Finite-Source Modeling to Constrain Complex Fault Geometry of the South Napa Earthquake

Abstract:

The August 24, 2014 South Napa M_w6.0 earthquake ruptured to the north along the West Napa fault, producing strong shaking in the deep sediments (~1000 m) of the Napa Valley. Aftershock locations define the roughly north-northwest striking fault plane, but details regarding the dip of the fault and the strike of the northernmost section of the fault remain less certain. Preliminary inversions based on the inversion of broadband data from the BDSN, PBO GPS, and InSAR observations show a 13 km long rupture initiating at a depth of 11 km and propagating unilaterally to the northwest and up-dip (Dreger et al., 2015), however this kinematic finite-source model assumes a single planar fault. Field reconnaissance and LiDAR imaging reveals a complex system of sub-parallel faults at the surface. The main western branch, which had the largest surface offsets, shows a marked change in strike approximately 9 km north of the epicenter (Bray et al., 2014). In addition, there is evidence of surface offsets on eastern branches of the West Napa fault system (Bray et al., 2014). Whether this complexity persists, or how the faults may link at depth, is an open question. However, based on the limited preliminary dataset of Dreger et al (2015), we conclude that the fault dip is nearly vertical. In this study we investigate complex fault models that account for changing strike, and en echelon west dipping faults that may join at hypocentral depth by means of finite-source inversion of regional broadband, local strong-motion and GPS and InSAR geodetic data sets. We build on the data set used in the prelimary model of Dreger et al, (2015) by incorporating local strong motion waveform data as well as the static displacement dataset of Funning (2015). For the strong motion data, we incorporate stations located on “rock” (Vs30 > 700 m/s). Green’s functions for the “rock” stations are computed with a modified Gil7 model to account for the lower velocity in the upper 30 meters. Preliminary results from a kinked model in which the strike 10 km north of the epicenter rotates 20 degrees more northerly reveal a slip model that is highly consistent with the results of the Dreger model in terms of the directivity and area of slip, but the modified Gil7 model tends to produce higher peak slips around 120 cm.