Seismicity Patterns of the September 28th Parkfield Event

The densely instrumented Parkfield section of the San Andreas fault provides an outstanding data set to investigate the frequency-magnitude distribution for the pre main shock and aftershock period of the September 28th, MW=6.0 Parkfield event. The Parkfield segment has been subject to extensive studies of the frequency-magnitude distribution and strong heterogeneities of b-values have been documented - anomalously low b-values of 0.5 in the asperity region below Middle Mountain and b>1.1 in the creeping section at shallower depth. Detailed investigations have also shown that large parts of the Parkfield segment have stationary b-values. In this study we address the questions of 1) whether these patterns prevail for the aftershock sequence of the latest Parkfield event and 2) whether the aftershocks occur in low-slip or increased shear-stress regions. Prior to studying the effects on b-value, we investigate the quality of the catalog as changes in seismicity parameters are closely related to natural or man made transients in the frequency-magnitude distribution. We focus on the assessment of the magnitude of completeness as a function of time. Detecting localized temporal changes of b-values in the presence of strong spatial heterogeneity is challenging, thus we map the b-value distribution as a function of sampling radius to obtain the most suitable resolution, compare the results for the time period before and after the main shock using differential b-value maps and establish significant changes using the Utsu test. With this approach, we strive to further establish a link between b-values and stress in the crust. In a second approach, we use a quantitative test for the spatial relationship between the aftershock distribution and the main shock rupture properties. Aftershock hypocenters are projected onto the main shock fault plane and coseismic slip and stress drop values obtained from finite source models are interpolated to their respective location. The null hypothesis is: aftershock hypocenters are randomly distributed on the main shock fault plane and are not correlated with main shock properties. Keeping the locations of the aftershocks, a series of slip distributions is simulated using a spatial random slip model and the respective coseismic stress-drop distribution. The number of aftershocks located in low-slip or increased shear-stress regions are determined to show whether or not an apparent correlation between main shock properties and aftershock locations exists. Preliminary results indicate that the null hypothesis cannot be rejected.