Nearly two-thirds of the microseismicity occurring at Parkfield, California, consists of repeating clusters of closely located events exhibiting waveform cross-correlation values exceeding 0.95. This observation, coupled with a high probability for an M ≈ 6 earthquake within the next decade (Working Group on California Earthquake Probabilities, 1988), make this segment of the San Andreas fault system a prime location to search for temporal variations in coda Q associated with crustal changes that may occur during the earthquake cycle. The coda comparison analysis method described in Part 1 of this study (Aster et al., 1996) is applied to data from four borehole stations of the Parkfield High Resolution Seismic Network (HRSN) for the period January 1987 through February 1994. We examine 21 of the nearly 300 repeating clusters observed to date. The restriction of the data set to only the most highly similar sources is observed to be critical; small differences in the source process, even for event pairs with centroid locations within approximately 20 m, significantly increase the spurious noise level in the parameter estimates. Absolute 68% confidence bound estimates obtained on the repeatability of coda Q for all microearthquake pairs are ±10% for frequencies below 20 Hz. Tighter constraints are obtainable by reconstructing the temporal history of coda Q from the first difference measurements between microearthquake pairs. These functional reconstructions show that the mean value of coda Q has not varied by more than approximately 5% from 3 to 30 Hz at the 1σ level during the study period. However, other observables, such as an increased level of microseismicity, suggest that a preparatory stage for the next Parkfield earthquake has begun, indicating that coda Q is so far showing no sign of sensitivity to preparatory processes at Parkfield.

First Page Preview

First page PDF preview
You do not currently have access to this article.