Seismicity, Earthquakes Mechanisms, and Seismic Wave Attenuation in the Northeastern United States

The historical seismicity, defined here as covering the time period 1534-1975, was examined using two regionalization algorithms for the purpose of defining seismic zones. The frequency regionalization, which is a two-dimensional spatial filter applied to the earthquake catalog (with aftershocks removed), reveals three major seismic zones in the area. The first zone, termed the Western Quebec Seismic Zone, spans an area from Lake Champlain in VT to the PQ-ONT border. The mean return time for a magnitude 6.0 (mb) earthquake in this zone, using a least squares estimator, is 188 years with a 41% probability of occurrence in 100 years. For an mb 6.5 earthquake, the mean return time is 536 years with a 17% probability of occurrence in 100 years. The second zone, termed the Charlevoix Seismic Zone, is a concentrated area of activity which has experienced some of the largest earthquakes in the study area. Here, the mean return time is 231 years with a 35% probability of occurrence in 100 years. The third major zone is termed the Boston-NH Seismic Zone, a band of seismicity running from the Lakes Region of central NH to eastern MA. Here, the mean return time of an mb 6.0 earthquake is 408 years with a 39% probability of occurrence in 200 years. For a magnitude 6.5 event, the mean return time is 1060 years with a 17% probability of occurrence in 200 years. The energy regionalization of the area shows that most of the seismic energy release has taken place in four small areas: Timiskaming, ONT; Cornwall, ONT-Massena NY; Charlevoix, PQ; and Cape Ann, MA. These zones are not surrounded by regions of lower seismic energy release. This may mean that the physical processes responsible for the events may be very small in spatial extent.

The instrumental dataset, covering the time period October 1975 through September 1981, shows that in most (but not all) cases the distribution of seismicity is space stationary, i.e. instrumental epicenters cluster in areas of historically active seismicity. Earthquake locations computed from network data are in most cases accurate to within 5 km. In some areas, in particular the Cape Ann area, the occurrence of earthquakes is much lower than in the past. Focal depths are known for only a handful of events which have occurred near seismic stations or have been studied with aftershock surveys. West of the Appalachians, in the Grenville Province, earthquakes occur at depths ranging form the near surface to almost 20 km. In the Appalachian Province, earthquakes are confined to the upper 10 km of the crust.

Fault plane solutions were determined for ten earthquakes in the study area using P-wave first motion data and crustal models applicable to the source areas. In addition, a literature search was undertaken and a dataset compiled which includes 53 earthquake fault plane solutions and 18 non-seismic stress measurements (hydrofracturing, overcoring, fault slip and core offsets, and pop-ups). This dataset was used to produce a crustal stress map for the NEUS-SEC. The area is characterized by a horizontal compressive stress field; however, the direction of this stress field is not uniform across the entire study area. In the Grenville Province, the compressive stress field is highly uniform and trends in an ENE-WSW direction. Earthquakes in this area show primarily thrust faulting on NW-SE trending fault planes. However, in the Appalachian Province, the compressive stress field is highly non-uniform. Earthquakes in this area show both thrust and strike-slip motions. If we interpret the dataset for the largest, best constrained events, the scatter remains. There may be an underlying compressive stress field in this area, but it may be modified by crustal inhomogeneities, such as the presence of crustal blocks, or by topographic loading stresses.

Seismic wave attenuation was measured in the study area from the time decay of coda wave amplitudes on narrow bandpass filtered seismograms. The frequency band of interest was 0.75 to 10 Hz. Qc was found to increase with frequency across this band, but there was also a difference between this frequency dependence for short and long lapse times of coda wave propagation. For short lapse times, corresponding to wavepaths primarily in the upper crust, Q increases from 400 at 3 Hz to 1300 at 10 Hz. For long lapse times, corresponding to wavepaths in the lower crust and upper mantle, Q was found to vary from 660 at 1 Hz to 1500 at 10 Hz. If we interpret this dataset in terms of a model incorporating both scattering and anelastic attenuation, we find that the minimum mean free path in the crust is about 75 km over all frequencies, whereas in the mantle, the minimum mean free path decreases from 400 km at 0.75 Hz to 90 km at 10 Hz.

These Q measurements were then used to develop and test a ground motion attenuation model for New England. We began by taking an intensity attenuation model and converting it to an equivalent particle velocity attenuation model using a velocity-intensity correlation. The resulting model successfully predicts the peak horizontal velocities observed the 19 January 1982 Gaza, NH earthquake. The model also compares favorably with the theoretical seismic wave attenuation assuming Lg propagation and the Q values measured in this work. These models were then used to compute the ground motions for four hypothetical NEUS-SEC earthquakes.

In summary, the seismic characteristics of the Grenville and Appalachian Provinces were found to be quite different. In the Appalachian Province, earthquake epicenters scatter across broad areas, are shallower, and exhibit more varying focal mechanisms than in the Grenville. The attenuation and scattering of seismic waves is also greater in the Appalachians. Potential ground motions may be much less predictable in this province.