Earthquake Hazard and Risk
Hazard and risk are two fundamentally different concepts. In general, hazard is a phenomenon that has potential to cause harm. Phenomena are both natural and manmade. For example, earthquakes, hurricanes, tornadoes, and floods are natural hazards; whereas car crashes, chemical spills, train derailments, and terror attacks are manmade hazards. Risk, on the other hand, is the probability (chance) of harm if someone or something that is vulnerable is exposed to a hazard. In quantitative terms, hazard is defined by three parameters: a level of hazard (severity), its occurrence frequency, and location. An example might be a fatal car crash (severity) in every month (frequency) at a specific intersection. Risk is defined by four parameters: a probability, level of severity, time period, and location. For example, in health sciences, risk is defined as the likelihood (probability) of getting cancer (severity) if an average daily dose of a hazardous substance is taken over a 70-year lifetime. In the financial world, risk can be the probability of losing a certain amount of money (severity) over a period. The figure below shows the 23 core global risks estimated by the World Economic Forum (2007) over a 10-year exposure time. This figure also shows that risks are more useful for policy decision-making than hazard.
Seismic hazard is assessed from instrumental, historical, and geological records (or observations) and expressed in terms of a level of hazard and its occurrence frequency: a seismic hazard curve, at a site. The figure below shows hazard curves in terms of earthquake magnitude (a) and peak ground acceleration (b) for San Francisco, Calif. Earth scientists, seismologists in particular, play a key role in seismic hazard assessment.
Seismic risk depends not only on seismic hazard and exposure, but also on the models used to describe the occurrence of earthquakes. High seismic hazard does not necessarily mean high seismic risk, and vice versa. For example, seismic hazard is high in the California desert, but seismic risk is low because of few exposures (people or buildings). On the other hand, seismic risk could be high in some areas, such as Pakistan and Iran, because of high exposure, even though the hazard may be moderate. The common model used to describe earthquake occurrences is the Poisson distribution (independent of the history of previous earthquakes; time-independent). Time-dependent earthquake occurrence models have also been used in seismic risk analysis. Different models result in different risk estimates.
The figure below compares seismic hazard and risk posed by a magnitude 7.7 earthquake in the New Madrid Seismic Zone with hazard and risk posed by Hurricane Katrina in the Gulf Coast area.
The hazards posed by earthquakes in the New Madrid Seismic Zone can be expressed as a magnitude 7.7 earthquake, a level of ground motion (i.e., peak ground acceleration), or a modified Mercalli intensity at a site, or as total damage. Similarly, the hazards posed by Hurricane Katrina in the Gulf Coast area can be quantified as a Category V hurricane, a level of flood or wind speed at a site, or total damage. The geologic record suggests that the recurrence interval of earthquakes with magnitude of about 7.7 in the New Madrid Seismic Zone is about 500 years, and the historical record indicates the recurrence interval of Category 5 hurricanes is about 100 years in the Gulf Coast area. If the occurrences of earthquakes and hurricanes both follow a Poisson distribution, the risk posed by earthquakes in the New Madrid Seismic Zone can be estimated to be about a 10 percent probability of exceedance in 50 years for a magnitude 7.7 earthquake, a specific level of ground motion or MMI at a site, or total losses (life and money); the risk posed by Hurricane Katrina in the Gulf Coast area can be estimated to be about a 39 percent probability of exceedance in 50 years for a Category 5 hurricane, a specific level of flood or wind speed at a site, or total losses (life and money). Although total damage from a magnitude 7.7 earthquake is similar to that from a Category 5 hurricane, the hurricane risk in the Gulf Coast area is about four times higher than the seismic risk in the New Madrid Seismic Zone. Thus, risk is more useful for policy decision-making.
Assessing seismic hazard and risk is difficult because of insufficient data, especially in the central and eastern United States, where the data on large earthquakes are very limited. This lack of data results in large uncertainties for the seismological parameters that are the basis for assessing seismic hazard and risk. Seismic hazard is commonly assessed either by probabilistic seismic hazard analysis or deterministic seismic hazard analysis. The fundamental difference between PSHA and DSHA is in how the uncertainties are treated: either implicitly (PSHA) or explicitly (DSHA). Although PSHA has been more widely used, our studies show that it may not be appropriate because it has some intrinsic drawbacks (see Current Research for more detailed information).
Most damage during an earthquake is caused by ground motion (see Fig. 1). A commonly measured ground motion is peak ground acceleration (PGA), which is expressed as a percentage of the acceleration of gravity (g). The larger an earthquake's magnitude, the stronger the ground motion it generates. The level of ground motion at a site depends on its distance from the epicenter—the closer a site is to the epicenter, the stronger the ground motion, and vice versa. Ground motion from a major earthquake in the New Madrid Seismic Zone is expected to be much stronger in western Kentucky than in the central and eastern parts of the state. Strong ground motion could also induce secondary hazards such as ground-motion amplification, liquefaction, and landslides under certain conditions.
The local geology and soil also play important roles in earthquake damage. Soft soils overlying hard bedrock tend to amplify ground motions —this is known as ground-motion amplification(see Fig. 1). Amplified ground motion can cause damage, even to sites far from the epicenter. Ground-motion amplification contributed to the heavy damage in Mexico City during the 1985 earthquake (see Fig. 2) and in the Marina District of San Francisco during the 1989 Loma Prieta earthquake; both areas were more than 100 kilometers away from the epicenters. Figure 2 shows the total collapse of Juarez Hospital in Mexico City, caused by the amplified ground motion during the earthquake of Sept. 19, 1985. Many communities in Kentucky are set on soft soils, especially those along the Ohio and Mississippi River Valleys. Those communities are prone to ground-motion amplification. For example, most of the damage in Maysville during the Sharpsburg earthquake of July 27, 1980, was caused by amplified ground motion.
Soft sandy soils can be liquefied by strong ground motion —a process called liquefaction. Liquefaction can cause foundations to fail. Figure 3 shows that sandy soil was liquefied and behaved like fluid during the Nisqually, Wash., earthquake of Feb. 28, 2001. Communities in Kentucky set on soft soils, especially those along the Ohio and Mississippi River Valleys, may be prone to liquefaction.
Strong ground motion can also trigger landslides —known as earthquake-induced landslides —in areas with steep slopes, such as eastern Kentucky. The slope failure shown in Figure 4 was caused by the Nisqually earthquake of Feb. 28, 2001.