The humbling legacy of the Tohoku earthquake

A post by Chris RowanA year ago on Sunday, one of the biggest earthquakes ever recorded ruptured the subduction megathrust that dips beneath the east coast of Japan. The rupture displaced the seafloor by tens of metres and generated tsunami waves up to 20 metres high, which hit the coast less than an hour later with devastating effect. My abiding memory of the morning I woke up to news of the earthquake, other than the near-constant pinging of my iPhone with alerts for all the aftershocks, is watching a helicopter video of the tsunami, having overwhelmed any coastal defenses, rushing across the Japanese countryside, sweeping away everything in its path. It was hard to believe wasn’t just something generated in a computer for a new disaster movie.

There were 19,000 casualties, with an additional 3,155 people still listed as missing; more than 370,000 buildings were destroyed or damaged; 300,000 people are still homeless. The estimated total cost to the Japanese economy is more than 300 billion US dollars. All this in a country that is well aware of its violent seismic history, and has taken the risk of future large earthquakes very seriously indeed. This investment was not in vain: compare the casualties and damage caused by the Tohuku event to the magnitude 9.2 Sumatra–Andaman earthquake on Boxing Day 2004, and the resulting tsunami that killed 230,000 people; or the much, much smaller magnitude 7 earthquake that struck Haiti in January 2010 and killed over 300,000. Japan’s strict building codes, sea walls, warning systems and preparedness drills were surely mostly responsible for this order-of-magnitude difference, and yet it still wasn’t enough. That is a humbling reminder of how even our most advanced societies are still at the mercy of the planet we occupy.

But the Tohuku earthquake has handed earth scientists an even more humbling lesson. One reason that the tsunami overwhelmed Japan’s coastal defences so thoroughly is that it met seawalls whose designers had underestimated how big a wave they might have to repel; an underestimate that was based on our faulty understanding of how big faults actually behave. Here’s how we thought it worked: long plate boundary faults such as the subduction megathrust beneath Japan could be divided into discrete segments that behaved quasi-independently. Each segment produced earthquakes of a characteristic size with a characteristic average repeat time, although an earthquake on one segment could affect the timing of rupture on its neighbours due to changes in the local stress.

This understanding was based on our instrumental and historical records of earthquake activity. These records have relatively short durations compared to the timescales of strain build up and release on large faults, so we’ve always known that extrapolating recent behavour over longer time periods might not give us the full picture. But what Tohoku and other recent megaquakes have shown is that it’s not just parameters such as the length of the earthquake cycle on individual segments that can vary over longer timescales, but the nature of the segmentation itself. The really big earthquakes seem to be cases where several segments of a fault all rupture in concert; even more disconcerting is the fact that magnitude 7 and 8 earthquakes every few decades or centuries along a particular stretch of the subduction zone does not rule out a magnitude 9 occuring in the same region every couple of millenia. Rather than being a simple saw-tooth wave of slow strain build-up and rapid release, we’re starting to see that faults sing a much more complex song, with a number of different frequencies mixed in. For most of the time we’ve been recording earthquakes, we’ve been hearing the high notes; we’re only now becoming aware of the base line – the longer earthquake cycles that govern larger, more potentially catastrophic ruptures.

The past decade has been full of rude surprises for geologists: starting with the 2004 magnitude 9.2 Sumatra–Andaman earthquake, the earth has rung with the vibrations of 5 quakes of magnitude 8.5 or greater. As Thorne Lay puts it in this week’s Nature, The Tohuku earthquake is just the most recent in this sequence, all of which “have violated some theories of where and when great earthquakes can occur and what their consequences can be.” We’re starting to realise that there is a lot more tectonic and seismic complexity than we thought in the space between the smooth motion of plates over million-year timescales and the jerky motion of faults over decades, centuries and millennia, and that looking at the long-term behaviour of dangerous faults using tools like paleoseismology (which had uncovered evidence of past events that were similar in scale to last years tsunami, even before it happened) is just as important as measuring contemporary deformation using GPS.

But the most important lesson is this: that no matter how much we think we know, our planet still has an infinite capacity to surprise us.

Categories: earthquakes, geohazards, tectonics
Tags: , , ,

Comments (4)

  1. terry says:

    Is it possible that some great quakes in the past, before instrumentation, were at or over M9?

    The last decade seems anomalous but I have my doubts that it’s “unusual”. I would love to see a project that really combed through the historical record and reanalyzed some great quakes. I realize that’s a huge project though, with rather poor data overall.

    • Chris Rowan says:

      It’s not that magnitude 9 earthquakes are unusual in themselves, it’s just that they were thought to occur on sections of a subduction zone that didn’t rupture very often, thus allowing large amounts of strain to build up. In contrast, the part of the megathrust that ruptured last March had a known history of rupturing in magnitude 7-8 events, so was not thought to be at risk of producing a magnitude 9. Obviously, we were wrong about that, and the paleoseismology indicates that it does generate magnitude 9 events and tsunami every 1000 years or so, as well as smaller segments rupturing in magnitude 7-8 events in between. It’s the mixing of these two modes on the same part of the fault that is the big surprise.

      • terry says:

        I gather subduction zones around the world are now getting reassessed?

        We already know about Cascadia’s potential and I’m aware of some research in the Caribbean. I wonder where else?

  2. Lyle says:

    The modern study of earthquake deposits has also assisted in the understanding. There was an interesting article in the Japan Times showing the runup in 869 and 2011 and they were nearly identical:
    In a way we are fortunate that Japan has at least a 1200 year record of major events to date them exactly, although in a general sense the idea that there have been 3 major events in the last 3000 or so years is seen from just the deposits left. Interestingly the researcher was going to meet with government officials to show the maps of earlier events on March 23, but which obviously never happened.
    Clearly more costal swamps need trenches dug thru them to see if there are any deposits.