Attabad, landslide dam failure imminent(?)

Dave Petley continues to inform the world of the events at Attabad. Compare this image of the situation yesterday with the photo below of the situation today. It is helpful to locate a few markers: (1) the white rock projecting toward the waterfall from a position just to the right of the low point in the road at the top of the photo; (2) the prominence at the top left where it looks like a large rock sticks out into the lake just where it starts flowing into the spillway, and (3) two large rocks on the slope on the left side of the steep waterfall. The upper rock closest to the water is grey with a white "nose", and the other rock (directly below it in the photo) is flat and white. (4) There is also a convenient marker, a black vertical line toward the bottom center of the photograph.

Using these markers, you can see that the water from the lake is accelerating toward the waterfall just above the "grey" rock. The waterfall over the steepest part of the slope is turbulent, frothy, and has no exposed rocks to speak of. The gully at the bottom of the waterfall is an area of erosion. Compared to the next photo taken at a slightly different angle, the water disappears behind a ridge, but it's not possible to see what is going on behind this ridge.



Now look at today's picture. For scale, there are two people just above the white marker rock  on the right side of the river.  The upstream conditions where the lake starts to flow into the spillway are about the same, but the situation is changing dramatically between the grey boulder on the right and the pointy rock over by the road. The channel has noticeably widened, exposing rocks in the channel. A critical place to look is at the changes in the channel between the three rocks that I've defined as reference points. It is widening significantly in this area. Dave proposes that the water is now eroding the landslide itself instead of just the spillway materials. He says "The waterfall ....has create(d) a set of rapids that are clearly eroding back up the channel. The head of the rapids are close to the saddle. The key point is probably the location where the channel became notably steeper--this is where scour accelerates. Once this point is reached and passes the saddle, the rate of flow will start to increase and we might well see the breach developing. Unless the top of the rapids is being impeded by a large boulder or similar, this will probably develop quite quickly....downstream communities need to be prepared for a rapid breach."

There are three ways that this channel is eroding: expansion of the width, deepening of the channel by bottom erosion, and retrogressive erosion. I was fortunate enough to be able to document the processes that Dave is describing today in a study of one of the rapids in the Grand Canyon in 1983. A debris flow (technically different than a landslide) roared down Crystal Creek in 1966 and clogged the Colorado River, forming one of the most difficult, and deadly, rapids on the Colorado. Discharges through the Grand Canyon are controlled by the discharges through Glen Canyon Dam. In 1983 a series of events forced the Bureau of Reclamation to increase the discharges to three times the level that had ever passed through Crystal Rapids.

The cartoon on the upper right shows what happened, and happens as floods of different sizes attack barriers such as debris flows or landslides. In the Grand Canyon, debris flows episodically block the path of the Colorado, and so the starting point for the erosion is arbitrary. At low discharges, the river will erode a relatively narrow channel through the saddle point of the debris flow. At the rapids of the Colorado, this saddle point is always on the opposite side of the river from the side-canyon of origin of the debris flow. However, landslides can have such high velocities and so much mass that they shoot across the main river channel and bank up against the far side. In this case, the saddle point isn't necessarily against the far wall as shown here, but can be in the middle or even proximal part of the landslide.

Increasing discharges widen the channel, and scour out the bottom (not well shown in these cartoons). There can also be erosion through the debris flow either from the upstream side (not likely in most cases because flow velocities are low on the upstream side) or the downstream side. The latter is the regressive failure, also called headward erosion, that seems to be dominating at Attabad.

When the headward erosion reaches the saddle point the floor of the channel will be dramatically lowered at this critical point. Water velocities will increase causing erosion to increase, and this is the time when the landslide lake may pour out and into the valley of the Hunzen River.

We who are watching this unfold hope that the people of the Hunza Valley are safe.

Water flowing through the spillway at Attabad

On May 29 water started flowing through the spillway at Attabad. The photo here, from Dave's Landslide blog, shows an early stage of the overtopping. The spillway is working well at these low discharges. There were some reports of landslides into the lake today, and estimates that these slides had increased the lake level by 2 feet, but I don't know if these have been verified. As of this a.m. the discharge through the spillway is estimated to be 4 cusecs (cubic meters per second).

The lake level is rising about 2.5 cm per hour, and inflow into the lake is estimated to be 71 cusecs. There is thus about a 7 cusec difference between inflow and outflow. This water is stored in the lake causing its level to rise. As it rises, the discharge through the spillway increases. One limiting case is that the spillway channel will stay at the current width, in which case the increase in discharge is directly proportional to the rate at which the lake level rises and increases the depth of the water flowing through the spillway channel.

At the other extreme, if the spillway erodes and widens, then the discharge will increase as both the area increases and the depth increases. Erosion at the bottom of the channel would increase the depth of the flow through the spillway even further. It is this later case that worries the landslide experts because the discharge could increase so rapidly under these conditions. Dave is updating his blog frequently, so tune in there rather than here!

Update on the Attabad landslide

According to the latest report on Dave's Landslide Blog, there is less than 2 meters freeboard before the lake behind the landslide dam reaches the engineered spillway. Monitoring has been difficult because no one is allowed on the dam due to the hazard. I recommend his most recent report in which he addresses the difficulty of predicting exactly when the overtopping will occur, discusses some likely scenarios, and criticizes the shape of the spillway (too narrow, too shallow, and too steep). (Remember that there's not a lot of time for engineering solutions to be implemented--in this case, less than 4 months.)


Before and after photos of the erosion of a landslide dam. Photos from Dave's Landslide Blog, but not sure where he first got them.

He compares the current situation at Attabad with a similar situation that he monitored at Tangjiashan, China, two years ago. (Here's a direct link to his posting on that at a comparable time before breaching. Read forward and backward to see the whole history.) For those who want material for teaching, this is a great example of what the failure of a landslide dam looks like. The dam was formed in 2008 by the Sichuan earthquake. In this case, the spillway was engineered to start eroding when the lake overtopped and started flowing through it. It was not as steep as the Attabad spillway (this is good), but was also too narrow. Five hours after flow started, the dam failed, but the engineering solution had worked fairly well, and Dave is hopeful that the outcome at Attabad will be like this, rather than more catastrophic. Unfortunately, the dam at Attabad is narrower, and so it will be more prone to fail entirely instead of just along the spillway axis. It is also composed of smaller particles than the Tangjiashan dam and will therefore be more erodible.

I have not been able to find out details of evacuations and preparations downstream. Some information is contained on Dave's WWW site in earlier postings.