Investigations of Passive Seismic Body-Wave Interferometry Using Noise Auto-correlations for Crustal and Upper Mantle Structure

It is known that the positive lags of the auto-correlation for the seismic transmission response of a layered medium correspond to the reflection seismogram (Claerbout, 1968). In this study, we investigate the use of ambient seismic noise recorded at selected broadband USArray EarthScope Transportable Array (TA) stations to obtain effective reflection seismograms for frequencies up to 1 Hz. The goal is to determine the most suitable parameters used for the processing of ambient seismic noise for the identification of crustal and upper mantle reflections and to minimize unwanted artifacts in the noise correlations. In order to best retrieve the body-wave components of the Green's function beneath a station, a number of processing steps are required. We first remove the instrument response and apply a temporal normalization to remove the effects of the most energetic sources. Next we implement spectral whitening. We test several operators for the spectral whitening where the undulations of the power spectrum are related to the strengths of later arrivals in the auto-correlation. Different filters are then applied to the auto-correlation functions, including Gaussian and zero phase Butterworth filters, in order to reduce the effect of side lobes. Hourly auto-correlations are then stacked for up to one year. On the final stack, Automatic Gain Control (AGC) is applied to equalize the correlation amplitudes in the time domain. The robustness of the resulting ambient noise auto-correlation is first tested on selected TA stations in Nevada, where we are able to identify PmP and SmS arrivals similar to those found by Tibuleac and von Seggern (2012). We then investigate noise auto-correlations applied to selected USArray TA stations in the central US.