Scenarios Earthquakes for Urban Areas Along the Atlantic Seaboard of the United States: Seismicity

Earthquakes with magnitudes as large as those in California (M_w > 8.0) have occurred in the U.S. east of the rockies, albeit at a low rate and dispersed over a large area.

Using a seismicity catalog NCEER 1991 (Seeber and Armbruster, 1991) for the entire eastern U.S. (Figure 1), an area defined here as the region east of the Rocky Mountains, we can attempt to fit the reported seismicity to a ‘Gutenberg- Richter’ frequency- magnitude relation of the form

In this equation, N is the expected cumulative annual number of earthquakes (as an average for the entire eastern U.S.) with magnitudes M. The average recurrence period, for events with magnitudes M, is then T = 1/N. We define the probability P(%) that during an exposure time, Dt, an event with magnitude M will occur, corresponding to an average recurrence period T(M). This probability is:

Using numbers derived from the eastern U.S. earthquake catalog, we compute estimates of the average recurrence periods and of occurrence probabilities given in Table 1. There are large uncertainties to these estimates which we do not discuss here.

Magnitude M	Rec. Period T (years)	Probability P(%) event M will occur^a in time t (years)
Magnitude M	Rec. Period T (years)	t=10	t=20	t=30	t=50	t=100	t=1000
5	3	98.0	100.0	100.0	100.0	100.0	100.0
6	22	37.0	61.0	75.0	90.0	99.0	100.0
7	180	5.5	10.0	16.0	24.0	43.0	100.0
8	1470	0.7	1.4	2.0	3.3	6.6	49.0
a. Probabilites are for the occurrence of events in the entire U.S. region east of the Rocky Mountains without distinction exactly where the events would occur.

It is noteworthy that the computed average recurrence periods T for the larger magnitudes (M 7) appear long compared to the short times lapsed since such events have last occurred historically. For example, the M 7 Charleston SC earthquake of 1886 occurred only 106 years ago, while the average T for an M 7 event to occur anywhere in the eastern U.S. based on the overall seismicity is computed to be about 180 years; the M 8 New Madrid earthquakes of 1811/12 occurred only about 180 years ago, while the average recurrence period for such an event to occur anywhere in the eastern U.S. is estimated to be T = 1,470 years. From this comparison we conclude that the estimates for recurrence periods for earthquakes with magnitudes M 7 given in Table 1 may be upper bound values. Actual recurrence periods for these large magnitude events could be shorter than indicated in Table 1, which is based on a fit of equation (1) to a limited data set under the assumption of a constant b value over the entire magnitude range.

Example: NY Metropolitan Area. The average estimates given in Table 1 are for the entire U.S. east of the rockies. But seismicity is not uniformly distributed. Also, the urbanized areas occupy only a fraction of this vast region. Therefore, it is of interest to ask the question: how close to a given city may an earthquake of a given magnitude occur during a given period of exposure? As an example, we use the seismicity in the New York City metropolitan region, in a seismic zone known as "Manhattan Prong." The seismicity of the Manhattan Prong (including portions of the adjacent Newark Basin and Hudson Highlands) is fairly well documented (Figure 2). In this region, Lamont-Doherty Earth Observatory has operated a telemetered seismic network for more than 20 years. The historic seismicity (300 yr) is taken from the NCEER 1991 catalog (Seeber and Armbruster, 1991). We use the combined network and historical seismicity data. The specific source area considered has a dimension of 50 x 130 km (source area S = 6,500 km²). The cumulative annual number of earthquakes that occurred as a function of magnitudes equal to and larger than M, normalized to an area S = 6,500 km², is displayed in Figure 3. The surrounding area has an only slightly reduced level of seismicity.

A straight-line fit is made to the observed seismicity rates by the Gutenberg-Richter frequency-magnitude relation

with a = -2.305 ?0.642 and b = 0.775 ? 0.153. Note that n and 10^a are now normalized to number of events per year per km². Instead of relation (3), a non-normalized form (4) can be derived which yields the expected cumulative number of earthquakes, N, during T (years) in the area S (km²) as

A specific source area S'(km²) = pr² is needed to produce one event, i.e. N = 1, of magnitude M or larger, in a period of T (years):

We now ask at which average distance d would occur an event M from a point that floats in the open-ended region characterized by seismicity parameters a and b. Expressing the total area S' in terms of a median-probability (50-percentile) distance d, one obtains S' = 2pd² (or

). Inserted into equation (5) this yields:

Equation (6) yields for any given recurrence period T a unique set of magnitude-distance combinations (M-d).

Using equation (6) with the numerical values for a and b applicable to the Manhattan Prong, we obtain:

Inserting the recurrence periods of T = 500, 1000, 2500, and 5000 years, respectively, into (7), and choosing the fixed magnitude values M = 5, 6, 7 we obtain the following hazard-consistent distances d (km) from a central location within the Manhattan Prong seismic source shown in Table 2.

Magnitude M	Average Recurrence Period T (years)
Magnitude M	500	1000	2500	5000
5	22	16	10	7
6	54	38	24	17
7	131	92	58	42

We see that the median distances d increase with magnitude, and decrease with average recurrence time. The longer one waits, the closer a large earthquake may strike.

Scenario Earthquakes for Urban Areas Along the Atlantic Seaboard of the United States

Seismicity