Earthquakes within plates: we don’t know when, and we may not know where

ResearchBlogging.orgA post by Chris RowanEd has already given the lowdown on a new study in Nature which might lead to a rethink on earthquake hazards in the continental interior. Plate tectonics treats plates as entirely rigid entities, but continental crust is too weak, and too riddled with faults left over from when it was close to a plate boundary, for it to entirely hold up when subjected to the stresses of plate motion. So although a very large proportion of the Earth’s earthquakes occur at plate boundaries, there is also some seismicity – including some very large shocks – within plate interiors. The problem is working out where this intraplate deformation is going to occur, and to do so seismologists rely on data which serve them well at plate boundaries – the historical record of large earthquakes, and the location of low-level seismic activity which indicates the build up of tectonic strain.

What Stein and Liu argue in their paper is that away from the plate boundaries, these tools provide a very misleading picture. In apparently active parts of the continental interior like the New Madrid area, all the abnormal seismicity can be regarded as a long-lived aftershock sequence; rather than indicating any new elastic strain being built up by external forces, which could eventually produce another large earthquake in the future, the seismicity is just a local tectonic response to a historically recent large earthquake (in New Madrid’s case, it was a series of magnitude 7-8 earthquakes in late 1811 and early 1812), and will eventually die off with time.

This conclusion is a little worrying, since it implies that the next big intra-continental quake might well occur in what presently seems to be a seismically inactive region, which, given the density of old faults cutting through your typical chunk of continental crust, could be almost anywhere. We already know the difficulties of predicting when big earthquakes are going to occur, but it seems that in the middle of plates, predicting where they are going to happen might also be a bit more tricky than we thought. However, a caveat remains: the proposed length of a typical intra-continental aftershock sequence is hundreds of years, which is much longer than our instrumental records, and even historical records in many places. The authors do point out that earthquake patterns in China, which has the best historical record, is of single large quakes in different areas (with last year’s Sichuan quake being the most recent) rather than a series of large earthquakes associated with a particular fault; perhaps palaeoseimology can show whether a similar pattern holds further back in time and on other continents.

Stein, S., & Liu, M. (2009). Long aftershock sequences within continents and implications for earthquake hazard assessment Nature, 462 (7269), 87-89 DOI: 10.1038/nature08502

Categories: earthquakes, geohazards, geology

Comments (4)

  1. Harold Asmis says:

    I’ve written about this, but the implications stretch physics, and all experience with rock mechanics. It implies that all of those past quakes happened in ‘virgin’ rock, and are in a state of permanent stress lock. If overall rates are the same, we will get new quakes in unfractured, seismically quiet rock. This cannot be. It is more likely that they have missed the obvious mechanism of stress relief, and water-based stress corrosion. Which implies that these zones just grow, and you will never measure regional ‘strain accumulation’.

  2. Brian Shiro says:

    Stein and I worked on this problem 10 years ago when I was his student. This is the latest in a long line of papers from him that make the case for a relatively low seismic hazard in middle America. His position is somewhat at odds with the USGS’s estimates of hazard in the New Madrid region. I tend to agree with Stein because I know he’s one of the top scientists in geophysics, and I’ve worked with some of the GPS data and models he’s using to make his case.

  3. Chris Rowan says:

    It implies that all of those past quakes happened in ‘virgin’ rock
    I beg to differ; clearly these quakes are occurring on reactivated faults. The authors discuss this. It’s just that there are plenty of old faults to choose from.

  4. Harold Asmis says:

    If you take the rate of ‘new’ big earthquakes, and combine it with the half-life of the old earthquakes, there would be a lot of action building up. Do we truly have no clue on where the next one is most likely to occur?