Tuesday, January 27, 2009

The dam and the Sichuan Earthquake

I get Science via Pony Express plowing through snow over the Continental Divide, I suspect, so I didn't see this news article until a couple of days ago:

A Human Trigger for the Great Quake of Sichuan?

The article is behind a paywall, so I'll try to summarize it: Last year's devastating M 7.9 earthquake in Sichuan, China may have been triggered by the Zipingpu Dam. At least, that was implied by an AGU talk (which I didn't see) by Christian Klose, and by an independent article in the Chinese journal Geology and Seismology. Klose's argument is that the weight of water changed stresses on the fault in exactly the wrong way: it increased the shear stresses (which make the fault more likely to slide), and decreased the normal stresses (which prevent slip). His evidence includes the type of fault slip (especially the initial slip) and the depth of the majority of the aftershocks. I don't have quite the right background to evaluate his evidence (especially based only on the abstract), but Science reporters Richard Kerr and Richard Stone describe the reaction at AGU like this: "Klose's listeners were intrigued but far from convinced. They wanted to hear more details about changing water levels and local, lower-level seismicity."

This isn't the typical fluid-induced earthquake, caused by an increase in pore fluid pressure. The Sichuan earthquake nucleated 20 km below the ground surface, and that's an awfully long distance from the water source near the surface. It sounds as though the size of the reservoir, the location of the reservoir compared to the fault, and the direction of fault slip may have combined in just the wrong way... if the reservoir was what made the fault slip then.

The dam would still be a trigger, rather than a cause, for the earthquake. The cause, ultimately, is the collision of India with Asia. (As Klose states in his abstract, "This region has been tectonically loaded for >10kyr.") But the dam could explain some of the surprising things about the earthquake - for instance, why a fault that was thought to slip every 2000 to 10,000 years would have gone now, with such horrible consequences.

It's intriguing and scary... and terrible. The earthquake was tragic enough. If humans were responsible for triggering it... well, I don't want to think about it.

(Except that I'm teaching a course inspired by John McPhee's book The Control of Nature in May, and I'm trying to decide which possible case studies to use this time.)

(H/T to Sciencewoman, who told me to watch for this news article... two weeks ago. I've finally seen it!)


Anonymous said...

hmmm ... this is all pretty intriguing ... my gut reaction is skepticism, but I need to learn more

Anonymous said...

passo per caso nel tuo blog
un saluto from Italy, ciao

Callan Bentley said...

I'm using The Control of Nature as required reading in my Environmental Geology course this semester. First discussion on its contents is in two weeks; I'll let you know how it goes (perhaps via a post on NOVA Geoblog.

kurt said...

The Zipingpu dam looks like a fairly ordinary dam from posted photos
(e.g., http://www.flickr.com/photos/taylormiles/2495597793/sizes/o/ ) in a region with steep topography, so the reservoir is relatively long and deep. Xiao Fan reports the dam to have a "height of 156 metres, a storage capacity of 1.126 billion cubic metres, and the reservoir is built in a fault zone."
The focus of the earthquake was at a depth of 19 km (ftp://hazards.cr.usgs.gov/maps/sigeqs/20080512/20080512.pdf ignoring the typo in the box)

If the stress from the weight of stuff is given by ρgh, then the reservoir water (G = 1) exerts a stress of (1000kg/m3)(10m/s2)(150m) = 1,500,000 N/m2 and the 20 km of rock (G = 2.6) overlying the focus exerts (2600kg/m3)(10m/s2)(20000m) = 520,000,000 N/m2.... meaning the water exerts about 0.3 percent of the overlying weight on the fault.
Seasonal water level fluctuations in the reservoir to prevent flooding during the rainy season are likely to be less than the full depth of the reservoir. Does this seem like a reasonably hair trigger? It is a fault under great tectonic stress…. I’m no structural geologist, so I don’t know.

Since average rock has a density 2.6 times greater than water, completely filling/emptying the 150 meter deep reservoir is equivalent to adding/removing 58 meters of rock. That would be an impressive landslide, but nothing unprecedented. I would imagine steep mountains made of fractured rock along a fault in such a wet climate might frequently experience massive mass wasting. It would be tricky to tease out a relationship of ancient landslides causing seismicity in the past, though, since earthquakes can obviously induce mass wasting (the record would be chicken/egg).

As a non-structural geologist, I’m curious. As you point out, at 20 km depth, we’re not talking about increasing pore fluid pressure, rather the added weight of the reservoir pushing down. This is a reverse/thrust fault due to horizontal compression, so strong σ1 is horizontal and weak σ3 is vertical. Wouldn’t increasing the vertical load increase σ3 and so decrease the differential stress, shrinking the Mohr circle and reducing the likelihood of slip?

Kim said...

Kurt - yes, the directions of the stresses from loading would seem to be wrong to trigger a thrust fault. The dam, at least, is on the footwall of the fault, though - I don't do mechanical modeling of stress fields, so I don't have experience with more complicated 3D fields (than the box model from a college structure class), but I wonder if the exact location of the weight is the problem?

The abstract gives the numbers that Klose used - 320 million tonnes of water, an increase in shear stress of > 1 kPa, and a decrease in normal stress of -4 kPa. It doesn't mention the orientation of the fault (though I could probably find that from aftershocks and/or the focal mechanism if I looked).

Callan - I've used The Control of Nature a lot in classes. I've started looking for other case studies to mix in, so the class keeps connected to things students know from the news.

Jim L. said...

Intersting article, and it seems like it might make sense. Did they do a comparison with how long the damn has been there?

Anonymous said...

The full news piece appears in various venues around the web, for those who don't have access to Science (e.g. here).

This wasn't on my radar screen. Thanks for pointing it out!

Lockwood said...

This is indeed tragic, if true. I hope you and the rest of the geobloggers out there keep us'ns on the wrong side of the paywalls informed regarding further discussion on this. I'm with Brian: skeptical, but stranger things have been shown true.

I used Control of Nature for a "Geology for Secondary Science Teachers" Class some years back. Once my students got past the idea that I actually expected them to read more than one book for a graduate-level class (gasp!) they seemed to like it. I think it's McPhee's best, and I've read almost all his books.

Elli said...

As one of your students that read The Control of Nature during an intro class, it was probably my favorite reading of that entire class. I'm planning on using the Iceland case-study in my new intro earthquakes & volcanoes course in the fall.

I've also had professors assign readings from Irons in the Fire for an optical mineralogy class that I'm contemplating using during mineralogy this semester.

kurt said...

The dam being on the footwall side of the fault is an interesting observation. Visually, I can see the concept of pushing down on the footwall reducing normal stess on the thrust fault, but that hinges on the assumption that the hanging wall is rigid enough to stay "erect," at least a little, during footwall depression. I imagine the hanging walls of thrust faults lose most of their shear strength (on a large scale) when they push over bends in the footwall, though (like when one transfers a sheet of flattened pie dough from wax paper into a pie plate). Without rigidity, the hanging wall would just droop down like a wet noodle laying on the footwall.

Depressing the footwall would have to occur either by compression of the footwall rock, or by adjustments on faults in the footwall (i.e., shearing along existing fractures). The compressibility of rock seems like a trivial effect in this case, especially considering the relatively small weight of water.
Shifts on faults seem more productive. If this happens, it really makes me wonder about the importance of landslides triggering earthquakes in an area because the effected area would be doubled if the footwall was also vulnerable.

Thank you for the post (and for the comments). As you know, Kim, I am just an ore jock, so I know about as much about structural geology as someone who watches House on TV knows about medicine. ;-)

Kim said...

Kurt - I would love to hear what Michele or Juliet or another rock mechanics guru thinks about the idea of the dam triggering an earthquake. There's been a lot of discussion of earthquake triggering that's seemed counterintuitive to me (ever since the Landers earthquake) - but sometimes when things are counterintuitive, it means my intuition is wrong. I'd like to hear the thoughts of someone who thinks about stresses more than I do.

kurt said...

Another speculative article on the Wenchuan earthquake is making its rounds. Dave's landslide blog has a good analysis of the paper.