On the nonuniqueness of receiver function inversions

TLDR

The study investigates the resolving power of teleseismic P‑wave receiver functions by modeling synthetic waveforms with a time‑domain inversion starting from a range of initial velocity models. The authors accelerate the inversion using Randall’s efficient differential seismogram algorithm, impose smoothness constraints via Shaw and Orcutt’s jumping technique, and apply the method to station RSCP in Tennessee. Over 235 inversions show that receiver functions are primarily sensitive to high‑wavenumber velocity changes and a depth‑velocity product, that the typical slowness range cannot resolve the depth‑velocity ambiguity without a priori information, and that the results agree with Owens et al.

Abstract

To study the resolving power of teleseismic P waveforms for receiver structure, we model synthetic waveforms using a time domain waveform inversion scheme beginning with a range of initial models to estimate the range of acceptable velocity structures. To speed up the waveform inversions, we implement Randall's (1989) efficient algorithms for calculating differential seismograms and include a smoothness constraint on all the resulting velocity models utilizing the “jumping” inversion technique of Shaw and Orcutt (1985). We present the results of more than 235 waveform inversions for one‐dimensional velocity structures that indicate that the primary sensitivity of a receiver function is to high wavenumber velocity changes, and a depth‐velocity product, not simply velocity. The range of slownesses in a typical receiver function study does not appear to be broad enough to remove the depth‐velocity ambiguity; the inclusion of a priori information is necessary. We also present inversion results for station RSCP, located in the Cumberland Plateau, Tennessee. Our results are similar to those from a previous study by Owens et al. (1984) and demonstrate the uncertainties in the resulting velocity estimate more clearly.