AbsoluteS-velocity estimation from receiver functions

Abstract

S U M M A R YWe present a novel method to recover absolute S velocities from receiver functions.For a homogeneous half-space the S velocity can be calculated from the horizontal slowness and the angle of surface particle motion for an incident P wave. Generally, the calculated S velocity is an apparent half-space value which depends on model inhomogeneity and P-waveform. We therefore, suggest to calculate such apparent half-space S velocities from low-pass filtered (smoothed) receiver functions using a suite of filter-parameters, T. The use of receiver functions neutralize the influence of the P-waveform, and the successive low-pass filterings emphasize the variation of S velocity with depth.We apply this V S,app. (T ) technique to teleseismic data from three stations: FUR, BFO and SUM, situated on thick sediments, bedrock and the Greenland ice cap, respectively. The observed V S,app. (T ) curves indicate the absolute S velocities from the near surface to the uppermost mantle beneath each station, clearly revealing the different geological environments. Application of linearized, iterative inversion quantify these observations into V S (z) models, practically independent of the S-velocity starting model. The obtained models show high consistency with independent geoscientific results. These cases provide also a general validation of the V S,app. (T ) method.We propose the computation of V S,app. (T ) curves for individual three-component broad-band stations, both for direct indication of the S velocities and for inverse modelling.In applied receiver function analysis it is often stated that receiver functions are not sensitive to the absolute levels of the S velocity (e.g. Ammon et al. 1990;Kind et al. 1995;Schlindwein 2006;Tomlinson et al. 2006). In the present paper, we show that this is not entirely correct. We present a simple transform which clearly emphasizes the absolute S-velocity information present in receiver functions. The effect of the free surface on an incoming teleseismic P wave plays a key role in this method.A plane P wave incident on a free surface is reflected as a P wave and a converted SV wave. The particle motion (or polarization) observed by a three-component seismograph on this surface is the superposition of the incoming and the two outgoing waves. As a result the apparent incidence angle (i P ) defined by the surface particle motion is different from the true P wave incidence angle (i P ). The relation between true and apparent incidence angles was early quantified in Wiechert (1907, eq. 128) (see also e.g. Nuttli & Whitmore 1961) from which followswhere V P and V S are compressional and shear velocities of the half-space and p the horizontal slowness (ray parameter) of the P wave. Eq. (1) can be rearranged towhich defines the half-space S velocity as a function of the observed apparent incidence angle and slowness of a P wave. No assumptions are made concerning V P or V P /V S . Bostock & Rondenay (1999) derived an equivalent but less simple expression (their eq. A6) from the free-surf...

Crustal S-Wave Velocity from Apparent Incidence Angles: A Case Study in Preparation for InSight

“…Due to the variable station quality, this resulted in similar numbers of individual usable receiver functions per station (Table 1). Data from station BFO have already been used as an example by Svenningsen and Jacobsen (2007) in their study of lithospheric S-wave velocities derived from P-wave polarization measured by receiver functions.…”

“…A set of second-order zero-phase Butterworth low-pass filters is applied to the receiver functions to obtain the variation of apparent incidence angles with period (Hannemann et al 2016). Following Svenningsen and Jacobsen (2007), the corner periods of the filters T f are selected to be logarithmically distributed. When directly comparing results with Svenningsen and Jacobsen (2007), it has to be taken into account that, for an equivalent filter band, the corner period of the Butterworth filters used here is twice the corner period of the cosine filters used by Svenningsen and Jacobsen (2007).…”

“…The S-wave velocities obtained at various periods are then apparent rather than true sub-surface velocities, with more and more information from larger depths contributing at increasingly longer periods. Those apparent S-wave velocities can be inverted for the true S-wave velocity structure of the subsurface (Svenningsen and Jacobsen 2007).…”

“…Signals from nuclear explosions, which have comparatively simple source-time functions and well-known source properties, have been used in the study of P-wave polarization to avoid these complexities (Krüger 1994). More recently, Svenningsen and Jacobsen (2007) have suggested to use P-wave receiver functions instead of the raw seismograms to measure φ P . The deconvolution included in receiver function processing effectively removes source and distant wave propagation effects, so this approach allows using a larger amount of data and a more automated processing.…”

“…Svenningsen and Jacobsen (2007) showed that the linearized inversion of apparent S-wave velocity curves calculated from P-wave polarization for crustal and upper-mantle S-wave velocities is independent of the starting model, in contrast to the direct inversion of the receiver function waveforms (Ammon et al 1990). After adjusting the method to the boundary conditions at the ocean bottom, Hannemann et al (2016) successfully applied it to an OBS data set with one to five receiver functions per station, thus demonstrating its usefulness when only a small number of event recordings is available.…”