Introduction

The focal mechanism is based on waveform modeling of the broadband time histories at SLM and CCM. Because of the small event size, time histories are band pass filtered to remove low frequency noise.

Because the SLM waveform at 20.8 km and the CCM waveform at 88.6 km were simple, a simple 40 km thick crust was initially used. Comparing observed traces to synthetics Green's functions for depths of 5, 10, 15 and 20 km indicated a depth > 15 km because of the lack of any observed sP arrival between P and S. However, to obtain the correct partition of the SV pulse onto the vertical and radial components, a shallow low velocity layer, approximating the Paleozoic section was added to create the following model used by hspec96:

H     Vp    Vs    Rho  Qp     Qs     EtaP  EtaS  FrefP FrefS
00.5  5.00  3.00  2.6 1000.0 1000.0  0.00  0.00  1.0  1.0
39.0  6.15  3.55  2.8 1000.0 1000.0  0.00  0.00  1.0  1.0
00.0  8.15  4.67  3.3 1000.0 1000.0  0.00  0.00  1.0  1.0

Because of difficulty in working at high frequencies, a visual search technique was used rather than a sophisticated inversion technique. A shell script was written to search through a range of strike, dip and rake. Then for each focal mechanism a synthetic three-component time history was created for comparison. Both synthetic and observed time histories were bandpass filtered the same way using SAC. An acceptable mechanism was based on matching the low P-wave amplitudes and the relative partition of amplitudes among the three components.

Figure 1 shows the observed and synthetic vertical,Z, radial, R, and transverse, T, traces for SLM. These ground velocities (cm/s) have been bandpass filtered between 0.25 and 4.0 Hz with 3 pole Butterworth filters.

SLM waveform fit

Figure 2 shows the observed and synthetic vertical,Z, radial, R, and transverse, T, traces for CCM, a distance of 86 km. These ground velocities (cm/s) have been bandpass filtered between 0.5 and 4.0 Hz with 3 pole Butterworth filters. These data were not used to define the focal mechanism, but rather to test the applicability of the mcehanism in matching the level of motion and also the relative amplitudes of the main phases. An exact fit was not attempted because of the complexity of the real data, and the fact that a prominent Moho S-reflection is seen, which requires a better crustal model. Prominent phases on the synthetics are P at 15 sec after the origin time, sP at 20s, S at 26 sec and the Moho relfection at 30 sec. The observed Z and R components show evidence of the sP phase, which is a good depth indicator, and a very strong direct S wave on the Z and R components.

The fit at CCM is less than perfect, but the phase alignment and approximate amplitudes are reasonably correct.

SLM waveform fit

Mechanism

The double couple focal mechanism has DIP=35, STRIKE=-17 and RAKE=50 for one nodal plane and a seismic moment of 0.5E+20 dyne-cm (Mw=2.46). Figure 3 shows the mechanism. A lower hemisphere equal-area projection is used.

Focal mechanism

Fig. 3. (a) nodal planes and predicted P-wave first motion polarities, with symbol size indicating relative amplitudes; (b) S-wave polarizations, with pressure and tension axes indicated; (c) observed P-wave first motions indicated (TYS AZM=271 AIN=151 compression; CCMO AZ=353 AIN=132 Compression; FVM AZM=179 AIN=66 compression). Nodal planes are constrained by the SLM modeling.
Last Changed January 16, 1998