atquake

Two recent papers (Shearer and Stark, 2011; Michael 2011) unfortunately perpetuate a long-standing notion that statistical analysis of the ~ 100 year instrumental earthquake catalog informs us about clustering of M~9 subduction earthquakes.  This fallacy, perpetuated now for decades, has misled us into categorizing subduction zones by seismic capacity based on plate age and convergence rate.  Shearer and Stark (2011) make the same error committed in the 1970’s by turning to the instrumental catalog to “test” the possibility of clustering of M9 earthquakes.  There are a number of serious flaws in this approach.  With regard to the greatest earthquakes, recurrence times can be many hundreds of years.  Cascadia varies from 240-500 years, with gaps as long as 1200 years.  Cascadia has likely had two superquakes in 10,000 years, with long term cycling and clustering of events that is now becoming apparent (Goldfinger et al., 2011;submitted).  NE Japan likely had its penultimate M~9 event in the year 869, (Minoura et al. 2001).  During the intervening ~1000 years, numerous smaller earthquakes in the 8.2-8.4 range used only a small fraction of the accumulated strain, requiring the eventual superquake of March 2011 (forecast by  Ikeda, 2005).   The Sumatran subduction zone (Sieh et al., 2008), Cascadia and NE Japan apparently each have long term energy cycling, with groups of smaller events punctuated by larger events and long time gaps in their histories.  These three zones are the only three with paleoseismic records long enough to make these observations.

A second problem is the authors, who are well aware of the problems with short records, attempt to circumvent what we consider to be a brick wall by using smaller earthquakes with higher frequencies.  This approach simply does not address the question of whether clustering of M9 earthquakes exists, but instead answers an entirely different question, that is have global rates of M> 7 earthquakes clustered in the 20^th century.   We see no direct connection between these two distinct questions.  One need look no further than Cascadia, which has a b value of near zero, to see the fallacy of this assumption.

A third problem is that the basic observation that M9 events have clustered twice in the 1957-1965 period, and again 2004-2011 is questioned because a mechanism is not known.   Numerous arguments have been made against both static and dynamic triggering, the only two obvious options, and tests of both of them have been made as well using, again, smaller earthquakes.  Geology has a rather sordid history of throwing out observations for lack of a good mechanism.  Plate Tectonics and the Missoula Floods come to mind.  Lack of a mechanism is not evidence, it’s simply a neutral observation that may or not be relevant.

To evaluate global clustering of M9 earthquakes, long paleoseismic records from more subduction zones are required, and it is unlikely that seismology can address this question.  Statistical tests do not address this problem because they use a much larger range of earthquake magnitudes, addressing a different question.  As with plate tectonics, the current absence of evidence for a mechanism, is not evidence of absence of global clustering.

References

Goldfinger, C., Nelson, C.H., Morey, A., Johnson, J.E., Gutierrez-Pastor, J., Eriksson, A.T., Karabanov, E., Patton, J., Gracia, E., Enkin, R., Dallimore, A., Dunhill, G., and Vallier, T., 2011, Turbidite Event History: Methods and Implications for Holocene Paleoseismicity of the Cascadia Subduction Zone, USGS Professional Paper 1661-F, Reston, VA, U.S. Geological Survey,  332 p, 64 Figures.

Goldfinger, C., Ikeda, Y., and Yeats, R.S., 2011, Superquakes and Supercycles, AGU fall meeting and submitted paper.

Ikeda, Y., 2005, Long-term and short-term rates of horizontal shortening over the Northeast Japan arc, Hokudan International Symposium on Active Faulting: January 17-24 2005, Hokudan City, Japan.

Michael, A. J., 2011, Random variability explains apparent global clustering of large earthquakes, Geophys. Res. Lett., 38, L21301, doi:10.1029/2011GL049443.

Minoura, K., Imamura, F., Sugawara, D., Kono, Y., and Iwashita, T., 2001, The 869 Jogan tsunami deposit and recurrence interval of large-scale tsunami on the Pacific coast of northeast Japan: Journal of Natural Disaster Science, v. 23, p. 83-88.

Shearer, P.M., and Stark, P.B., 2011, Global risk of big earthquakes has not recently increased: Proceedings of the National Academy of Sciences,  published ahead of print December 19, 2011, doi:10.1073/pnas.1118525109.

Shishikura, M., Sawai, Y., Okamura, Y., Komatsubara, J., Tin Aung, T., Ishiyama, T., Fujiwara, O., and Fujino, S., 2007, Age and distribution of tsunami deposit in the Ishinomaki plain, Northeast Japan: Annual Report on Active Fault and Paleoearthquake Researches, p. 31-46.

Sieh, K., Natawidjaja, D.H., Meltzner, A.J., Shen, C.-C., Cheng, H., Li, K.-S., Suwargadi, B.W., Galetzka, J., Philibosian, B., and Edwards, R.L., 2008, Earthquake Supercycles Inferred from Sea-Level Changes Recorded in the Corals of West Sumatra: Science, v. 322, p. 1674-1678.

It seems that the more “advanced” a society becomes, the shorter it’s memory.  The Andaman Islanders did  better in the 2004 quake than anyone else, and the previous big quake was hundreds of years prior.  Native Americans not only have a memory of the last Cascadia earthquake 311 years ago, they have a memory of the explosion of Mt Mazama (Crater Lake) ~ 7600 years ago.    We on  the other hand can’t remember much that happened before Twitter and Facebook.

Modern societies intentionally discard anything old.  So instead, we have to rely on science, not social memory for fill in what we have forgotten.  Even in Japan, the stone tablets warning of past great tsunamis (see http://www.google.com/hostednews/ap/article/ALeqM5hnlFOddxHicXMy-m1x3Lnd5OfLtg?docId=3186d9e8c263410eb1c1e60efabc62b1) were ignored mostly, and even they were not old enough to record the even larger tsunamis of the past, as many of them were washed away by the recent one.

Subduction zone earthquakes may have recurrence times of 500-1000 years, so human memory is really of little use.  In Cascadia, the 1700 AD earthquake was the most recent one, but not at all the largest.  The largest event probably occurred ~ 5900 years ago, and may have been ~ three times the energy of 1700.

Why was the Mw=9 earthquake in the Tohoku area a surprise?

The issue is that the maximum earthquake scenario is generally unknown for most places, whether they be in California, Sumatra, Japan or Cascadia.  This is because historical records are generally too short to include the full range of earthquakes that have occurred. This was the case in Japan, where the Japan trench earthquakes were thought to run in the M=8.4 range based on historical records.  Even though those records are long in Japan, they aren’t long enough to guarantee the safety of something critical like a nuclear plant.  It turns out there probably was a similar event to the March 11 event in the year 869, and another large one in 1611, but that wasn’t well known when the plants were built, and still we do not know much about them.  The same is true for most fault systems of the world. The construction standards are set for a ‘maximum credible earthquake” which in many regions is little more than a guess.  That may sound harsh, but these educated guesses are made all the time by expert opinion, and are rarely made on the basis of detailed field investigations, which is what are needed.  One solution to this is simple paleoseismology, which we now know how to do to get records 10,000 years long in some cases.  This wasn’t known in the 70′s, but it is now and is a very cheap way to directly answer the key questions, and remove much of the guesswork.

In Cascadia, the famous “orphan tsunami” in 1700AD was most likely not launched by the largest earthquake produced by the Cascadia Subduction zone.  That earthquake, viewed through the lens of 10,000 years of past events, looked to be pretty average.  Two  much larger events may have occurred ~ 5800 years and 8800 years ago.  So going back in time to the previous few earthquakes is not enough.  In Cascadia, more than ten earthquake cycles are needed to establish an MCE, maybe more.  Clearly the two largest events in Cascadia are the 11th and 16th events back in time,  5800 and 8800 years ago.  These events must be quite a bit larger than the others, and are recorded as “large” at numerous sites, including those with no modern sediment supply at all, so are independent of climate, catchment basin size, recurrence time or other issues.

Are the nuclear plants in California and elsewhere safe from the maximum earthquake that could occur on nearby faults?  I don’t think anyone knows.  From the geology side, we certainly don’t know what the maximum credible earthquake is beyond an educated guess, and even if we did, just knowing earthquake “size” is only part of the story; how does the ground respond, what are the peak accelerations, would it generate a tsunami or a landslide generated tsunami?  What other cascading failures could conceivably occur?  Like an airplane crash, it’s not usually a single catastrophic  failure, but some small unanticipated failure that leads to a chain of problems that the engineers did not anticipate.  In Japan, the containment vessels were not breached by the earthquake or the waves, it was the lack of power for the cooling pumps that let to the problems.  It’s not easy to think through all the possible failure cascades and design for them.  The loss of the Challenger was a good example of lots of engineers and lots of oversight, but they just forgot that o-rings get brittle when cold, and then there was the pressure to launch on time.  Nuke plants and airplanes are enormously complicated systems, and it’s not easy to think through every possible problem, but both are truly “failure is not an option” situations that we just need to admit we’re not that good at, and take a lot more care.

The Tohoku earthquake, March 11 2011

It was the third earthquake of the week, and at first, seemed no larger.  Earlier in the week we had been rattled by a 7.4 earthquake far to the north along the Japan trench.  In the JAMSTEC cafeteria we watched the tsunami warning issued by JMA along the Japanese Pacific coast.  Early the next morning, an other smaller but closer earthquake shook me awake ~ 6:30 am.  So the Sendai earthquake was the third of the week.  When it happened, about 40 of us were in a workshop for those involved in studying the Sumatran earthquakes of 2004-2010.  My first thought was that it might interrupt the workshop, and it was Friday afternoon just before 3:00, so not much time left for an interruption.  But instead of stopping at a few seconds, or maybe a minute as the earlier ones had, this one kept going, and going, and going.   In a room full of seismologists, we timed the gap between the P-wave and S-wave arrivals, and then started thinking about whether we should get out of the building.  The desks looked really flimsy, so duck and cover didn’t look good at all.  On the other hand, were were possibly in one of the safest buildings in Tokyo (NE outskirts in Kashiwa-Chiba).  The corners of the room had pillars at least a meter square, and looked really solid.  But we went outside anyway.  It’s amazing how much you can do in five minutes.  After about a minute of shaking, were were all outside in the courtyard, watching the flagpole on the roof of the 7th floor whipping through 60 degrees, and the dry rattle of the trees with last years leaves as they shook.   Since we were all there, we snapped a quick group picture during the quake.

The mainshock lasted 5-6 minutes, an eternity.  Then we felt strongly the very large aftershock that felt (and turned out it was) very close to our location.   I never realized you could feel the difference between the different types of waves; The P-wave is like a jackhammer under your feet, the S wave much more like an ocean wave.  We all felt a little seasick as the S waves went on for many minutes.   A few minutes later, we all watched the first wave of the tsunami arrive in Miyako on live streaming video on Kenji Hirata’s cell phone, while the ground was shaking.

Studying earthquakes in one thing, but riding through a great earthquake was different.   The aftershocks were nearly continuous for the next 12 hours or more.    It’s a long time for the earth to feel like the ocean.    Two days later, the wingtips of the airplane bounced up and down on the ramp with more aftershocks just before leaving for home.  Nothing more useless than an earthquake geologist just after a a huge earthquake, so we all flew out,and our Japanese colleagues of course stayed to deal with a new reality in Japan.  The attention had to turn to rescue and recovery, and our work will come later.