triggered landslides that killed at least 12 people (Reuters image below of the Inn at Komanoyu hot springs hotel in Kurihara that was hit by a landslide. Seven people were killed here).

14th June 2008: Heavy rainfall in Arunachal Pradesh has triggered landslides in and around the town of Itanagar that have killed about 20 people.

Monday, 16 June 2008

Landslides from the 14th June 2008 "Iwate-Miyagi Inland Earthquake" in Japan

At 08:43:46 local time on Saturday 14th June 2008 a magnitude Mw=6.8 (USGS initial estimate) struck eastern Honshu in Japan. Although this earthquake (now being called the Iwate-Miyagi Inland Earthquake) was not particularly large, it occurred at a shallow depth (10 km) below an area that is quite mountainous. This is an areas that is very landslide prone anyway, being subject to active tectonic processes and frequent typhoon rainfall events. Once again, the consequences have been a large number of landslides.

Two days on some pictures are starting to emerge of the impact of the earthquake. The following three AP and Reuters images show quite well some of them:

The impact of all three on the road network us very clear - reports suggest that this has had a major effect on the rescue and recovery operations, as well as being responsible for a number of fatalities. The landslide in the first image is particularly interesting, but is slightly difficult to understand. Is this a large scale lateral spread?

Japan has a remarkable concentration of landslide specialists, so they are well-equipped to deal with this type of problem. Professor Sassa of the International Consortium on Landslides has commented on a large landslide at Aratozawa Dam in Kurihara, Miyagi Prefecture:
"Kyoji Sassa, head of the International Consortium on Landslides, said the volume of collapsed earth near the dam is believed to be more than 10 million cubic meters. Sassa believes that this is comparable to the largest landslides that occurred as a result of the 2004 Niigata Prefecture Chuetsu Earthquake, which also struck in a mountainous region, and to the volume that fell in single landslides after last month's cataclysmic earthquake in China's Sichuan Province. "It may have been a 'deep landslide,' in which a large amount of earth completely collapses from a point deep in the mountain slope," Sassa said of the landslide set off by Saturday's quake."

Meanwhile, the Daily Yomuri reports that:

"Eleven dams created by landslides that occurred when Saturday's Iwate-Miyagi Inland Earthquake hit are threatening to collapse, seriously affecting rescue efforts downstream. In response, the Construction and Transport Ministry on Sunday dispatched sediment control experts to the area to conduct a full-scale study on the possible danger posed by the so-called quake lakes building up behind such natural dams. According to the ministry, two quake dams have been confirmed on the Sanhasamagawa river and the Nihasamagawa river, five on the Hasamagawa river in Kurihara, Miyagi Prefecture, and four on the Iwaigawa river in Ichinoseki, Iwate Prefecture. Most of them stretch about 100 to 200 meters wide. Those rivers are feeders of the Kitakamigawa river. Once the water level rises and water spills out from a quake lake, its banks and the natural dam instantly collapse and a wave of mud could sweep into lowlands located downstream. "The danger of a collapse [of quake dams] changes all the time due to rain and aftershocks," a Construction and Transport Ministry official said. "We have to be very careful." The official said that the ministry plans to drain water from the quake lakes using heavy machinery to prevent a secondary disaster. There are some houses downstream of quake lakes on the Iwaigawa river. "Because we've got information that the water level is gradually rising, we may have to tell residents to evacuate depending on circumstances," an Ichinoseki municipal government official said.

They have rather a nice image of one of the sites. This is clearly not on the scale of Tangjiashan, but potentially hazardous nonetheless:

Landslides in Darjeeling, India

In September 2007 intense monsoon rainfall in the Kalimpong region of Darjeeliing in Northern India triggered large numbers of landslides, killing at least seven people. This has brought to the fore the issue of environmental degradation and the consequent landslides in Himalayan environments. Darjeeling is a microcosm of problems arising throughout the Himalaya, with rapidly increasing numbers of landslides causing major problems.

Darjeeling is probably most famous for its tea plantations, which thrive there because of the rolling topography, moderate temperatures and high rainfall. These conditions are great for tea but bad for slope stability. Preservation of the landscape can minimise the occurrence of landslides, and indeed the tea plantations generally do this reasonably well if managed carefully. However, many Himalayan regions are currently facing a near perfect storm of conditions that promote landslides - rapid population growth (which means that there are more people needing more houses and more food), deforestation, uncontrolled building to low standards, intense monsoon rainfall, and the construction of low cost roads with poor engineering standards. The upshot is all too predictable - increasing numbers of landslides with a growing level of impacts. The image below from Google Earth shows the town of Kalimpong, located on a high ridge. The slopes below are scarred with active landslides, but more importantly much of the landscape is made up of old landslides and landslide scars. Kalimpong has been seriously affected by active landsliding in recent years.

Now, as the SW monsoon rains start to build, the issue of landslides in Darjeeling is becoming pressing once again. In Darjeeling though there is one glimmer of hope, which is a remarkable campaigning group called Save the Hills, led Wing Commander Praful Rao (Retired), who campaign to raise awareness of issues associated with slope instability in Darjeeling. They have a very detailed blog in which they document both the magnitude of the challenges and the range of their activities to reduce landslide occurrence and to increase preparedness and resilience. I thoroughly recommend taking a look as it represents a model of community-level action in response to landslide threats. The main blog is here:

I particularly enjoyed their images of historic landslide events, which can be found here. Darjeeling is fortunate to have such an active group, and I hope that it is able to make a difference. Many other places would benefit hugely from the activities of such organisations.

Thursday, 12 June 2008

Debris flow video from Hong Kong

Thanks to a former student for bringing the following video to my attention. Hopefully it will play directly when you click on the box below - if not then take a look at:

This channelised debris flow occurred on Lantau Island, near to Hong Kong airport, on 7th June when the area was affected by an exceptional ("black") rainfall event. In a single hour Hong Kong observatory recorded 140 mm of rain, the highest intensity on record there. During the day, which saw over 300 mm of rain across most of the area, about 400 landslides were triggered on Lantau alone, including one that killed two people.

Helpfully, the Hong Kong Observatory make rainfall isohyet charts available online. I have annotated below the map for 7th June and the approximate location of the landslide. Note that the landslide occurred in an area in which rainfall totals were at their highest (>400 mm).

Hong Kong Observatory isohyet map of the rainfall distribution for 7th June 2008. The approximate location of the landslide in the Youtube video is indicated.

A couple of words about the landslide. This is technically a channelised debris flow. It appears to have started as a comparatively small shallow slip in weathered materials high on the slope. The mobile material has then picked up (technically this is termed "entrained") debris and water in the channel. Once such flows start, and assuming that they have a steepish slope to travel down, they build momentum and volume to create a highly turbulent and destructive pulse of material that moves very rapidly. Slides of this type typically occur in pulses, as can be seen in the video, which is of course particularly nasty for people in the way. Needless to say these types of flow are very dangerous, particularly if they overflow the channel banks. Fortunately, in this case the flow appears to have remained mostly in the channel itself.

Hong Kong has a long history of these types of landslides, and is adept at managing them. To enhance this the Hong Kong government has just started a large project to identify natural terrain areas prone to these failures and to mitigate areas of high risk.

Wednesday, 11 June 2008

Deaths and research in landslides

Now that the Tangjiashan crisis is over, it is clearly time to move on to other things. I will post
a retrospective on that event when the dust has settled. For now I think we should allow the Chinese to bask in the success of their achievement.

To change the topic, I thought it would be interesting to post an analysis that I did a couple of years ago. The analysis was simple but quite fun - I looked at the landslide fatality database for 2003 and 2004 by calculating the proportion of fatalities by large geographical area. I then looked at the field areas of the research presented in the Springer journal Landslides and in papers presented at a selection of international landslide conferences, and worked out the proportions as per the fatalities.

Now ideally there should be some relationship between the locations of landslide deaths and the locations of landslide research, one would hope. Care is needed as:
1. One year of landslide data might not be representative; and
2. The sample of the areas of research might not be representative.

However, the plot does show how far we are from an equitable situation:

In my simplistic world I would hope that most areas plot close to the central line on the graph. In most cases this is clearly not so. The regions with really serious landslide problems in this period - Central America, SE. Asia and S. Asia has a tiny proportion of the research. Regions that dominated the research into landslides - North America and Europe - had very few landslide fatalities. Only East Asia looks to be in the right place amongst those regions with large numbers of both research projects and fatalities.

Two things emerge:
1. Care is needed because in 2004 landslides in Haiti caused huge problems, which mean that for this period Central America dominates the statistics. If we were to take the period 2002-2006 East and South Asia would dominate the fatality count.
2. Only E. Asia appears to have the balance right between research and fatalities. This reflects the large landslide programmes in China, Japan, Hong Kong and Taiwan.

Of course the picture might look different if this were to be examined through the lens of economic losses rather than fatalities, but not I suspect if we were to normalise by GDP or per capita income. All of which serves to highlight the fact that the landslide research community should shift its focus if it really wants to make a difference by saving lives.

I will try to do a better analysis of the full dataset over the summer, but thought it would be interesting to share this initial analysis.

Tuesday, 10 June 2008

Tangjiashan - problem over

Xinhua is now reporting that the flood wave has passed through the city of Mianyang safely. This appears to mark the end of the crisis. I have to say many, many congratulations to the Chinese authorities and everyone involved for their handling of this event. To manage to mitigate this lake in this way is probably an unprecedented achievement.

Tangjiashan - let's be a little careful here...

I note from comments below that now that the landslide dam at Tangjiashan has failed the disaster is considered to be over. Please let's be a little careful here. It is true that the flood waves has now passed through the upper parts of the river. Below is the graph of the reported discharges as per the comments left on earlier postings:

The water level is clearly dropping. However, the key issue is what happens downstream - it will take some time for the flood wave to reach these downstream areas. A wave of this size will inevitably cause considerable damage along the channel. We will know that the disaster has been averted when the flood wave disperses on the plains. Until we know this has occurred we must be careful. However, the Chinese government has been effective in moving people to date, so all the indications are that they are managing this crisis well.

Note that there is also a secondary risk that this wave will trigger further landslides.

Tangjiashan - latest image

This is a new image released by Xinhua of the collapse at Tangjiashan. It is clear that the dam has now effectively failed:

The rate of scour is clearly very fast and the lake level is still very high indeed (it has apparently dropped 10 m). CCTV has just reported that the smaller dams downstream have now been overtopped and washed away.

Updated: Tangjiashan is failing very rapidly

Latest graph on Tangjiashan flow rates:

It is clear that the failure of the dam is now very rapid and is uncontrollable. It is essential that anyone in a vulnerable location downstream is evacuated immediately. It appears that the workers on the dam have been successfully pulled off the site by helicopter, which is a relief.

Update: Correspondent Andy is now reporting that cracks have been observed on the dam, which suggests that it is failing internally.

Tangjiashan flow rate graph

This is a quick, back of the envelope graph of the reported flow rate at Tangjiashan.

The dramatic rise in flow rate is clear, but so far this is not catastrophic. The next data point is the key one - can anyone provide it?

Updated: Tangjiashan is breaching

Andy has posted a comment on the blog saying "Latest news: At 7:20am 6/10 (Beijing time), the outflow reached 377 m3/s and lake level was 742.80 m compared with 742.96m 12 hours ago. Before hit the post button, I checked the news one more time. The new outflow number is 500 m3/s! Is it a good or bad news that the outflow increases so fast?"

Update: Xinhua has confirmed that the rate of flow has indeed increased.

Update: Ripen reports that at 09:00 local time the flow had increased to 1210 cumecs. The breach is clearly occurring very fast now. The next three hours are critical.

Update: Aerial image of the breach, from

Note that the very steepest part of the channel is just below the crest. The magnitude of the flow will be determined by whether this migrates to the crest. The situation is on a knife-edge.

I hope all of the soldiers have been pulled off the dam now!

Monday, 9 June 2008

Erosion on the lower parts of the Tangjiashan dam

Two days ago I posted this picture AP image of the channel at Tangjiashan:

AP image captioned: Water from the Tangjiashan "quake lake" flows down a landslide site in the quake-devastated Mianyang City, southwest China's Sichuan Province.

Today, the following Reuters image has been released, apparently showing the same location:

Reuters image entitled: Water flows through a sluice channel of the Tangjiashan quake lake in Tangjiashan, Sichuan Province June 8, 2008 in this picture distributed by China's official Xinhua News Agency. The water level in the quake lake stood at 741.82 metres above sea level at midday on Sunday, still 1.45 metres higher than the sluice, with the lake's volume exceeding 240 million cubic metres, Xinhua News Agency reported. Picture taken June 8, 2008.

The caption says that the picture was taken yesterday. Clearly there was massive erosion on the downslope face. The flow yesterday was reported as c. 10 cumecs, whereas today it is estimated to be 50 cumecs. If these dates are correct then the erosion rate must now be very rapid.

A further image shows the site from above:

Reuters image entitled: Water flows through a sluice channel (R) of the Tangjiashan quake lake as another channel is being constructed in Tangjiashan, Sichuan Province June 8, 2008 in this picture distributed by China's official Xinhua News Agency. The water level in the quake lake stood at 741.82 metres above sea level at midday on Sunday, still 1.45 metres higher than the sluice, with the lake's volume exceeding 240 million cubic metres, Xinhua News Agency reported.

Note two things here. First, the magnitude of the erosion. Second, as suspected yesterday, the team on the dam appears to have filled in the channel that they cut before (see the bulldozers on the left - compare this with the first picture (above) in this post).

The critical issue is how quickly this erosion is moving upslope. I suspect rather rapidly now as the flow in the channel increases and the water level in the lake rises.

Sunday, 8 June 2008

Updated@ Tangjiashan images deepen the mystery

NB - I have now updated this with an image at the end that shows this more clearly.

A quick image search today showed up the following Reuters image, entitled: "Water from the Tangjiashan "quake lake" flows down in quake-hit Mianyang, Sichuan Province June 7, 2008. Chinese troops on Saturday eased pressure on a swelling "quake lake" threatening hundreds of thousands of people, but a smaller lake burst its banks in a show of destructive force."

According to the caption this was taken on 7th June - i.e. yesterday.

Now, lets compare this to the earlier image that we had by putting them side by side, with the older one on the left and the newer one on the right:

A few things to note here (see annotated image below):
1. There does appear to be more flow in the second image, though not vastly so;
2. The activities of the excavators in the second image appear to be to deposit soil and rock, not to be excavating it (highlighted with the red circle below). Are they filling something in here?
3. The small waterfall adjacent to the blue trucks in the first image has eroded back a considerable distance in the second one (highlighted with a green circle below). This shows that erosion is occurring, but through a nick point.

The timing of the image is crucial here. If it was indeed taken yesterday, and the nick point eroded back that far during the day, then the erosion back of this nick point becomes a very possible failure mechanism for the dam.

Updated - here is another Reuters image that shows this more clearly:

An update on Tangjiashan

I thought that this would be a good time to review the situation with Tangjiashan now that water has been flowing for over a day.

1. Water flow is now apparently about 21.5 cumecs - up from 10 cumecs at about the same time yesterday. This is still not enough to balance inflow, so the lake level continues to rise. Of course as the lake level increases the rate of flow should also increase, so inevitably this will go up.
2. It appears that for at least part of its course the flow is not following the man-made channel but has created its own route.
3. The rate of down-cutting is slow, perhaps because of the presence of boulders in the channel bed. These boulders, if they are persistent through the landslide mass, provide a source of hope.
4. There is some evidence that some water is preferentially flowing into the dam mass rather than through the whole length of the channel. This will be wetting up the landslide mass.

Some thoughts:
a. So far so good in terms of no disaster.
b. The Xinhua press releases have fairly consistently talked about draining the lake - the assumption being I think that the channel would slowly down-cut to allow this (?). This does not appear to be happening. I think that this may be bad news in the medium term as this will mean that somehow a channel will need to be constructed to cope with the peak flows in the rainy season. This is a very tough challenge.
c. The wetting up of the landslide mass may also be a threat. A big concern would be the possibility of a slope failure in the dam mass as this wetting occurs. This is pretty much the worse case scenario
d. Xinhua is reporting that "A relatively strong aftershock rocked the dam of the Tangjiashan "quake lake" for about 20 seconds at 6:51 p.m. on Sunday and caused massive landslides on surrounding mountains. The magnitude of the aftershock was not immediately available and its impact on the dam was under surveillance." The USGS suggests that this was a magnitude 5.0 earthquake at a depth of 10 km (i.e. a large but not exceptional event). The "massive landslides" comment is hard to interpret, but deeply worrying. A large slide into the lake could cause an instant over-top situation. The dam is unlikely to survive this intact.

I have noticed that the international media reports appear to imply that now that the lake has flow through the sluice the problem is over. See for example the Washington Post, which reported that:

"'Quake Lake' Drainage Eases Danger in China "

This is wrong as the danger has not yet eased. If the lake level was falling then maybe, but it is still rising. As the Chinese media have rightly suggested, the situation remains critical. If the lake level does not start to at least stabilise, and preferably fall, soon then the danger level will rise substantially. This remains a deeply worrying situation.

As ever comments are welcome from all. The discussion on previous posts has been terrific - I hope that we can develop this further.

Saturday, 7 June 2008

What is going on at Tangjiashan?

First thanks to both Ripen and Hideo Ototake for their emails and images on this issue.

I remain somewhat confused about what is going on at Tangjiashan. We know that the sluice now has flow within it - that is not in doubt. But what is happening with that flow is very hard to understand.

First, let's work out the flow rate needed to maintain a constant lake level. The earthquake happened on 12th May and now it is 7th June - this is 26 days. In this time the lake has, according to Xinhua, accumulated 229.5 million cubic metres of water. By my calculation this is approximately 102 cubic metres per second.

So, to maintain the level of the lake at a constant point the channel plus seepage should be transporting this volume of water. More than this and the lake will slowly empty, less and it will fill (this is certainly not rocket science!).

Now let's look at the images that are available at the moment. In this post I highlighted one that appears to show a low level of flow. The image below has also appeared here. This clearly shows a level of flow that is well below 100 cubic metres per second (from now on let's call this a cumec for convenience). Furthermore, Xinhua has now said that:
"Experts calculated that the water flows at seven to eight cubic meters per second, far more than the previous two cubic meters per second."

Now the figure of less than 10 cumecs is absolutely consistent with the image above, which is clearly of this order. There is no way that this is the 100 cumecs required to balance average inflow.

So where is the excess water? I cannot work this out! Possibilities are (as I see them - there may be more):
1. In-flow is only <10 cumecs now because it has not rained for a day or so. I do not believe that this is conceivable;
2. The lake is filling so slowly that only <10 cumecs is escaping. This is possible - but suggests that there is a great deal more flow to come (which means that the current situation is a "phoney war");
3. The excess is being dealt with by seepage through the dam (in which case then the flow is now essentially out of control - this is not necessarily bad news, but it could be);
4. The volume of the lake has been seriously over-estimated (this does not seem likely).

There may be other options. Suggestions please?

Whatever, something does not add up here, and that is very worrying to my mind...

Finally, Ripen asks what the channel is on the right side of the above image? Is this a former attempt to cut a channel by the army? Is it erosion by seepage (there appears to be water within it)? Is it a natural feature associated with the landslide (I don't believe this). Or is it something else (again, suggestions welcome please)?

In conclusion, something feels wrong here. I hope it is just that I am not getting a good picture of what is going on, which means that I am missing something vital. This could well be the case, or...

Updated: Tangjiashan - flow rate information

Xinhua is now reporting that the discharge in the sluice has now reached 40 cubic metres per second. This means that it is less than half that required to balance inflow at the current rate (c. 100 cubic metres per second). So far at least the channel is apparently stable. The same report indicates that the lake volume is now 229.5 million cubic metres.

This is good news. The lake should continue to fill slowly until the water depth is greater enough to drive enough water through the channel to balance the inflow. The potential for erosion will increase, but so far so good. No heavy rainfall is expected for at least a week, which provides time enlarge the channel further to deal with the summer floods.

Updated: an anonymous poster (thanks!) has drawn my attention to this AP image, which is captioned:
Water from the Tangjiashan "quake lake" flows down a landside site in the quake-devastated Mianyang City, southwest China's Sichuan Province.

I suspect that this was taken soon after flow started as this is clearly not 40 cubic metres per second. I guess that the channel on the right us the one that is referred to in the Xinhua posts as having ongoing works to widen it. Note that the main flow, before it starts to split, does appear to be eroding the channel bed at least to a certain degree.

Updated: Flow begins through the channel at Tangjiashan

Various media sources are reporting that water began to flow through the channel at Tangjiashan today. For example, Reuters is reporting that:
"Chinese troops began easing pressure on a dangerously swollen "quake lake" on Saturday, with water gushing into a man-made sluice in an operation monitored by satellite."

The dam and channel now has to survive two critical phases. The first lies in the next day or two, with increasing volumes of flow through the sluice. The key issue will be the rate of erosion of the channel bed. If the rate of erosion is slow, then we can breathe a sigh of relief. Ideally the channel bed will slowly erode, releasing the water at a manageable rate. It could be that no bed erosion will occur, in which case the worry would be the second critical phase, which will be the first really intense rainfall event in the rainy season. This of course would trigger a flood down the river that would seriously challenge the channel. Personally, I suspect that this is comparably unlikely as erosion of the channel bed looks probable.

Thanks to Edward Vielmetti for his many updates over the last few hours. He has just pointed out that Xinhua has recently stated that:

"The long-awaited drainage of China's Tangjiashan "quake lake" started at 7:08 a.m. Saturday, when its water flowed into a manmade sluice channel. A Xinhua reporter at the commanding center saw water passing the sluice channel via satellite monitor. The flow was rapid, steady and gradually increasing in volume. The channel bifurcated, and some of the water diverged into the separate small channel at 9:30 a.m. Soldiers of the armed police force are still working at the exit of the sluice channel to expand the channel and make it steepy so that the flow could be speeded up."

The statement that they wanted to make the channel "steepy" (presumably steeper?) is concerning if true - the aspiration should be to slow the flow down to reduce the erosion rate, not to speed it up.

Updated: Xinhua is now reporting that:
The dam of Tangjiashan quake lake faces no danger of collapsing in the foreseeable period, a top military commander at the site said Saturday. "The dam has no collapsing danger in the foreseeable period," said Fan Xiaoguang, deputy commander of the Chengdu Military Area Command.

I interpret this as being an indication that, so far at least, the rate of erosion is low. If true then this is would be very good news indeed.

Friday, 6 June 2008

Tangjiashan - delayed again

Xinhua is now reporting that the flow through the sluice has now been delayed through the use of a 60 cm high temporary dam (see below). This is an interesting development - I wonder why they have decided to do this at this late stage. Delaying the flow has a risk associated with it as the body of water will be larger, and thus the flood is potentially more damaging. Presumably they feel that the risks posed by this delay are more than compensated for by the larger channel, which reduces the chances of the dam eroding away quickly. I have noted before that in my view the channel is too small, but I am surprised that they have done this at this late stage.

The decision not to use dynamite, noted at the bottom of the text below, is undeniably sensible!

Original Xinhua text:

Water level of China's main quake lake Tangjiashan topped the lowest point of 740meters by 0.20 meters on a sluice channel at 4:00 a.m. Saturday, but it was prevented from overflowing by a temporary dam added to protect workers on emergency construction downstream. More than 100 armed police were airlifted to Tangjiashan to broaden and deepen the sluice channel on Friday afternoon, said an expert at local commanding center on early Saturday morning. The temporary dam was 0.6 meters high, said the expert. Local headquarters said rainfall was heavy at the lake, with an average of two millimeters per hour. The swollen lake was formed by a massive landslide that blocked the flow of the Jianjiang River in Beichuan County, Mianyang City, following the May 12 earthquake in the country's southwest. It held more than 220 million cubic meters of water and posed a threat to about 1.3 million people downstream. Some 600 armed police and soldiers worked for six days and nights to dig a 475-meter channel to divert water from the lake. More than 250,000 people in low-lying areas in Mianyang have been relocated under a plan based on the assumption that a third of the lake volume breached its banks. Two other plans require the relocation of 1.2 million people if half the lake volume is released or 1.3 million if the barrier fully opened. The swollen quake lake has put China's longest oil pipeline at risk. The pipeline, winding from Lanzhou via Chengdu to Chongqing, was 60 kilometers downstream from the lake. With a capacity of transferring six million tons of oil each year, the pipeline provides 70 percent of product oil to Sichuan and neighboring Chongqing Municipality. If the line was cut, refined oil in storage could only supply Sichuan for three days, whereas repair work would take 30 days. Emergency plans have been drafted, according to colonel Liu Yongzhan. "Gate valve of the pipeline will be shut to prevent potential pollution, and a float bridge shall be later set up, so that a temporary line would ensure resumption of oil supply within48 days," he said. Engineers have feared that the lake could burst and cause a flood, citing unforeseen factors such as fresh landslides, heavier than expected rain and the instability of the mud and rock dam. The plan of using dynamite to blast the lake had once been under discussion but was later announced as given up, according to Liu Ning, chief engineer of the Ministry of Water Resources. "We will not use dynamite if water in the lake could spill over naturally," he said. "Blast might incur unexpected result."

Tangjiashan - here we go...

Xinhua has reported in the last 20 minutes that:
"Water level of China's main quake lake Tangjiashan climbed to the critical point of 740 meters on a sluice channel at 0:00 a.m. Saturday, but the long-awaited drainage hasn't started yet."

Given that we know that it has been raining in the area this afternoon, over-topping must be about to start. Hold onto your hats...!

And can I just say good luck to everyone in the path. I hope to wake up tomorrow to hear that this hasn't even made the news. If so, the Chinese authorities will have achieved the almost impossible.

Tangjiashan - overtopping is imminent

Xinhua is reporting that over-topping of the Tangjiashan dam is now absolutely imminent. They Chinese media have reported that the water level in the dam rose to 739.52 metres above sea level at 4 a.m. GMT today. This is 48 centimetres from the bottom of the channel. It appears that further evacuations are now underway.

More as and when information becomes available.

Wednesday, 4 June 2008

Updated: Tangjiashan (Beichuan) dam - a summary of what we know

In the next 24-48 hours the water level at Tangjiashan should reach the spillway and flow should start. I thought therefore that it would be useful to summarise what we know about the site:
  • The landslide dam is 124 metres high;
  • At the time that the over topping begins the lake will have a volume of about 205 million cubic metres of water (UPDATED - the volume is now estimates as 211.6 million cubic metres as at 2 pm (loca) Wednesday. The lake level had 1.37 m to go before reaching the spillway)
  • The channel that has been cut appears to be about 5 (now known to be 7 m) wide at the bed (though note that the film with the journalist in the bed suggested less than this) and about 15 m at the top. It is 12 m deep The bed is unlined and unprotected. It is formed in silty materials with a few clasts (stones);
  • Failure can occur through any of three mechanisms:
    • The water can reach the spillway and start to flow over the dam;
    • The two known seepage points (total reported flow = 10 cubic metres per second) in the dam could cause collapse;
    • A landslide into the lake could trigger a wave that would overtop the dam. A dangerous slope has been identified at Xuanping township.
  • Downstream there are three further (smaller) landslide dams in the Beichuan area;
  • The Chinese government has relocated about 250,000 people. In total, about 1.3 million would be threatened by total collapse.
Reuters image of the top of the dam showing the lake and the spillway under construction

These are the best and worst case scenarios for the dam as I see them:
Best case: water flows over the spillway and erodes it slowly. The lake level reduces gradually over a few weeks;
Worst case: a large landslide into the lake triggers a wave that overtops the dam, causing complete collapse. The entire volume of the lake is released very rapidly. The resultant flood pools behind and then breaks through the downstream dams, creating a devastating debris flow.

The most likely scenario at this stage is that water will flow over the spillway. A ball park flow rate appears to be about 100 cubic metres per second when there is no rainfall. This channel will not be able to cope with this as far as I can see, leading to rapid erosion and therefore collapse of the dam. I suspect that most of the lake volume will be released in a few days at most. The water is likely to pool briefly behind the landslide dams downstream, which will probably increase the size of the flood wave once released.

Where will the water go? I have tried to produce a quick Google Earth summary image below of the anticipated flow path (click on the image for a better view):-

The flood is likely to be large, fast and very damaging in the mountains, but should spread at least a little and slow on the plains. We can only hope for the best.

I am travelling to Nepal this afternoon, but will do what I can to keep this updated from there.

Tuesday, 3 June 2008

Tangjiashan - CCTV footage of the channel

There is a very good film (in Chinese) of the Tangjiashan landslide. It is available at:

The first 6 minutes or so are worth a look. Early in the video they fly a helicopter over Beichuan, which shows the damage caused by landslides there. They then land the helicopter on the landslide mass and the reporter goes into the channel. It appears to be about 5 m wide at the base with steep walls. The material is silt by the look of it, with just a few clasts.

The Chinese goivernment has done a fine job to build a channel at all, but this is not going to be able to contain the flow.

Tangjiashan - is the dam leaking?

Quite alarming reports started to emerge yesterday that the dam at Tangjiashan might now be "leaking". For example, a correspondent in Japan, Hideo Ototake, sent me a link to this website, which suggests that a leak started at the bottom of the dam in the afternoon of 1st June. Reuters today is carrying this report, which states that:

"The largest "quake lake" formed by China's most devastating earthquake in decades is not expected to start draining until Thursday due to a lack of rain, state media said. The water level at Tangjiashan was 735.78 meters on Monday, only 1.24 meters higher than a day earlier, because of a lack of rain and multiple leaks on the natural dam made of mud and rock, the newspaper said."

Multiple leaks?

In addition, the Hideo Ototake also pointed me to a new and rather splendid aerial image of the lake that is available here (click on the image for a better view).

This image shows the lake and the dam, with the dry river valley below. As far as I can tell this image predates the construction of the channel. What is slightly alarming is the area at the toe of the dam that I have enlarged here:

This image might suggest that at the toe of the dam there is what appears to be a damp area? This is very unclear, but this is one interpretation. Is this the start of seepage?

There are strong reasons for concern here. If the dam is showing signs of seepage then there is a risk that the dam will fail through this mechanism. This is a worse case scenario as collapse under this mechanism can be very rapid indeed. Failure is not necessarily catastrophic, but there is a reasonable chance. So if the flow rate starts to increase through these seeps then the authorities will need to move all of the population in the affected areas without delay. Note also that if seepage has started then this suggests that the core of the dam is more permeable than I had anticipated. This is not good news at all when the overspill channel starts to erode.

Assuming that seepage failure doesn't occur the spillway is expected to start handling flow on Thursday at the latest. It will be an interesting couple of days.

Monday, 2 June 2008

May 2008 Landslide map

The map below shows the distribution of fatal landslides for May 2008. At the moment I am not including the landslides associated with the Sichuan earthquake as I cannot yet estimate the impacts of them. Clearly the numbers of landslides and the numbers of fatalities will dwarf the numbers below. I will try to ascertain the magnitudes in due course, though this is very challenging. I have included reported rainfall-triggered landslides in the aftermath of the earthquake, though again this is likely to be an underestimate.

The statistics are:
Number of fatal landslides (excluding those triggered by the Sichuan earthquake): 21
Number of fatalities: 280

May was once again very substantially below the average number of fatalities for may in 2003-2007, which is 336 deaths . In my post on the April figures I said that May "could be an interesting month!" I was right, but for the wrong reasons. The distribution is again fairly well-scattered, but as we come into the wet season it is increasingly focussed on Asia. The red triangle represents Sichuan.

May 2008 fatal landslide locations (click for a larger version)

The patterns that were starting to emerge in the annual map last month are now strengthening. In particular the inevitable clusters in the Himalayas, western Central America, the Philippines and Indonesia are now clear. The cluster in the Himalayas is still focused on the west side at the moment as this area gets heavy rain in the early part of the year. The monsoon rains have now started in the very southern part of the Indian subcontinent and will now start moving northwards. This will drive the growth of this cluster. There are certainly more landslides in Central America than is normal.

2008 fatal landslide locations to the end of May (click for a larger version)

So what of the month ahead? We, expect to see the number of fatal landslides increase markedly, especially in Asia. As the rains start in Central China landslides seem inevitable. As the La Nina conditions fade, we might expect to see fewer landslides in South America, but more in Asia.

Sunday, 1 June 2008

Tangjiashan Lake, Beichuan - so now we wait...

Xinhua is reporting that the spillway at Tangjiashan Lake is now finished and that the soldiers constructing it have withdrawn, leaving just a few hardy and brave souls to raise the alarm.

First, we must recognise the awesome task that has been accomplished here - 135,500 cubic metres of debris have been shifted to create a channel that 475 m long and "up to" 10 m wide. To do this in the circumstances is utterly extraordinary, putting the responses of most other countries to disasters to shame. But where does it leave the situation at Beichuan and downstream?

First, I have tried to collate some images of the channel. All originate from Xinhua. In the first, excavators are shown digging the channel. Now, this is unfinished so great care is needed, but note:
  1. The material is very fragmented and soft. There are almost no boulders in the channel bed and in the walls of the channel;
  2. The channel is very narrow indeed. It was unfinished when the image was collected, so might not be the final product, but the walls have been well-graded, so it probably is not far from it (?);
  3. The gradient is quite steep, meaning that water flow is likely to be rapid.

In the second image, soldiers are seen working on the channel banks. Rocks are being moved in chains - it appears that this might be to armour the channel bed? Again, the channel is very narrow and steep, and interestingly it also has a very clear bend in it.

The final image shows one end of the channel, which here is broad and quite shallow. Again the material is fine grained, with almost no obvious boulders. The channel gradient is quite low.

There is little more information available to me at present. The channel is expected to start to carry water in the next three to four days. It seems to me (and to one or two other landslide people that I have spoken to ) that the channel will almost certainly erode rapidly once it is carrying water. The steepness of the gradient and the fine grained material appear to almost guarantee this. It seems likely therefore that the authorities have decided that the release of much or all of the water in the lake is inevitable. The key question will be how quickly that erosion occurs, and thus how big the flood wave actually is. The fine-grained nature of the material suggests to me that it could well be rapid, although this could be tempered by the size of the blockage. Note that the blockages downstream will also play a role, allowing the water to pond and then break through, which is likely to increase the maximum size of the flood wave. It seems likely that the flood wave will be very large, but frustratingly it is difficult to say just how big this will be. Evacuating as many people as possible downstream seems to be to be prudent.

Xinhua remains somewhat intriguing in its reporting, saying: "Two other plans require the relocation of 1.2 million people if the half of the lake volume was released, or 1.3 million if the whole landslide blockage is washed away."

I cannot quite understand what this means. It seems to me that the river will in due course cut a channel right through the dam - i.e. all of the water will be released. The key question is how quickly this occurs. Past experience suggests that this could be anything from a small number of hours to months or even years. If I was a betting man I would tend towards sooner rather than later. I hope that the authorities understand that if a catastrophic collapse were to start it would happen in a very short period of time. Thus, the evacuation would need to be very rapid indeed. Hopefully (and probably I guess?) this won't be the case, but...

We must also be very careful to ensure that we understand that this is just one problem of many. We mustn't lose sight of the fact that:
1. There are c.27 other remaining dangerous landslide dams. All need to be dealt with. It would be an extraordinary achievement to mitigate them all, so an unexpected collapse of at least one is surely likely. It worries me that there is so little information about other areas. Given the numbers of landslides that we know have occurred, I wonder if we are even sure that all of the valley blockages have been identified. I guess we have to trust the Chinese on this, and they have some incredibly competent people, but concerns linger.
2. Landslides continue to occur in the aftershocks. These represent a substantial hazard.
3. We are getting closer to the wet season. In heavy rainfall landslides will be a substantial problem, as will the movement of sediment in floods and debris flows. Dealing with the valley blockages is clearly the immediate crisis, but there are many more just around the corner.

Finally, I will be in and out of my office for the next few days but can be contacted by email. I would really like to hear from anyone who can provide more info, and am willing to discuss the above in more detail as and when appropriate. I intend to continue to post on here as much as I can.