The collapse of landslide dams - the challenge in Sichuan

In view of the ongoing deep concern about the stability of the >30 landslide dams in Sichuan (now known in the press as "quake lakes"), and in particular the one upstream at Beichuan, I thought it might be helpful to provide a brief review of what we know about landslide dams.

First, there us a great deal of literature on this topic. For example, Schuster and Costa (1991) detailed 463 examples of landslide dams worldwide. Since then many more have
been described. Most landslide dams ultimately collapse, most commonly through overtopping. Here the water starts to flow over the surface of the dam. In most cases this causes the dam to start to erode, which releases a little more water. Usually this creates a feedback loop that leads to complete release of the water in a short period. Occasionally the dam fails because of seepage or piping through the blockage itself or through porous sediments below the dam or because of a slope failure in the dam wall. These slope failures are sometimes triggered by earthquakes. Finally, failure sometimes occurs because of another landslide entering the lake. failure of the second landslide is often caused by the elevated ground water levels created by the lake, or by an earthquake aftershock. Such failures are generally exceptionally catastrophic.

Failure is not inevitable - and indeed some landslide dams last for hundreds or even thousands of years. These are the exceptions though, usually arising because of specific conditions. Schuster (1995) looked at 187 examples worldwide, finding that 55% of failed within one week of their
formation, and 89% had failed with a year one year.

So why do some dams survive? Usually survival occurs because of some or all of the following factors:
1. The dam consists of large boulders that are not eroded by flow. This is the case so far at least in the Young River landslide dam in New Zealand (Figure 1), which formed in September 2007 and so far has not collapsed.
2. The dam or surrounding slope allow release of water through piping or seepage without eroding.
3. The surface area of the lake is sufficiently large that evaporation balances inflow.
4. It is mitigated by humans (Figure 2).

Figure 1: Stuff.co.nz image of the Young River landslide lake in Mt Aspiring National Park, New Zealand.

Figure 2: Spillway being constructed on the Hattian Bala landslide dam in Pakistan in 2006.

Unfortunately, the conditions required for 1-3 are comparably rare. They do not seem likely in Sichuan given the materials and the rainfall.

Dam collapse events are very dangerous indeed. Collapse usually occurs rapidly, causing a disastrous flood. The magnitude of the flood wave can be very large, which is exacerbated by the fact that the water picks up the debris from the dam, changing from a flood to a debris flow consisting of water, soil, rocks, trees, etc. The statistics are quite frightening - for example the artificial breaching of a landslide dam on the Bairaman River, Papua New Guinea induced debris flow that was 80 m deep immediately downstream of the dam. 39 km downstream the flood wave was still 8 m high. The collapse of the‘La Josefina’ rockslide dam on the Paute River in Ecuador, 26 dams after it was formed, induced massive flooding along a 100 km stretch of the river. The 1786 collapse of a landslide dam on the Dadu River in Sichuan is believed to have killed over 100,000 people. Finally, the failure of landslide dam near Mount Cayley in Canada created a debris flow that travelled at about 35 metres per second. The flow was sufficiently violent to propel wood splinters that became impaled in trees that they hit.

Mitigating landslide dams
Mitigating the threats associated with landslide dams is by no means simple. The following approaches are usually adopted:
1. Monitoring: the most immediate response is usually to put in place monitoring, often in the early days consisting of an observer with a suitable communication system. This is often accompanied by closure and evacuation of potentially-inundated areas. This monitoring is often then automated to create a warning system, but this requires that people in affected areas are aware of the hazards, receive the warning when it occurs, and know how to react. Achieving this is far from trivial.
2. Syphons and pumps: In some smaller lakes short term management of the problem can be achieved using syphons and / or pumps to draw down the water level. This is not a suitable approach for long term management due to cost and reliability, but it can be effective in the short term.
3. Tunnels and spillways: The most common mitigation measure is to create a spillway - i.e. a channel across the top of the dam (at the lowest pint obviously) that will allow prevent overtopping. This is usually done by digging a channel from the downstream side to the lake. The channel must be armoured to prevent erosion. In some cases a tunnel is created for the same purpose. This is clearly more expensive. Great care is needed to prevent catastrophic release of the water.
4. Stabilising the dam and adjacent slopes: sometimes the dam is stabilised to ensure that it does not collapse due to slope failure. This is usually undertaken by reinforcing and reprofiling the dam surface, sometimes with an added layer to reduce seepage.
5. Removal: Occasionally it is necessary to remove all or part of a landslide dam, for example to reopen a river for navigation. Clearly this is not usually practicable for very large landslides. Enormous care is needed when undertaking this type of exercise to ensure that a catastrophic collapse does not occur.
6. Flow management: Sometimes attempts are made to manage the flow when collapse occurs. This is a hazardous activity as the flows can be exceptionally large and energetic.

Mitigation of landslide dams in Sichuan
So, where does this leave Sichuan? Well, the situation there is an absolute crisis, with bthe lake filling rapidly and rain forecast. Xinhua today reports that:
"About 1,800 armed police officers and People's Liberation Army (PLA) soldiers were hiking on Sunday toward an expanding "quake lake" in southwest Sichuan Province, hoping to blast away its landslide barrier before it bursts and causes a flood. The rescuers have 10 kilograms of dynamite each and are expected to arrive at the site on Sunday night," a PLA spokesman told Xinhua early on Sunday. The team were about two kilometers from Beichuan county seat at 5 p.m. Their trekking would be long and hard, with high mountains to climb and potential landslides. "We've found a local guide in Beichuan County who is willing to take us there," said Gong Juncang, an officer. "We tried once before, but couldn't make it to the top after 12 hours." The local meteorological bureau forecast high winds and thunderstorms for the area on Sunday and Monday. The Tangjiashan quake lake, which is in danger of bursting as water builds up in it, is one of the more than 30 such lakes in rivers blocked by landslides from the earthquake and thousands of aftershocks. Its barrier is in danger of bursting as the water level rose by nearly 2 meters on Saturday to 723 meters, only 29 meters below the lowest part of the barrier, which measured 752 meters high."

There can be little doubt that this lake needs mitigation very quickly and one has to sympathise with the plight of the Chinese authorities, who cannot get access to it at the moment due to the weather and the amount of debris, However, blasting the dam with dynamite would be a grave risk. I suspect that the view is that triggering a collapse now, catastrophic though that might be, is preferable to allowing the lake to overtop and collapse with a much larger volume. It is an appalling dilemma.