NEWS AND VIEWS

Where prediction is not possible

For those - such as Californians - who live at plate boundaries, earthquakes are a fact of life. Others, who may count themselves lucky, should think again.

London not universally accepted (and may not years for the recurrence of a magnitude 7 READERS elsewhere may be forgiven for apply to all faults), at least one can say earthquake on the New Madrid zone - not knowing that last week Britain had how much of a 'slip deficit' is likely to have have bred complacency about seismic what may have been its biggest earthquake accumulated at various parts of a plate­ protection east of the Rocky Mountains. this century. No one was injured, and the boundary fault , and put an upper limit on In fact , residents of the eastern United reported structural damage was limited to the size of earthquake that may occur. States are only now beginning to realize ruptured gas mains, cracked masonry and Away from plate boundaries, life for that their seismic risk (a term that takes collapsed chimneys. With a magnitude of seismologists is more difficult. Most into account not just intrinsic earthquake about 5, the event on the Anglo-Welsh intraplate earthquakes cannot be assigned hazard, but factors such as population border was only one-hundredth the size of to an identifiable fault zone, in part concentration, soil conditions and build­ last October's earthquake on the San because most are too small to rupture the ing codes) is comparable with that of Andreas fault; indeed, California has a Earth's surface, but also because the low Californians. As a result, Memphis (the few earthquakes this size each year. level of intraplate seismicity makes it largest city near the New Madrid zone) The difference is that the San Andreas difficult to identify 'active' zones. Most has just adopted seismic resistance require­ fault marks the boundary between two continental interiors are characterized by ments, and New York City is about to crustal plates, whereas Britain sits well a rather uniform state of stress, arising follow suit. within the borders of a plate that stretches from forces acting on the edges and base The buzz of regulatory activity may give from the Mid-Atlantic Ridge to Japan, of the plate (see M. L. Zoback et al. the impression that the intraplate earth­ and from the Arctic Ocean to the Mediter­ Nature 341, 291-298; 1989). Typically, quake problem is in hand, at least in prin­ ranean. The plate boundaries are where the deformation rate that results is only ciple; that what we lack in predictive skills almost all the action is: plot the world's one-thousandth of that at a plate boun­ can be made up for by engineering. But earthquakes and active volcanoes on a dary such as the San Andreas; to generate there is a worrying loose end, which calls map and the plates emerge, as in a game of large earthquakes the strain must some­ into question the ability to estimate the connect -the-dots. how be concentrated. seismic hazard of' an intraplate region: But what of the stray dots in the Arch Johnston, of Memphis State intraplate seismicity seems to be inher­ middle - the intraplate earthquakes? University, has pointed out that the ently patchy in time and space, and more They signal the failure of the simplifying largest intraplate earthquakes are associa­ so than can be explained simply by the assumption of plate tectonics, that the ted with areas of continental crust that statistics of small numbers. plates are rigid monoliths which deform have been weakened by an episode of A particularly striking example, only only at their edges. And whereas plate stretching and thinning, during the form­ recently recognized, is the Meers fault in tectonics provides a framework in which ation of a rift or a new ocean basin. (This is southwestern Oklahoma. A conspicuous to understand - and perhaps ultimately bad news for the heavily populated fault scarp, 30 km long, bears witness to a to predict - plate-boundary earthquakes, eastern seaboard of the United States, faulting event about 1,200 years ago (A. J. there is as yet no analogous theory for which was stretched during the early Crone & K. V. Luza Geol. Soc. Am. Bull. earthquakes in 'stable' continental crust. stages of the opening of the Atlantic 102, 1-17;1989). If the event was accom­ No seismologist can say, even after the Ocean.) But the correlation breaks down panied by an earthquake (which must be fact , why last week's earthquake occurred at smaller, yet still damaging, magni­ confirmed by showing that the fault where or when it did, or why it was the size tudes; notably, some of Australia's largest extends downwards several kilometres it was. onshore earthquakes have occurred in into the crust), the earthquake would have Of course, these questions cannot be very old, unrifted crust. had a magnitude of about 7. Yet today, answered with certainty for any earth­ If all intraplate earthquakes were as unlike the New Madrid zone, the area quake, but at least at plate boundaries one benign as last week's event in Britain, the surrounding the Meers fault is essentially can see (and measure) the direct cause of failure to understand them would repre­ aseismic. In the New Madrid zone, the the activity. For example, the Pacific plate sent an intellectual challenge and no likelihood of a major earthquake is is moving northwards with respect to more. But historical records show that estimated by extrapolation from the North America at about five centimetres past intraplate earthquakes have been as frequency of smaller earthquakes, but a year. If, instead of sliding smoothly by powerful as the great San Francisco earth­ how does one estimate the seismic hazard one another, the two plates are locked quake of 1906. In New Madrid, Missouri, of the Meers fault? together by friction on the fault, then three magnitude 8 earthquakes occurred A further obstacle to understanding is stress will accumulate at the boundary. in a three-month period in 1811-12, thrown up by the observation that, before When the fault does slip, it does so cata­ damaging structures 1,500 km away on the the event 1,200 years ago, the Meers fault strophically, in an earthquake whose size Atlantic seaboard. And, given modern had not moved for at least 100,000 years. depends, in part, on the amount of slip . population densities, even a modest earth­ If recurrence times can be as long as Experience with the San Andreas has quake in an urban area can be devastating: 100,000 years (and if there can be no led to the definition of several fault seg- ' last December's magnitude 5.5 event in seismicity in the meantime), there must be ments with different behaviour: some Newcastle, Australia , killed 12 people and areas of the continents capable of signifi­ creep aseismically; others slip frequently caused A$I,500 million worth of damage. cant seismic activity about which we know in small earthquakes; still others seem Seismologists are fond of saying, "Earth­ nothing at all. Frustrating for seismolo­ to store up energy for longer times and quakes don't kill people; buildings do". gists, perhaps, but there should be plenty rupture in great earthquakes. Although The long gaps between large intraplate of work for structural engineers. this idea of 'characteristic earthquakes' is earthquakes - for example, 500-1,000 Laura Garwin NATURE · VOL 344 . 12 APRIL 1990 583