The Ataköy vertical array (Turkey): insights into seismic wave propagation in the shallow-most crustal layers by waveform deconvolution

Abstract

S U M M A R YA vertical array of accelerometers was installed in Ataköy (western Istanbul) with the longterm aim of improving our understanding of in situ soil behaviour, to assess the modelling and parametric uncertainties associated with the employed methodologies for strong-motion site-response analysis, and for shallow geological investigations. Geotechnical and geophysical investigations were carried out to define the subsoil structure at the selected site. Data associated with 10 earthquakes (2.7 < M < 4.3) collected during the first months of operation of the array were used to image the upgoing and downgoing waves by deconvolution of waveforms recorded at different depths. Results have shown that the velocity of propagation of the imaged upgoing and downgoing waves in the borehole is consistent with that of S or P waves, depending on the component of ground acceleration analysed but independent of the chosen signal window. In particular, an excellent agreement was found between the observed upgoing and downgoing wave traveltimes and the ones calculated using a model derived by seismic noise analysis of array data. The presence of a smaller pulse on the waveforms obtained by deconvolution of the horizontal components suggests both internal S-wave reflection and S-to-P mode conversion, as well as a not normal incidence of the wavefield. The presence of a pulse propagating with S-wave velocity in the uppermost 25 m in the waveforms obtained by the deconvolution of the vertical components suggests P-to-S mode conversion. These evidences imply that, even when site amplification is mainly related to 1-D effects, the standard practice in engineering seismology of deconvolving the surface recording down to the bedrock using an approximate S-wave transfer function (generally valid for vertical incidence of SH waves) might lead to errors in the estimation of the input ground motion required in engineering calculations. Finally, downgoing waves with significant amplitudes were found down to 70 m and even to 140 m depth. This result provides a warning about the use of shallow borehole recordings as input for the numerical simulation of ground motion and for the derivation of ground motion prediction relationships.

Seismic‐Wave Propagation in Shallow Layers at the GONAF‐Tuzla Site, Istanbul, Turkey

“…As in previous deconvolution studies (Mehta et al, 2007;Parolai et al, 2009), our results were insensitive to the data window selection. Accordingly, we chose to use windows containing the complete signal of the event.…”

“…This approach also allowed us to determine the velocity structure between the borehole sensors. Several algorithms exist for interferometry, but here we focus on the deconvolution technique (e.g., Trampert et al, 1993;van Vossen et al, 2004;Mehta et al, 2007;Parolai et al, 2009;Nakata and Snieder, 2012).…”

“…We tested two different regularizations. The first is used, for example, by Mehta et al (2007), Parolai et al (2009), andNakata andSnieder (2012), and the second is the method of Helmberger and Wiggins (1971) and Dey-Sarkar and Wiggins (1976). The first turned out to be more stable for our data and is given by E Q -T A R G E T ; t e m p : i n t r a l i n k -; d f 3 ; 3 1 3 ; 2 1 0 G 1;2 ω ≈ Wω Uz 1 ; ω Uz 2 ; ω ;…”

“…It is applied, for example, for imaging purposes (Vasconcelos and Snieder, 2008b), retrieving building response to ground shaking (Snieder and Şafak, 2006;Bindi et al, 2015), and for determining near-surface velocities (van Vossen et al, 2004;Parolai et al, 2009;Nakata and Snieder, 2012). Deconvolution interferometry has also found applications in estimating site-specific Q values.…”

“…The regularization parameter ε refers to a constant added to the denominator to prevent numerical instabilities in equation (2) (e.g., Parolai et al, 2009). It is chosen as a percentage of the average spectral power of Uω at the station selected as reference.…”