Reflection Response from the Autocorrelation of Ambient Noise as Recorded on the Transportable Array
Thursday, 18 December 2014
In performing autocorrelation of ambient noise, the station acts as both a receiver and a source of seismic waves. This allows for the computation of the crustal reflection response. Two years of broadband data are processed in 10-hour segments from 25 Transportable Array (TA) stations in the central U.S. Multiple bandpass filters are evaluated, and one with corners between 1 and 2.5 Hz is used to remove the greatest range of noise without compromising visibility of the reflections. The power spectrum of the ambient noise (amplitude of the stacked autocorrelations) shows resonant peaks in the frequency domain presumably caused by harmonic sources near the station. These resonance peaks are a source of noise when interpreting the autocorrelations in the time domain, since they are not represented in the structure response. To improve the reflection response, the peaks are removed through a short-term average (sta) and long-term average (lta) algorithm applied to the amplitude spectrum. The data are migrated using a simple 1-D velocity model derived from Catchings (1999) to compute two-way travel times of hypothetical reflectors in the crust. The horizontal components are migrated using the S wave velocities and the vertical component using the P wave velocities to model the expected shear and P wave reflections respectively. Comparing images with the S and P waves reveal similarity in depth of crustal reflectors, which gives confidence that the observations are crustal discontinuities. The data for stations within the Mississippi Embayment reveal a discontinuity at or near the estimated sediment thickness at each station. Reflections in the data are present at the approximate depth of the Moho as obtained from the Earthscope Automated Receiver Survey for each station.