Computation of Three Component Receiver Functions (R3) by Beamforming of Optimal Regional Vertical Component Seismograms Before Simultaneous Iterative Deconvolution
Abstract
In receiver function (RF) analysis, we interpret Ps conversions and reverberations from broadband seismograms of teleseismic events (from 30 to 90 degrees away). To produce RFs, the horizontal component of the seismogram is deconvolved by the vertical (which, because of the near vertical ray paths, are assumed to be the P-source function). Noise in the vertical component (i.e. scattering from localized 3-D structure, s-waves energy, ambient noise, etc.) will result in a noisy horizontal RF with possible spurious peaks. Many investigators produce RFs by rotating the seismograms into the presumed or derived ray path of the P-wave (L-component) prior to deconvolution. Our goal is to make a cleaner RF by way of beamforming to produce the cleanest P-waveform from the seismogram before deconvolution. Rather than using a reference model to compute delay time for beamforming, cross-correlation will be used to compute delay times of the P-wave across stations. A regional wave-front will be fit to the time delays to estimate a regionally corrected backazimuth (BAZ) and ray parameter. This data derived BAZ and ray parameter will be used to rotate the seismograms into the P-ray path (L-component). A cleaner estimate of the incoming P-waveform can be made by beamforming the vertical component of the seismogram from a given station with those recorded at neighboring stations (from upwards of three hundred km away). The time delays derived from cross-correlation will be used to align P-wave recordings from stations as much as 300 km away. This beamed vertical component is then averaged from stations across different geological terrains. As a result, local variation in delay times of P reverberations will not be coherent across stations and stacking (beaming) will remove these P-reverberations from the beamed P component (assumed source function). Further improvement can be made by producing a correlation matrix for all shifted P-waveforms from each station and excluding those with low correlation coefficients from the beam. Because we have removed (or minimized) the P-reverberations from the source function (L component), it becomes necessary to rotate all three components to be parallel with the P-wave ray parameter to minimize the amplitudes of P-reverberations on the SV and SH components. We then deconvolve all three components of the rotated local seismogram by the beamed vertical component and produce a three component receiver function (RF3). The vertical component of the RF3 will contain local P-reverberations, thereby contributing P-arrival times that are independent of the Vp/Vs ratio. When combined with radial (Ps) components of the RF3, this can reduce the ambiguity in trade-off between estimates of depth and Vp/Vs ratio to a given horizon. This method of computing RF3s will be tested in both areas of complex geology (southern California) and relatively simple geology (midcontinent USA). Comparison of RF3s with traditional receiver functions may yield better interpretations than using either method independently.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2011
- Bibcode:
- 2011AGUFM.S23A2239P
- Keywords:
-
- 7205 SEISMOLOGY / Continental crust;
- 7218 SEISMOLOGY / Lithosphere;
- 7294 SEISMOLOGY / Seismic instruments and networks