When the magnitude of the earthquake off Tohoku, Japan, was calculated to be 9.0, it joined the ranks of "megathrust" earthquakes -- exceptionally large ruptures involving 5 to 10 meters of slip motion on fault zones more than 100 kilometers in length. These massive earthquakes occur along Earth's subduction zones, that is, where two 150-kilometer thick plates of solid rock collide. One plate dives beneath the other to reunite with Earth's hot rocky mantle, and, in the case of major earthquakes, this movement consequently shudders the planet.
According to recalibrations of old seismograms by the US Geological Survey (USGS), 11 "megathrust" earthquakes with an 8.5 magnitude or greater occurred worldwide in the twentieth century. Ten of these 11 earthquakes occurred offshore or near a coast, nearly all with tsunami damage. So far in the twenty-first century, five such megathrust earthquakes -- with severe tsunami damage in four cases -- have occurred offshore.
Proper use of earthquake science advises against an overreaction to the Tohoku disaster, but also spotlights further dangers that policymakers must take into account.
Reading Earth. GPS monitoring tells us that Earth's tectonic plates move slowly and gradually, but the big jerks at their boundaries are rarely predictable in the short term. The recent clustering of megathrust earthquakes has precedent: Seven of the eleven twentieth-century megathrusts clustered in a 15-year interval from 1950 to 1965, and several of these seven ruptured adjacent segments of the subduction zones connecting the Kamchatka Peninsula in eastern Russia, the Aleutian Islands, and mainland Alaska. No megathrust earthquake occurred between 1965 and 2004, during which time advancements in seismometry upgraded from scattered stations of mechanical sensors with paper records to today's dense telemetered global networks of induction-feedback sensors with digital recording. Today's generation of earthquake seismologists have scant twentieth-century megathrust data for comparative studies.
A megathrust hiatus is not unusual either: The USGS recognizes only two such events in the nineteenth century. History tells us that more megathrust earthquakes could occur in the next decade, but we have no evidence that the recent rate of nearly one megathrust per year will persist for longer than that.
Projecting for rarities. The rarity of megathrust earthquakes makes them difficult to mitigate. Prior to this month, the largest recorded Japanese earthquakes -- in 1896 and 1933 off Sanriku, to the north of Tohoku -- each released five times less energy than the Tohoku event this month.
Japan is not the only perilous case. Without a fortuitous tsunami measurement in Japan in the year 1700, we would not know for certain that the US Pacific Northwest is also capable of a 9.0 megathrust event. Nearly every subduction zone that lacks bathymetric complications could conceivably sustain a megathrust earthquake. Whether we have evidence for a prior event or not, the world looks much more dangerous than in 2004, before the 9.3 Sumatra-Andaman earthquake killed more than 220,000 people. Still, megathrust occurrence is limited by the slowness of plate motion; 300 to 500 years are necessary to build up the tremendous strains that they release.
Seismologists distinguish between earthquake prediction and earthquake forecasting. Earthquake prediction is short-term, conjuring Hollywood images of sirens and radios ordering citizens to sleep in tents outside their houses. Earthquake forecasting is more like "pledge night" on PBS: earnest pleas for public investments to mitigate a distant possible catastrophe. The difference between the two, however, is that earthquake forecasting is possible now, and earthquake prediction isn't. Seismologists do not yet understand how earthquakes begin to rupture. Though much has been learned from laboratory studies of rock fracture and field studies of earthquake zones, neither practical nor theoretical models have thus far captured properly the dynamics of an imminent large earthquake.
Earthquake forecasting has progressed greatly. All large earthquakes have aftershocks, which are smaller earthquakes that relieve the residual stresses that develop in the chaos of the rupturing fault zone and the secondary faults that surround it. Since around 1990, seismologists have noticed that aftershocks tend to cluster in volumes of rock in which shear stresses on faults are predicted to increase rather than decrease. The net stress-change of an earthquake is negative, but typically a cross-like pattern of increased stresses is predicted by theory, with pockets of higher stress projecting beyond the rupture termination, as well as angled perpendicular to the fault. The stress increase is a few bars (one bar approximates atmospheric pressure), which is tiny compared to ambient stress in the rock. That this small increment of stress largely determines the pattern of aftershocks is no comfort: It tells us that a large rock volume was very close to failure before the earthquake.
Aftershock patterns benefitted earthquake forecasting when Ross Stein and his USGS colleagues discovered that the stress increments of past large earthquakes were good predictors of where the next large earthquake would occur. Long after the aftershocks subsided -- months, years, or decades after -- another earthquake of similar size often broke within the next segment of the fault zone, where stresses had been increased only slightly in relative terms. How time-delayed stress-triggering occurs is a mystery, but it has been documented worldwide.
An irregular series of large, damaging earthquakes shook the North Anatolian Fault in the twentieth century from east toward the west across modern-day Turkey, reaching the Sea of Marmara in 1999 with the Izmit earthquake. Stress increments from Izmit have loaded the fault segment next to Istanbul. The 6.6 magnitude San Fernando earthquake in 1971 loaded the nearby fault that caused the 6.7 Northridge earthquake in 1994. More germane to Japan, the 9.3 Sumatra-Andaman megathrust earthquake in December 2004 loaded the next subduction-zone segment to the south, and this segment generated an 8.6 megathrust event only three months later in March 2005. No prediction can be made today for Japan, but it is safe to forecast a sharply increased probability for a major earthquake on the broad, simple subduction-zone segments both north and south of the Tohoku rupture zone. The segment to the south lies offshore the Tokyo metropolitan area.
Adding strength to future policy. Global media are focusing on the implications of future earthquakes like Tohoku, but policymakers should not discount the dangers posed by earthquakes similar to the 6.3 earthquake beneath Christchurch, New Zealand in February this year. Though 10,000 times less energetic than Tohoku, this unusually shallow, unusually intense earthquake devastated a small, but highly urbanized surface area directly above the rupture zone. Although it occurred within an earthquake-prone nation, the Christchurch earthquake is instructive because, though associated with a larger 7.0 earthquake five months prior in New Zealand, it occurred in a region with little recognized earthquake hazard. The earthquake risk associated with stable geologic regions was recognized by the US Atomic Energy Commission in the early decades of nuclear energy. For years the commission supported regional seismographic networks throughout the United States to monitor small earthquakes and identify active fault zones.
The unfolding nuclear crisis in Japan seems not to be caused by a tragically flawed reactor design, as with Chernobyl, or an operational lapse, as with Three Mile Island. Few if any nuclear power systems could have survived the Tohoku earthquake and tsunami without major damage. Lessons will be learned from this disaster that can mitigate damage to nuclear facilities from future earthquakes, but not eliminate the risk entirely.
For perspective, geologic evidence tells us that a business-as-usual fossil-fuel energy policy will almost certainly lead to a sea-level rise from melting ice caps, comparable in height to that associated with the Tohoku tsunami. As with earthquakes, the main uncertainty is the timing of ice-cap melt, but all coastlines will be affected -- not only those next to subduction zones. Unlike with tsunamis, the seawater won’t recede.
As the events in Japan have illustrated, the country faces a crisis from this megathrust earthquake, but it also faces an additional crisis from nuclear power plants sited in the area that was damaged most. Nuclear power facilities are often sited near the coastline to use seawater as a convenient source of coolant, and are thus vulnerable to coastal earthquakes and tsunamis.
There are few risk-free choices in energy policy -- and it might not be too soon to rethink natural-hazard policy so that it includes energy policy.