Friday, January 15, 2010

Can Earthquakes be Predicted? Yes, no, maybe probabilistically

Any skuleboy kno that earthquakes occur as tectonic plates slip on each other.

Everyone would prefer that the time they do his could be reliably predicted.

Experts agree that we are not yet in a position to make accurate predictions, but how close are we?

This is abstracted from Wikipedia on earthquake prediction
  1. The VAN method picks up seismic electric signals (SES) from conductive sensors in the ground, and claims to be able to predict earthquakes of magnitude >5, within 100 km of the epicentre, in a 2-hour to 11-day time window. Its claims are contested. (SES come from piezo electric effects from stressed quartz, which is used to provide the spark in non-flint cigarette lighters).

  2. There is evidence that the recurrence time of an earthquake in any place is related to its previous recurrence time.

  3. Satellites may be able to detect precursor low-frequency EMF changes prior to earthquakes.

  4. Animals such as dogs, cats, chickens, horses, ants and elephants have been observed to change behaviour in the days prior to earthquakes. More.

  5. 50% of major earthquakes have foreshocks, but only 5-10% of foreshock magnitude earthquakes are followed by a major quake.

  6. Ground water level changes occur before some earthquakes.
So we have some leads. The thing is that none of these, taken alone, has the predictive capacity required by the business community. Clearly, nobody is going to close down San Francisco for a week because some chickens are clucking uncontrollably, since they are going to look a bit sheepish is San Francisco fails to fall down in that week. (And if it does fall down, sceptics, being sceptics, are going to say it was a fluke and a coincidence). Susan Hough's book,
Predicting the Unpredictable: The Tumultuous Science of Earthquake Prediction , says there is no prospect of accurate prediction.

And yet...

The problem is that seismologists approach things on an analytic, single-modality basis. Experts are into their own thing, and obviously, a satellite man is going to look down (in both senses) on the lady who is fooling around with chickens and elephants. On the other hand, politicians have to make decisions to protect their populations to the best of their ability, and if I was mayor of a town or city sitting on a fault line (and many cities are located on fault lines, as they are mineral-rich) I would set up a collaboration between all modalities of prediction.

Each of the 6 methods taken on its own will only give a probabilistic estimate of an earthquake. Academic critics will demand 100% certainty before they will allow anything to go ahead, and 100% certainty is not available in this life. So, for their own ego aggrandisement and in pursuance of their obsessive compulsive tendencies, prediction skeptics will commit us to endless repetitions of what is now happening in Haiti.

Instead of having no predictions, and waiting forever for a single perfect predictive method that will never happen, it would be reasonable to work with what we now have, and take them in concert.

We have 6 methods available to us, each with its own uncertainties. If they could be integrated, we could get a very reasonable probability of what is about to happen. We can also give the affected population a probabilistic estimate. Early on, if one or 2 of the modalities are looking a bit ominous, low probability warnings might lead mothers and children to choose to take the weekend away from home, maybe go camping (tents do not fall on you). If four or five modalities are blinking, non essential workers might also decide to move out of the area, and essential workers might choose to wear helmets. If all six modalities are flashing red, the local authorities might care to issue a recommendation for everyone to clear off out of the danger area.

VERY IMPORTANT NOTE: compliance with these warnings is VOLUNTARY. Skeptics, individualists and free market fundamentalists are not required to comply with the warnings. They are at liberty to stay, and indeed other skeptics from outside the area will be allowed in, so that they can either crow to the cameras in the event of a false alarm, or, in the event that it was not a false alarm, so that they can submit themselves to evolutionary pressures.

This integrated, multi-modality early warning system should be audited for accuracy as time goes on. It might prove useless, or it might prove useful. If useless, it might cause a bit of loss to local businesses. If useful it might save thousands of human lives. ON balance, it is stupid not to try it out.

Summary: there are several modalities of earthquake prediction that, taken individually are subject to false positives and false negatives, but taken together, could provide effective probabilistic and voluntary warning of earthquakes.

PS: Is every window in San Francisco and other at-risk areas covered with adhesive film, so that if it shatters it does not fall and lacerate people in the street below?
I hope so.

See also:

No comments: