by P. Somerville, N. Smith and D. Dreger
December 1993, 84 pp.
Click on the link below for the full text:
A search has been made for the influence of critical reflections from the
lower crust on the attenuation of peak acceleration with distance in individual
California earthquakes. At close distances, the direct upgoing shear wave generates
the largest ground motion amplitude. However, shear waves reflected from the lower
crust become large in amplitude once the distance (and the angle of incidence) are
large enough to become critical, that is, when all of the incident energy is reflected
back to the surface. Strong ground motion amplitudes remained approximately uniform
in the distance range of about 50 to 80 km during the 1989 Loma Prieta earthquake,
and strong ground motion levels on rock and alluvium in the central San Francisco
Bay area exceeded those of empirical attenuation relations. It has been suggested
that these large ground motions were critical Moho reflections (SmS), and contributed
about equally with impedance contrast amplification effects in generating destructive
ground motions at soft soil sites in the central San Francisco Bay area.
The approach used in this study is to construct profiles of accelerograms having absolute
times for each of several major California earthquakes. Evidence for the presence of
critical reflections consists of large phases that arrive near the time predicted for
critical reflections from lower crustal layers such as the Conrad and Moho, and have a
horizontal phase velocity appropriate for critical reflections from the lower crust.
For selected events, the interpretation of these profiles of recorded accelerograms is
facilitated by synthetic seismograms that represent the effects of wave propagation in
horizontally layered models of the earths crust. The effect of these reflected phases
on the attenuation of peak acceleration with distance is then assessed.
We have analyzed profiles of accelerograms from seven large California earthquakes
and several aftershocks. For the older events, the strong motion recordings did not
have absolute time, and so it was difficult to identify critical reflections in these
recordings. Also, for the larger events, the source duration was sufficiently long that
it obscured the individual phases that we would like to identify. Nevertheless, we
generally found evidence of critical reflections in large, late wave arrivals that
may be SmS, and these arrivals were associated with a flattening of the empirical
attenuation curve. Analysis of aftershock recordings, which have briefer source
functions, provided clear evidence of the presence of these reflected phases.
These results suggest that critical reflections influence the attenuation of strong ground
motion throughout California, and are already included to an extent in standard attenuation
relations. However, the smoothly decaying functional form of most attenuation relations,
while fitting observed strong motion data when averaged over many events, may not
accurately describe the attenuation that is observed in a single event. The attenuation
relation of the Loma Prieta earthquake is the most prominent demonstration of this.
Since we do not expect crustal earthquakes in California to occur much deeper than 18
km, or the crustal thickness to be much less than 25 km, the Loma Prieta case may
approximately represent an upper bound on the departure of the attenuation relation
of an individual event from that of the larger strong motion data
set in California.
As a rule of thumb, the distance at which the attenuation curve for an individual
earthquake begins to flatten can be estimated from the critical distance of the SmS
phase, which is easily calculated from the crustal structure and the source depth.