The town that is in for a big shock

Press, 14 February 1986, Page 18

The town that is in for a big shock

Scientists are hoping for an earthquake in tiny Parkfield, California, in 1988. Dr Peter Smith, Reader in Earth Sciences at Britain’s Open University, explains why in an article which originally appeared in the “Guardian.” \

Earthquake prediction research has proved to be a depressing business. No-one has yet been able to devise a method of predicting even a small proportion of the world’s seismic events. It has even been impossible to find a way of anticipating a minority of the earthquakes occurring along the simplest of the three distinct types of boundary between the Earth’s lithospheric plates. Worse still, there is no known means of forecasting shocks within a single, wellstudied, local seismic zone. Yet American seismologists are confidently asserting that a moderate earthquake will occur near the California town of Parkfield, about midway between San Francisco and Los Angeles, sometime between now and 1993, most probably in 1988. What’s more, they claim to be able to specify the magnitude and location of the impending event with remarkable precision. The obvious way of trying to discover a viable prediction technique is to monitor phenomena likely to change in the period leading up to a quake. Ground level and tilt, tide and well-water levels, local magnetic fields, the elastic properties of nearby rocks and the rates at which radon gas is emitted from the ground are just a few of the things that have been observed to change before earthquakes. The problem is that such effects do not always occur; and even when they do, they seldom fall into a pattern regular enough to enable the time of the subsequent event to be pinpointed. There have, of course, been a few isolated exceptions. In 1966 the Japanese managed to predict the peaks of activity in a long series of earthquake swarms at Matsushiro, and in a much more spectacular way the Chinese were able to anticipate the large Haicheng shock of 1975.

But these were unusual achievements aided by the rare beneficence of nature. In each case there happened to be many different precursors and an adequate system (equipment and people) to monitor them. Despite more than 20 years of concentrated research in the United States, the U.S.S.R., Japan, and China, no-one has managed to extend these odd successes into a widely applicable technique. So in recent years the Americans have retreated to a cruder, but possibly more effective approach, based on earthquake recurrence intervals. The idea is simple, so simple, in fact, that it might hardly be expected to work. It is well known that there are many places at which shocks

occur time and time again. If the earthquakes at such sites were to occur at random times, this tendency to repeat would have no predictive value at all, except perhaps to identify potential danger zones. But what if the shocks were to occur at regular intervals?

To see why this is even a theoretical possibility, it is necessary to look at what happens in the simplest form of earthquake. Imagine a vertical fault in the Earth’s crust. Stresses exerted from below are trying to force the crustal block on one side of the fault in one direction and the block on the other side in the opposite direction. Nothing happens at first, for the fault is locked by friction. But as the stresses continue, the strain builds up until the friction is overcome. The fault then slips suddenly, producing a seismic shock, and the pent-up strain is released. The whole process then repeats and re-repeats, generating a series of quakes. Assuming no changes in the underlying stress patterns or in the fault characteristics, earthquakes should recur at regular intervals, in which case prediction becomes simply a matter of adding a certain number of years on to the date of the last shock. The regularity may not be perfect, however, for the Earth invariably turns out to be more complex than any simple model suggests. Moreover, even if the recurrence interval at a particular location is indeed constant, it could be indeterminable, because historical records may not go back far enough. Unfortunately, in many parts of California the recurrence intervals seem to be more than 70 years and sometimes more than 150; and so there are too few cycles within the State’s recorded history. But near Parkfield, things are different Earthquakes occurred there in 1857, 1881, 1901, 1922, 1934 and 1966, giving a recurrence interval only of 22 plus or minus three years. The 1934 event was slightly early, but the timing of the earthquakes is still remarkably regular.

From a numerical series, seismologists have concluded that the next Parkfield earth-

quake will strike in 1988 plus br minus five years. And simple though the method may be, the resulting prediction is being taken seriously. The National Earthquake Prediction Evaluation Council (N.E.P.E.C.), a scientific review panel established by Congress to assess the validity of predictions, has given this one its seal of approval. It is the first endorsement N.E.P.E.C. has given in its eight years of existence.

As a result, Parkfield, with fewer than 50 inhabitants, is now under siege. Seismometers have been installed to monitor seismicity (including possible foreshocks), laser systems are being used to detect crustal deformation, and dilatometers have been installed in boreholes to measure strain changes. Seismologists hope that now that the. broad prediction has been made, they will later be able ,to make more precise estimates of the timing of the earthquake in the light of data on conventioal precursors. At the very least they will be well placed to observe precisely what happens when an earthquake does take place. The sociologists and planners have also moved in. The California Governor’s Office of Emergency Services has commissioned a firm called Scientific Services Inc. to provide public response plans for the five local counties and for the state as a whole. In fact, a series of plans is being drawn up for warning times ranging from a few years to 15 minutes and for earthquake magnitudes ranging from 6 (which would affect only a few hundred residents) to 7 (in which case more than 150,000 people would be involved). As this is the first scientifically based prediction to be made public in California, it is the first time that the planning community has been called upon to prepare specific emergency procedures.

Faith in the validity of the prediction, and hence conviction that all the emergency-planning is necessary, has been bolstered by the fact that at least the last three Parkfield earthquakes appear to have been similar in magnitude, intensity, location, faulting mechanism and fault rupture length. The seismologists

concerned have therefore been , tempted to be quite precise 1 about the characteristics of the ’ supposed 1988 shock. They ex- < pect it to have a magnitude of J 5.5 to 6.0 and a focus within a specific region 25km long by [ skin wide by 3km to Bkm deep.

All the same, there are seep- ' tics. For example, Mr T. R. < Tdppozada, of the California De- ; partment of; Conservation, reck- ’ ons that the 22-year periodicity is > unproven. He argues that nine- < teenth century records of earthquakes in California are both ' incomplete ' and imprecise, t \ throwing doubt on the absence of , 'shocks at Parkfield between 1857 and 1881 and opening up the ' possibility that i an earthquake , recorded in 1885 may have had its epicentre there. He also claims that, far from being characteristic, the last two earthquakes (1934 and 1966) were actually half a magnitude weaker than earlier ones. If Mr ■■ Toppozada were proved right on either or both counts, the prediction would lose strength.

And even the scientists at the United States Geological Survey admit to . one complicating factor. In 1983 a magnitude 6.7 earthquake occurred at Coalinga, on a different .fault but only 40km northwest 'of Parkfield.

This event is*known to have x affected the pattern of creep \ along at least one section of the San Andreas fault and to have triggered an unusual swarm of small seisiiiic qvents just to the south-east of Parkfield. There is no knowing just how this will have influenced the time of the next earthquake at Parkfield itself. • I

But noneof this seems to have dampened 'the! confidence with which the seismologists have gone public With the prediction, a somewhat surprising confidence given the state of the science. Would the scientists concerned have been quite so forthcoming had the expected event been much larger and/or in a metropolitan area rather than near sparsely populated Parkfield? Possibly not, although the risk of appearing foolish has to be taken sometime, and it’s perhaps for the best that it should be taken where only a few hundred people are likely to be involved.

Even so, if the next Parkfield earthquake fails to occur, reasonably close to 1988, the seismologists will be in despair. Worse, they will lose credibility with both the public and Government, which could cost them research funds and co-operation in the future.