Earthquake Prediction Just as Tantalizingly Close as Ever There’s an article in last Friday’s issue of Nature describing some changes in the rocks near the San Andreas Fault that occurred in the hours before two small earthquakes. Here’s the BBC’s writeup; for those of you who can sneak behind the Nature paywall, the original article is here.

A similar study was published sixteen years ago, not in Nature but in Science. The first author on the 1992 Science paper, Paul Silver, was also the second author on this week’s Nature paper. While the recent study measured stress changes along a fault using precise instrumentation installed in a pair of very fancypants boreholes, in 1992, Silver and his coauthor Nathalie Valette-Silver exploited a natural system that is also very sensitive to changes in the squeezing of surrounding rocks: California’s Old Faithful, in the wine country town of Calistoga.

So before considering the evidence presented in Nature this week, I’d like to bring you all up to speed on the latest developments in geyser-based earthquake prediction. It’ll be a short post.

The geyser in Calistoga is a commercial tourist attraction. At least part of its plumbing was man-made; it erupts through a well bore drilled in the late 1800s. Today, visitors pay $8 to see it erupt, and to spook the fainting goats and pet the llamas while they wait for the main event.

The geyser owners have been tracking the length of time between eruptions at Old Faithful for many years. Silver and Valette-Silver looked at this data, and found that while the eruption interval is usually regular enough to justify the nickname “Old Faithful”, every now and then it does something weird. “Weird” is loosely defined – sometimes the geyser starts picking randomly between two possible wait times instead of one, and other times visitors get consistently more quality time with the fainting goats – but in any case, Silver and Valette-Silver found about 190 instances of weird behavior in the 28-year period of record, and 3 of them occurred shortly before moderate regional earthquakes. After each of those earthquakes the geyser went super ginormously haywire, dwarfing the irregularities observed in the one or two days prior. While there were 5 other biggish earthquakes during those 28 years, there was no signal at the geyser either before or after them.

That’s pretty much where things stand today. The Calistoga geyser data set is still one of the longest records available for any proposed precursory signal; while weird behavior precedes earthquakes at a slightly higher rate than you’d expect purely by chance, it’s not so much higher that you can’t sniff at it, if you happen to be the sort of person who sniffs skeptically or disdainfully at things. The paper’s accumulated citations are mostly from people thinking about geyser eruption dynamics. We still can’t predict earthquakes.

I haven’t read this new paper yet, but I’m willing to bet a case of good beer that in 12 years, it’ll have made a non-splash, just like the Calistoga geyser.

  • Silver, P.G., Valette-Silver, N.J. (1992). Detection of Hydrothermal Precursors to Large Northern California Earthquakes. Science, 257(5075), 1363-1368. DOI: 10.1126/science.257.5075.1363
  • Niu, F., Silver, P.G., Daley, T.M., Cheng, X., Majer, E.L. (2008). Preseismic velocity changes observed from active source monitoring at the Parkfield SAFOD drill site. Nature, 454(7201), 204-208. DOI: 10.1038/nature07111


  1. Elfriede Steiner-Grillmair wrote:

    As a professional geologist, I have worked on my own time and money for the past 30 years on earthquake prediction. I am happy to say that I succeeded in developing an easy to understand theory around it which, however, needs much more computer capacity than I can afford on my own. This theory would also include a new gravitational model (SG-gravity – standing for “shiftig gyration gravity” or “Steiner-Grillmair gravity”) as well as an entirely new cosmology which eliminates the meaningless “dark energy” concept and other hypothetical constructs. In his PRINCIPIA MATHEMATICA, Newton suspected the existence of this new SG gravitational model but was unable to explain it.

    Anyway, the tiny steps in potential earthquake prediction you describe above are great, but it sure would lead to much faster success if adequate funding were provided to try out my method.
    For more details try and look for article, blog and other posts by Spiral Path.

  2. Andrew wrote:

    Oh, Jim Berkland loves the Old Faithful database, even though it’s inconclusive. Last time I saw him he kept bringing it up while I was trying to eat.

  3. Matt wrote:

    That’s pretty neat. I know earthquake prediction can be pretty good in the few seconds between damage starting downrange from the epicenter and the start of the quake – if only because the p-waves provide a little advance warning before the damage starts. Is there anything with a reliably larger predictive range than that, for instance is there any way to predict an earthquake something like 5 minutes beforehand?

  4. Maria Brumm wrote:

    @Matt: No, there is no reliable way to predict earthquakes, and anyone who tells you otherwise is a crank.

    The p-wave earthquake early warning systems can only tell us about earthquakes that have already started, so they don’t count as “earthquake prediction” per se.

  5. Silver Fox wrote:

    In the early 70’s while still an undergrad, I had a huge file of earthquake prediction papers. I found it fascinating, but then not all that much really happened. The attempt to predict continues. Maybe someday…

  6. Elfriede Steiner-Grillmair wrote:

    This comment should answer Matt’s question of July 16, 2008, about a longer predictive range for earthquakes than just a few seconds. Yes, there is – my SG gravitational model can predict large earthquakes months and even years in advance.

    As I mentioned in my previous comment of July 16th, some of the details may be found at in a member article, blog and other posts in the forum by my member name “Spiral Path”. Unfortunately, at the moment I don’t even have enough money for groceries, let alone any available funds to sufficiently update my computer equipment so that I could write a proper paper or book about it.

    Too bad that most of the scientific community seems to hold Maria Brumm’s view that “there is no reliable way to predict earthquakes, and anyone who tells you otherwise is a crank” (see above post by Maria Brumm on July 16), as such a view will permit the needless death and suffering of hundreds of thousand of future earthquake victims.

    If you don’t mind my designation as a “crank”, you may write to me at , and if my computer still works by then you may also get an answer.

Post a Comment

Your email is never published nor shared. Required fields are marked *