Using seismic noise correlation to determine the shallow velocity structure of the Seattle basin, Washington

Abstract

Cross-correlation waveforms of seismic noise in the Seattle basin, Washington, were analyzed to determine the group velocities of surface waves and constrain the shear-wave velocity ( V S ) for depths less than about 2 kilometers (km). Twenty broadband seismometers were deployed for about 3 weeks in three dense arrays separated by about 5 km, with minimum intra-array station spacing of about 0.5 km. Cross correlations of only 9 days of noise recordings produced Green’s functions at periods of 2 to 6 seconds (s) for sites about 5 km apart. Usable noise correlations for shorter periods of 0.5 to 1.0 s were found for sites within the arrays separated by 1 to 2 km. We bandpass filtered the inter- and intra-array cross-correlation waveforms to determine Love-wave group velocities at periods of 0.5 to 6 s for paths within the Seattle basin and at 3 to 5 s for paths crossing the southern edge of the basin. We developed a non-linear inversion program to determine V S profiles that fit the observed group velocities for paths in the basin. We found that these group velocities are well fit by a variety of V S profiles, each with a distinct jump in V S at depths ranging from 0.9 to 1.3 km. This jump in V S is inferred to represent the top of bedrock. The observed group velocities are not matched by models with the top of bedrock at 0.7-km depth or shallower. The group velocities are also fit by a model with no large jumps in V S in depths less than 2.4 km. The V S profile for the middle of the basin from Stephenson and others (2017), with a depth to bedrock of 0.9 km, also adequately fits the group velocity observations, if a velocity gradient is added from 0.05- to 0.1-km depth. The results indicate that short (3-week) deployments of seismometers to record seismic noise may provide useful constraints on the V S of sedimentary basins.