Resolving Stress Drop Variation Along San Andreas Fault at Parkfield and Its Implication

Date: 4/25/2019

Time: 03:45 PM

Room: Cascade I

Earthquake stress drop is an important source parameter that is related to high-frequency ground motion and source scaling. Estimation of stress drop can be subject to significant uncertainty, such as sampling rate, source complexity, noise level, etc. The most significant source of uncertainty is the tradeoff between source term and path effects, which can systematically bias the stress drop results. In this study, we try to improve the stress drop estimations with improved stacking method, and spatial binning in both along-strike and along-depth directions. We collect data from both surface and borehole networks from 1984 to 2016, then we perform spectral analysis using an improved stacking-based spectral decomposition method that solves an empirical-correction-spectrum (ECS) based on a group of selected events, as well as spectral ratio analysis based on empirical-green’s-function (EGF) with carefully selected event pairs to address stress drop uncertainties and help resolving the tradeoff.

Using borehole data, we find that the stacking-based analysis can be systematically biased by the ECS correction, which relates to in-situ attenuation correction. In addition, spatial distribution of events is the root for stress drop spatial trends (such as depth trend), which can be ascribed to spatial attenuation variability. To reduce the issue of attenuation, we constrain the stress drop of lowest magnitude bin to obtain ECS for different spatial grids. The results significantly reduce the standard deviation of stress drops for all events. The improved stress drop observation implies increased stress drop in the creeping zone (NW of Parkfield) and decreased stress drop in the locked zone (SE of Parkfield) after the 2004 M6 earthquake. Future study will include assessing whether the stress drop change is related to attenuation changes by applying time-dependent ECS correction; analysis using surface stations with lower sampling rates and likely higher noise level; and comparison between stacking and individual pair EGF analysis. Updated results will be presented.

Presenting Author: Jiewen Zhang

Authors