Was the Sichaun earthquake ‘boosted’ by reservoir loading?

There’s been a lot of discussion about the effects of the Sichaun earthquake, such as the Tangjiashan ‘quake lake’ which has now been drained in a safe, if not strictly controlled, manner (the BBC also has some cool photos). But aside from Kim’s excellent summary of the tectonic context, we haven’t talked much about the earthquake itself. According to this article in the Washington Post, although the Sichaun region was by no means seismically inactive, no-one was expecting such a large earthquake there:

many scientists were caught by surprise by the magnitude of the China earthquake, estimated at 7.9 by U.S. scientists. Sichuan province has a history of earthquakes, but none so devastating. It was not near the top of anyone’s list of the most likely locations for a great quake. The data from satellites, which can track the motion of vast plates of the Earth’s crust, suggested a relatively moderate amount of strain building up in the rugged mountain front along the edge of the Sichuan basin.


Of course, to a certain extent we are suffering from the limitations of the fact that just like volcanoes, the cycles of elastic strain build-up and earthquake rupture associated with large faults occur over timescales of hundreds or thousands of years. This is much longer than the interval over which we have complete and accurate records of where, how often and how strongly any particularly region has been shaken up (although it’s likely that China’s historical records are amongst the best anywhere), and thus any estimates of seismic risk can easily miss faults that may not have ruptured when people were around and able to record the event, but are still be accumulating strain nonetheless.
However, it seems that another theory is doing the rounds: that the weight of water in the reservoir behind the Zipingpu dam (which is only a few miles from the epicentre of the Sichaun quake, and was damaged by it), may have caused an unusually large rupture. Fan Xiao, the person being interviewed in the excerpts below, is the chief engineer of Sichuan’s Regional Geology Investigation Team.

Wang Yongchen: Now, two weeks after the powerful Wenchuan earthquake, what are your thoughts?

Fan Xiao: This earthquake was a Richter scale magnitude 8 and, strictly speaking, it is very unusual for this area. The historic records show that the highest recorded earthquake in this area was magnitude 6.5, and no seismic activities of more than magnitude 7 occurred along the Longmenshan seismic belt before. (see map here)

Wang Yongchen: Do you think this unusual quake was triggered by the reservoir?

Fan Xiao: Very likely, given its geological conditions, and the epicentre being so close to the reservoir.

Wang Yongchen: So, you are saying that the reservoir is likely to have induced the earthquake, but how? What are the conditions that trigger an earthquake?

Fan Xiao: There are many conditions, but from past experience and cases, both inside and outside China, at least three conditions are really important. First, the dam should be high enough, at least higher than one hundred metres. Second, the dam should be big enough, generally with a storage capacity of 1 billion cubic metres or more, because a small reservoir wouldn’t create significant threats. Third, and very important, the reservoir should be located in a fault zone. The case of the Zipingpu meets all three conditions: it has a height of 156 metres, a storage capacity of 1.126 billion cubic metres, and the reservoir is built in a fault zone. It wouldn’t be crucial if the reservoir was located in an inactive fault zone, but it does matter if a reservoir is located on an active fault belt. It’s unfortunate that Zipingpu is exactly this case: it is located on an active fault belt, with the tail of the reservoir really close to Yingxiu, where the quake occurred.

I’ve read before about small-scale seismic activity associated with reservoirs, but as I recall, it’s pretty low-level: stuff that would only register on nearby seismometers. Still, I can see how additional loading from a water-filled reservoir could cause a nearby fault to rupture earlier than it would have done naturally. What is more difficult for me to believe is the implication that the unusual magnitude of the earthquake is also the result of it being triggered in this manner: how could what is, in the grand tectonic scheme of things, a pretty small shove, boost the strength of an earthquake by more than an order of magnitude?
Going back to the Washington Post article, Ken Hudnut of the USGS is quoted explaining that the Sichaun earthquake looks like it may not be the result of a single fault rupture. Instead there was some sort of ‘cascade’ effect, where the original fault triggered a rupture on a separate fault nearby, which went on to trigger a rupture on yet another fault. The energy from these three near-simultaneous quakes then reinforced each other to produce the intense shaking which caused so much damage. This makes me wonder: if this proves to be the case, could loading from the Zipingpu reservoir have resulted in faults more finely poised to rupture, and more likely to be triggered sequentially in this fashion? If so, perhaps – but only perhaps – the dam was a factor, after all.

Categories: earthquakes, geohazards, geology

Comments (15)

  1. Kim says:

    Hmmm. Hydro-tectono-stuff is Maria‘s expertise more than mine, but here are a few things that might be important:
    1) The problem with this reservoir shouldn’t be the weight so much as the increased fluid pressure in rock below the dam. (It was a thrust earthquake, so more weight alone – say, from dry rock – would increase friction on the fault plane and make it less likely to slip.) On the other hand, fluid pressure is known to make earthquakes more likely, and has historically caused magnitude 5-ish earthquakes (in the Denver, Colorado area, which isn’t nearly as tectonically active as the edge of the Himalayas).
    So if the dam triggered the earthquake, then presumably the water from the reservoir is hydrologically connected to the fault. (Perhaps there’s a network of fractures that connects to the fault – I don’t know if the fault itself cuts the surface; active thrust faults typically are blind.)
    2) A magnitude 8 earthquake requires a large area of fault to slip – a hundred kilometers long, and as deep as the crust is brittle. If slip occurred on three separate faults, presumably either the stress caused by the fault slip or the transient stress from the passing of the seismic waves set the adjacent faults to slipping. (And this is where I wish Maria was around to answer questions, because she’s a lot closer to the active research on this stuff than I am.)
    So my guess is that the water could have been the trigger, but that the size of the earthquake was the result of the existing fault geometries and the way that thrust faults interact. (But that’s just a guess – I hope some of the people who do mechanical modeling of thrust faults look at this, because it’s a fundamental question, and could be very important for understanding the seismic risk of any place with active thrust faults. That’s why people like Jim Dolan are looking at it and worrying – seismologists have considered the possibility of cascading thrust ruptures for the Los Angeles basin for the past fifteen years.)

  2. Divalent says:

    What broke the camels back? 1) a piece of straw, or 2) a coffin of lead bricks?
    My naive understanding of faults is that they rupture when the strain reaches a certain point, and that a dam/reservior could only affect *when* that point is reached, not *if* that point is reached. (And I naively suppose that which side of the fault it resides on would determine if its effect would be to advance or delay the time of rupture)
    Relative to the normal buildup of strain on a fault, what would the presence of the dam represent? A year’s worth? A decades worth? A month? etc.

  3. Harold Asmis says:

    I’ve read the article, and using physics, it’s not the ‘weight’ of the water, but the power of water injection that does the stuff. Using that, we can expect a dam to fire off some 6′s, or maybe a 7. The smoking gun would be if the initiating earthquake was within the ‘injection zone’ of the dam.
    If I was writing a new episode of Superman, I wouldn’t use a nuclear bomb, I would build a high dam!

  4. Miguel Vera says:

    Putting the dam scenario aside, I don’t think the magnitude and the earthquake itself were ‘that’ unexpected. I couldn’t help noticing that there were 3 journal articles published not so long ago considering in one way or another the possibility of a strong earthquake in the region in future years. One of them even considers the M>=7.0 possibility, although it seems to be based solely on statistics. Do you think we could be underestimating the tectonics of the region?
    The ‘cascade’ rupture, though, seems like a good explanation, but even then I’m not sure about the dam.

  5. Maria Brumm says:

    There are plenty of examples of earthquakes that jump from fault to fault without needing any reservoir loading; the Denali earthquake was one. You could probably find dynamic rupture modeling folks who’d tell you differently, but I don’t think we understand this process nearly well enough to do anything more than speculate about the role of fluid pressure.
    It’s the sort of speculation that would really improve after a few trials of the beer can experiment, I betcha. Divalent, that one’s for you especially – the way fluid pressure triggers earthquakes is by reducing the “threshold” value, not by adding more stress. (Faults don’t really have a constant threshold of strain that must be reached before an earthquake happens, but that’s still a decent first approximation for thinking about what’s happening.)

  6. djlactin says:

    hmmm…
    “Three Gorges Dam” will have a total volume of about 39.3 cubic km, roughly 3.9 x 10^10 Tonnes. It’s about 400 km downstream from the Beichuan epicentre. Does anybody know the tectonic history of the dam site? Nasty problems if the dam is breached: downstream cities include Wuhan, NanJing and Shanghai (total urban population ca. 35 million).

  7. Chris Rowan says:

    Good points, all – especially about the role of increased hydraulic pressure in ‘weakening’ faults (it’s been some time since I’ve played around with Mohr circles…). So could reduced friction at the slip interface result in a larger rupture, and hence a larger earthquake?
    When we finally have a geoblogger meet-up, the beer-can experiment should of course be the centrepiece. Multiple trials will be required for robustness’ sake, of course.

  8. Maria Brumm says:

    So could reduced friction at the slip interface result in a larger rupture, and hence a larger earthquake?
    If that effect is something that’s statistically different from just causing more earthquakes generally, it should result in a smaller b-value for reservoir-induced seismicity. (De-jargoning note: the b-value is a ratio of the number of small earthquakes to the number of large ones.) I’m not sure if anyone’s done that analysis or not, but it wouldn’t be difficult.

  9. Divalent says:

    Chris: “So could reduced friction at the slip interface result in a larger rupture, and hence a larger earthquake?”
    Maria: “If that effect is something that’s statistically different from just causing more earthquakes generally, it should result in a smaller b-value for reservoir-induced seismicity.”
    Or (from my naive perspective) if the reservoir also reduced the threshold, the quake you get might be smaller because there was less stain to be released, and so the overall effect might be hard to predict as a general rule (but instead may vary depending on the particular configuration of fault and reservior).
    Sounds like something that would benefit from being hashed out at a beer can demonstration. [BTW, does it work with beer bottles? I can't remember the last time I drank a beer from a can.]

  10. Maria Brumm says:

    IIRC bottles are a little too heavy to work properly. I think Guinness is probably the best beer that’s commonly available in cans, but next time I’m at the fancy beer store I’ll have to check to see if there’s anything else suitable for a snobby-beer beer can experiment…

  11. christie says:

    Chris – please contact me via email, I can’t find yours via google but I have a question for you re. sampling equipment
    christie.rowe@uct.ac.za

  12. djlactin says:

    p.s. transliteration note: Sichuan; not Sichaun.
    (Sichuan means ‘four rivers’.)

  13. John E Johnson says:

    Not to be the guy beating the GHG drum, but…Has anyone postulated global climate change as a cause or influencer of this earthquake? I read an article where the retreating ice sheets caused quakes in Sweden at the end of the last ice age. http://findarticles.com/p/articles/mi_m1200/is_n18_v150/ai_18848787

  14. The winter in China was exceptionally severe, thanks to global cooling, so when the frost thawed from the ground…
    Nah.

  15. llewelly says:

    John, according to the article you link, the quakes were a result of iso-static rebound after the ice sheet melted. So, that kind of quake cannot occur unless ice sheets are present and melting. Thus, it’s only a concern for Greenland, Antarctica, Patagonia, and parts of the Himalayas.