S U M M A R YHere, we evaluate the improvement in noise correlation functions (NCFs) gained by dividing ambient seismic records into shorter, overlapping time windows before correlation and stacking (Welch's method). We compare waveform convergence of short duration NCF stacks (e.g. 2, 5, 15 and 50 d stacks) towards the long-term (365 d) NCF stack. We observe short duration NCF improvement when applying Welch's method for non-pre-processed and running normalized time-series and short duration NCF degradation when applied to a 'one-bit' normalized timeseries. Surprisingly, non-pre-processed time-series provides the quickest convergence to a robust (year-long) NCF. Because of the simplicity of Welch's method, the improved NCF convergence and a minimal increase in computation, we recommend applying Welch's method for future ambient seismic field analyses. Using this approach will likely improve future NCF analyses, particularly for studies with limited duration recordings, high levels of intermittent local or site noise and studies attempting to evaluate temporal variations in subsurface structure.

No abstract

[1] The ambient noise correlation function (NCF) calculated between seismic stations contains, under appropriate conditions, accurate travel time information. However, NCF amplitudes are highly debated due to noise source intensity and distribution, seismic intrinsic attenuation, scattering, and elastic path effects such as focusing and defocusing. We prove with various numerical simulations that the NCFs calculated for a uniformly dispersive medium using the coherency method preserve accurate geometrical spreading and attenuation decay. We show that for a wide range of noise source distributions, the coherency of the noise correlation functions matches a Bessel function decaying exponentially with a specific attenuation coefficient. Conditions needed to obtain these results include averaging over long enough time intervals, a uniformly distributed seismic network, and a good distribution of far-field noise sources. We also show that the estimated attenuation coefficient corresponds to the interstation and not the noise-source-to-receiver structure.

We evaluate Rayleigh wave group velocity dispersion (5-40 s) around the Yellowstone Volcano Field with ambient noise tomography, measured from vertical component noise correlation functions. We include broadband data from 239 seismic stations (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012), including USArray's Transportable Array and the Noise Observatory for Imaging the Subsurface beneath Yellowstone (NOISY). Short-period (<13 s) group velocity anomalies are imaged for the Bighorn Basin (~25% slow) and Range (~20% fast), and the Yellowstone Plateau (~10% fast). Beneath the Yellowstone caldera, Rayleigh wave group velocities are~25% slower than the regional average with slow anomalies (<À15%) observed from 5 to 24 s. These values are consistent with a magmatic body being heated from below by an underlying plume.

If shaking from a local or regional earthquake in the San Francisco Bay region were to rupture levees in the Sacramento/San Joaquin Delta, then brackish water from San Francisco Bay would contaminate the water in the Delta: the source of freshwater for about half of California. As a prelude to a full shear‐wave velocity model that can be used in computer simulations and further seismic hazard analysis, we report on the use of ambient noise tomography to build a fundamental mode, Rayleigh wave group velocity model for the region around the Sacramento/San Joaquin Delta in the western Central Valley, California. Recordings from the vertical component of about 31 stations were processed to compute the spatial distribution of Rayleigh wave group velocities. Complex coherency between pairs of stations was stacked over 8 months to more than a year. Dispersion curves were determined from 4 to about 18 s. We calculated average group velocities for each period and inverted for deviations from the average for a matrix of cells that covered the study area. Smoothing using the first difference is applied. Cells of the model were about 5.6 km in either dimension. Checkerboard tests of resolution, which are dependent on station density, suggest that the resolving ability of the array is reasonably good within the middle of the array with resolution between 0.2 and 0.4°. Overall, low velocities in the middle of each image reflect the deeper sedimentary syncline in the Central Valley. In detail, the model shows several centers of low velocity that may be associated with gross geologic features such as faulting along the western margin of the Central Valley, oil and gas reservoirs, and large crosscutting features like the Stockton arch. At shorter periods around 5.5 s, the model's western boundary between low and high velocities closely follows regional fault geometry and the edge of a residual isostatic gravity low. In the eastern part of the valley, the boundaries of the low‐velocity zone and gravity anomaly are better aligned at longer periods (around 10.5 s) suggesting that the eastern edge of the gravity low is associated with deeper structure. There is a strong correspondence between a low in gravity near the Kirby Hills fault and low velocities from the ambient noise tomography. At longer periods, higher velocities creep in from the east and narrow the overall dimension defined by the lower velocities. Overall, there is a strong correspondence between the shape and location of low velocities in the Rayleigh wave velocity images, and geological and geophysical features.