Strong earthquakes cause damage and loss of life primarily because they usually strike without warning. For example, the vibration that unexpectedly shook the San Francisco area in 1989 caused 63 deaths, heavily damaged the Marina District, and caused the collapse of a double-decked section of interstate in Oakland, California ▼. This level of destruction was the result of an earthquake of moderate magnitude . Seismologists warn that other earthquakes of comparable or greater strength can be expected along the San Andreas Fault system, which cuts a nearly 1300-kilometer (800-mile) path through the western one-third of California. An obvious question is: Can earthquakes be predicted?

Short-Range Predictions

It would be extremely desirable if imminent earthquakes could be predicted in the way in which volcanic eruptions sometimes can. Unfortunately, despite substantial efforts in earthquake-prone countries such as Japan, the United States, China, and Russia, no reliable method exists at present for making short-range earthquake predictions.

To be useful, such a method would have to be both accurate and reliable. That is, it must have a small range of uncertainty with regard to location and timing, and it must produce few failures or false alarms. Can you imagine the debate that would precede an order to evacuate a large U.S. city, such as Los Angeles or San Francisco, and the expense of doing so?

Research on short-range earthquake prediction has concentrated on monitoring possible precursors—events or changes that precede an earthquake and thus might provide a warning. In California, for example, seismologists monitor changes in ground elevation and variations in strain levels near active faults. Other researchers measure changes in groundwater levels, while still others try to predict earthquakes based on an increase in the frequency of the foreshocks that precede some, but not all, earthquakes. To date, neither these nor other avenues of research have yielded a reliable method for short-range earthquake prediction.

In contrast to short-range predictions, which aim to predict earthquakes within a time frame of hours or days, long-range forecasts are estimates of the likelihood that an earthquake of a certain magnitude will occur in a given place on a time scale of 30, 50, or 100 years or more. Although long-range forecasts are not as informative as we might like, these data provide important guides for building codes so that buildings, dams, and roadways are constructed to withstand expected levels of ground shaking and for policy decision-making on disaster prevention and response. Most long-range forecasting strategies are based on the observation that large faults often break in a cyclical manner, producing similar quakes at roughly similar intervals. In other words, as soon as a section of a fault ruptures, the continuing motions of Earth’s plates begin to deform (bend) the rocks until they fail (rupture) once more. To determine the characteristic interval for a given fault and, thus, forecast future chances of rupture, seismologists study the historical and geologic record of earthquakes generated by that fault.

Paleoseismology

Paleoseismology (paleo = ancient. seismos = shake) is the study of the timing, location, and size of prehistoric earthquakes. Paleoseismology studies are often conducted by digging a trench across a suspected fault zone and then looking for evidence of ancient faulting, such as offset sedimentary strata or mud volcanoes. A large vertical offset of the layers of accumulated sediments indicates a large earthquake. Sometimes buried plant debris can be carbon dated, allowing for the timing of an earthquake recurrence to be established.

One investigation that used this method focused on a segment of the San Andreas Fault that lies north and east of Los Angeles. At this site, the drainage of Pallet Creek has been repeatedly disturbed by successive ruptures along the fault zone  ▼. Trenches excavated across the creek bed have exposed sediments that have been displaced by several large earthquakes over a span of 1500 years. From these data, researchers determined that strong earthquakes occur an average of once every 135 years. The last major event, the Fort Tejon earthquake, occurred on this segment of the San Andreas Fault in 1857, roughly 150 years ago. Because earthquakes occur on a cyclical basis, a major event in southern California may be imminent.

Paleoseismology has also revealed that powerful earthquakes (magnitude 8 or larger) and associated tsunamis have repeatedly struck the coastal Pacific Northwest over the past several thousand years. Each of these events is attributed to slippage along a section of the megathrust fault associated with the Cascadia subduction zone located off the west coast from southern British Columbia to northern California. The most recent event, which occurred in January 1700, generated a destructive tsunami that caused massive flooding in the coastal lowlands of western North America and was recorded as far away as Japan. As a result of these findings, public officials have developed extensive public education materials about tsunamis, updated building codes, and retrofitted some of the region’s existing buildings, dams, bridges, and water systems to be more earthquake resistant. For example, the Ocosta School District of southwestern Washington State built North America’s first engineered vertical tsunami refuge atop a new elementary school. The structure consists of a flat roof about 10 meters (30 feet) above ground, accessible from four heavily reinforced stairways built to withstand a magnitude 9.2 earthquake and resulting tsunami. The rooftop has space for roughly 2000 occupants from the school and surrounding residences. This initiative took cues from the 2011 Japan tsunami, but more importantly, was founded on scientific discoveries about Cascadia’s earthquake and tsunami risks.

Calculating Probabilities

The chance that an earthquake of a particular magnitude will occur at a given place within an estimated time frame, or the earthquake probability, is determined by studying the past frequency of earthquakes on a fault and the rate at which strain is accumulating along a section of that fault. This is a surprisingly difficult thing to study—detailed studies of many faults are lacking, and every earthquake that occurs along a fault can change the strain on a particular segment of that fault. Despite these limitations, scientists have produced forecasts using models that combine historical, field, and laboratory data to create hazard maps that show the high- to low-probability of a damaging earthquake striking a certain region ▼.

California’s San Andreas Fault runs diagonally from southeast to northwest for nearly 1300 kilometers (800 miles) through much of the western part of the state. Using paleoseismic and recent earthquake studies combined with improved mapping and new seismic models, scientists have determined California has more than a 98 percent chance of having a magnitude6.7  or larger earthquake in the next 30 years. The highest probabilities occur along the Hayward and Rodgers Creek Faults in the San Francisco Bay Area ▼. Southern California and east and south of Los Angeles, where some sections of the San Andreas Fault have not ruptured in more than 300 years, have similar earthquake probabilities.

Mitigation: Minimizing Earthquake Hazards

The statement “earthquakes don’t kill people; buildings kill people” succinctly expresses the fact that falling structures are by far the greatest cause of casualties during an earthquake. Thus, in regions where there is a known earthquake hazard, houses, bridges, dams, and other structures should be constructed to withstand at least moderate shaking. In addition, communities need to adopt building codes that require inspection and retrofitting of existing structures to withstand seismic shaking ▼. But due to prohibitive costs, the important step of retrofitting existing structures is often not a high priority. For example, a recent study indicated that more than half of the hospitals in Los Angeles County would likely collapse in a strong earthquake, and official mandates call for retrofitting of those buildings by 2030. However, due to the costs of retrofitting, only a handful of the hospitals have actually begun construction.

Earthquake-Resistant Structures

The importance of designing new buildings to resist earthquakes and retrofitting older structures is perhaps best illustrated by comparing two earthquakes of similar magnitude—the 1988 Armenian earthquake (M 6.8) and the 1989 San Francisco earthquake (M 6.9). Most of the buildings leveled by the Armenian quake were constructed of unreinforced concrete, which collapsed into rubble—killing an estimated 25,000 people ▼. Although the 1989 San Francisco earthquake was very destructive, the death toll (63 lives) was 400 times lower than that of the Armenian quake. The difference was due mainly to building practices; in California, most buildings are either wood framed or built with reinforced concrete that resists collapse, whereas in Armenia the buildings that collapsed were built of unreinforced concrete. In 2023, a pair of earthquakes (M 7.8 and 7.7) struck the Turkey-Syria region, killing over 50,000 people and destroying 345,000 apartments and hundreds of thousands of other buildings, leaving millions without housing or shelter. The bulk of the destroyed buildings were either old, non-retrofitted structures, or newer buildings not built to withstand shaking of the magnitude experienced in the earthquakes. In areas that are not seismically active but have the potential for experiencing a strong earthquake, the challenge is to establish appropriate building codes and educate residents on proactive measures to minimize casualties and property damage in the event of a strong earthquake.

Although the collapse of buildings is the largest cause of earthquake destruction, other hazards, such as ground subsidence caused by liquefaction, landslides, and tsunamis, can also be devastating. The catastrophic 2011 Japan (Tōhoku) earthquake (M 9.0) is a sobering reminder of this fact. Because of Japan’s strict building codes, the buildings, bridges, and other structures built or retrofitted since 1995 withstood the ground shaking of this powerful earthquake extremely well; only a few buildings collapsed as a direct result of shaking. It was the massive tsunami triggered by the quake that claimed the lives of 93 percent of the estimated 16,000 people who perished in this earthquake. Ground subsidence caused by liquefaction and landslides also caused significant damage to buildings, roads, and utilities, such as water and gas pipes. Japan’s earthquake preparedness is arguably the best in the world, in large part because the country has been struck by 19 major quakes (M 7-7.9) in the past 2 decades. Had the 2011 Japan earthquake occurred in a region where structures are less well engineered, the casualties attributable to collapse caused by shaking would have been considerably higher.

Earthquake Engineering

The application of engineering specifically to design structures that can withstand strong shaking has produced remarkable advances in making communities safer in the event of an earthquake. This practice has a long history. For example, many ancient temples in Greece were built on reinforced marble foundations specifically designed to be resistant to shaking.

Modern advances in earthquake engineering have resulted in innovative retrofitting projects. In Utah’s capital, Salt Lake City, engineers successfully isolated the base of the 100-year-old capitol building and installed flexible rubber dampers to prevent the building from shaking violently in case a major earthquake is triggered on the nearby Wasatch Fault. And in Taiwan, the 508-meter (1667-foot) high Taipei 101 skyscraper, which was built a few hundred meters away from an active fault, houses the world’s largest tuned mass damper ▼. This 660 metric ton (728 short ton) steel ball hangs from a pendulum-like structure suspended between the 92nd and 88th floors of the building. In the event of a major earthquake, or severe winds from a typhoon, the pendulum sways to offset the movement of the building and prevent structural damage.

Earthquake Warning Systems

Because P waves are less destructive and travel faster than both S waves and surface waves, they can be used to provide a type of earthquake warning system. Japan has operated such a system for over 20 years. One use of this system is to trigger automated shutdown of power-generating facilities and braking of high-speed bullet trains.

During the 2011 Tōhoku earthquake, this system was used to send an automated alert to the nation’s television stations as well as to more than 50 million telephones. In the area closest to the epicenter, this alert gave residents about a 10-second chance to take cover. In Tokyo, a city of around 9 million people located about 370 kilometers (230 miles) south of the epicenter, citizens had about an 80-second warning before the strong shaking began.

Other early warning systems are in place in Mexico, China, Italy, and Taiwan as well as several other countries. As of 2021, an early warning system was in place for California, Oregon, and Washington, with plans to expand to other U.S. earthquake-prone regions. For example, in the 2022 M 6.4 Ferndale, California earthquake, more than 3 million people were notified by phone with a message to “drop, cover, and hold on.” These warning systems can help to reduce injury and death; however, they are of limited value to people very close to an earthquake epicenter. Further, some argue that false alarms could potentially cause more havoc than the benefits derived from what is admittedly a very short warning time.

• Successful earthquake prediction has been an elusive goal of seismology for many years. Attempts at shorter-range predictions (for hours or days) use precursor events, such as changes in ground elevation or in strain levels near a fault; they have not been reliable.

• Long-range forecasts (for time scales of 30 to 100 years) are statistical estimates of the likelihood that an earthquake of a given magnitude will occur. Long-range forecasts are useful because they can guide development of building codes and infrastructure.

• Paleoseismology is a tool used to make long-range forecasts. Because earthquakes occur on a cyclical basis, determining how frequently they have occurred in the past can give some insight into when they are most likely to occur again, allowing scientists to calculate probabilities of future earthquakes.

• Mitigation strategies in earthquake-prone regions include building retrofitting, as well as construction of earthquake-resistant structures with elements that reduce shaking. The development of earthquake warning systems has the potential to also reduce death and injury from shaking.

• paleoseismology: The study of the timing, location, and size of prehistoric earthquakes.

• precursor: Relative to earthquakes, an event or a change that precedes an earthquake and may provide a warning.