Can seismologists predict earthquakes?
The answer depends on the time frame of the prediction. With our present understanding of the distribution of seismic zones and the frequency at which earthquakes occur, we can make long-term predictions (on the time scale of decades to centuries). For example, with some certainty, we can say that a major earthquake will rattle Istanbul during the next 100 years, and that a major earthquake probably won’t strike central Canada during the next 10 years. But despite extensive research, seismologists cannot make accurate short-term predictions (on the time scale of hours to weeks or even years). Thus we cannot say, for example, that an earthquake will happen in Montreal at 2:43 P.M. on January 17. In this section, we look at the scientiﬁc basis of both long- and short-term predictions and consider the consequences of a prediction. Seismologists refer to studies leading to predictions as seismic-risk, or seismic-hazard assessment.
A long-term prediction estimates the probability, or likelihood that an earthquake will happen during a speciﬁed time range. For example, a seismologist may say, “The probability of a major earthquake occurring in the next 20 years in this state is 20%.” This sentence implies that there’s a 1-in-5 chance that the earthquake will happen before 20 years have passed. Urban planners and civil engineers can use long-term predictions to help create building codes for a region codes requiring stronger, more expensive buildings make sense for regions with greater seismic risk. They may also use predictions to determine whether it is reasonably safe to build vulnerable structures such as nuclear power plants, hospitals, or dams in a given region. Seismologists base long-term earthquake predictions on two pieces of information: the identiﬁcation of seismic zones and the recurrence interval (the average time between successive events).
To identify a seismic zone, seismologists produce a map showing the epicentres of earthquakes that have happened during a set period of time (say, 30 years). Clusters or belts of epicentres deﬁne the seismic zone. The basic premise of long-term earthquake prediction can be stated as follows: a region in which there have been many earthquakes in the past will be more likely to experience earthquakes in the future. Seismic zones, therefore, are regions of greater seismic risk. This doesn't mean that a disastrous earthquake can’t happen far from a seismic zone they can and do but the probability that an event will happen in a given time window is less. To determine the recurrence interval for large earthquakes within a given seismic zone, seismologists must determine when large earthquakes happened there in the past. Since the historical record does not provide information far enough back in time, they study geologic evidence for great earthquakes. For example, recognition of a fresh, unweathered fault scarp or trace may indicate that faulting affected an area relatively recently. A trench cut into sedimentary strata near a fault may reveal layers of sand volcanoes and disrupted bedding in the stratigraphic record. Each layer, whose age can be determined by using radiocarbon dating of plant fragments, records the time of an earthquake (figure above). By calculating the number of years between successive events and taking the average, seismologists obtain the recurrence interval. Note that a recurrence interval does not specify the exact number of years between events, only the average number. Since stress builds up over time on a fault, the probability that an earthquake will happen in any given year probably increases as time passes.
Information on a recurrence interval allows seismologists to reﬁne regional maps illustrating seismic risk (figure above a and b).
Short-term predictions, specifying that an earthquake will happen on a given date or within a time window of days to years, are not and may never be reliable. Seismologists have considered, and discounted as unreliable, many supposed bases for short-term prediction. For example, a swarm of fore-shocks may indicate that rock is beginning to crack in advance of a main-shock, but such swarms can be identiﬁed only in hindsight. Precise surveys show that the surface of the ground may warp slightly prior to an earthquake, but no one can determine how much warping will take place before an earthquake will happen. Prediction studies focused on measuring changes in water levels in wells, radon gas in spring water, electrical signals emitted by minerals, or agitation of animals have met with similar skepticism. The concept of a short-term prediction should not be confused with the concept of an earthquake early warning system. An early warning system works as follows. When an earthquake happens, the seismic waves it produces start travelling through the Earth. Seismic stations closer to the epicentre may detect an earthquake before the seismic waves have had time to reach populated areas farther from the epicentre. The instant that seismic stations detect the earthquake, a computer approximates the epicentre location, then sends a signal to a control centre, which automatically sends out emergency signals to areas that might be affected. The signals shut down gas pipelines, trains, nuclear reactors, power lines, and other vulnerable infrastructure. The signal also sets off sirens and alerts broadcasters to send out warnings on radio, TV, and cell-phone networks to warn people that an earthquake is about to begin. Unless the focus is directly under the city, the warning may precede the arrival of the ﬁrst earthquake waves by several seconds, not a lot of time, but hopefully enough to prevent some infrastructure damage and perhaps enough for people to seek a safer location.