Joint inversion of spatial autocorrelation curves with HVSR for site characterization in Newcastle, Australia

Abstract

In order to investigate site characterization of Newcastle affected by the 1989 Newcastle Earthquake (ML = 5.6), we conducted microtremor array measurements. The spatial autocorrelation (SPAC) method was applied to estimate the S-wave velocity structure from observed microtremor data. Although the inversion of spatial autocorrelation curves is effective for estimating shallow S-wave velocity structures within the sedimentary layer, it is usually difficult to estimate the boundary between the sedimentary layer and bedrock due to a lack of amplitude of vertical components of microtremors at low frequencies. On the other hands, it is well known that horizontal to vertical particle motion spectral ratios (HVSR) have information which assists in resolving the bedrock depth and velocity. Thus, we have applied joint inversion of the spatial autocorrelation curves with HVSR to observed microtremors. Since observed HVSR curves are subject to fluctuations due to unknown Love and body wave contributions or other noise effects, there is difficulty in fitting absolute values of observed HVSR. Therefore, we evaluate observed HVSR and theoretical H/V spectra by zero-lag cross-correlation to fit the shapes of HVSR. The observed HVSR curve has accurate information for the deeper velocity structure and therefore, the estimated velocity model by joint inversion differs in velocity estimations at depth. It is concluded that the joint inversion of the spatial autocorrelation curves with HVSR is useful in practice for obtaining improved estimates of S-wave velocity models down to bedrock.