Microtremor seismic methods in a 2D valley environment: the Tamar Valley, Launceston (Tasmania, Australia)

The recording of ambient seismic vibrations (microtremors) is a valuable alternative to traditional active seismic methods and strong ground motion analysis when evaluating the velocity structure and pattern of resonance during site resonance or seismic hazard zonation studies. Microtremor observations recorded during two field surveys in 2006 and 2007 in Launceston (Tasmania, Australia) are processed using the spatially averaged coherency spectrum (SPAC) method, using centred hexagonal arrays (n = 6) in 2006 and centred triangular arrays (n = 3) in 2007. The shear wave velocity (SWV) structures evaluated at ten separate sites in Launceston show a great variability within the city limits; the depth to bedrock interface varying from 0m to 250m within a few kilometres. The presence of the ancient Tamar Rift Valley, formed by faulting during the Cretaceous and Tertiary periods and in-filled with soft sediments during the Tertiary period, was identified by Leaman (1994) using a gravity survey and is thought to induce an intricate pattern of resonance across the city. The rapid variations in geology in Launceston violate a fundamental assumption required for the interpretation of SPAC microtremor observations: that the geology can be approximated by a layered earth. It is the principal aim of this project to better understand the capabilities of the SPAC method to evaluate SWV profiles in valley environments. Conventionally, the SPAC method involves recording coherency spectra between pairs of vertical-component sensors azimuthally distributed around a centre sensor. The spatially averaged coherency spectra computed for available inter-station separations are then inverted for the soil shear wave velocity structure. The choice of the SPAC method to interpret microtremor spectra recorded in Launceston with centred hexagonal arrays is justified by a comparison with results obtained from the alternative array based frequency-wavenumber (FK) method. This comparative study demonstrates that the SPAC method is preferred for optimising the domain of validity of the frequency interval to interpret. The reliability of the coherency spectra recorded in Launceston is experimentally assessed by decomposing the spatially averaged coherency spectrum into coherency spectra recorded from pairs of sensors with varying azimuth. The behaviour of the observed coherency spectra with azimuth is found to provide some information concerning the distribution of the microtremor wavefield when using as little as n = 3 or n = 6 sensors in an array. The capabilities of SPAC observations to evaluate SWV structure in valley environments is demonstrated by orienting the SPAC arrays to record coherency spectra of the vertical component on pairs of sensors parallel and perpendicular to the valley axis. Observed coherency spectra recorded above the deepest point of the Tamar Valley in Launceston with two 50m radius centred triangular arrays, and coherency spectra simulated above a 2D model representation of the Tamar Valley with 50m and 100m radius centred triangular arrays suggest the axial component of the coherency spectra (axial-COH) gives a reliable evaluation of the shear wave velocity structure, and is adequate to locate the depth to bedrock interface at the deepest point of the valley. The transverse component of coherency spectra (transverse-COH) shows a discontinuity interpreted as a jump to a higher mode of propagation over the frequency interval 1.0Hz to 1.5Hz, identified as the SV mode of resonance of the deep and narrow Tamar Valley, and alters the spatially averaged coherency spectrum conventionally used to evaluate shear wave velocity profiles. The shear wave velocity profiles evaluated by SPAC observations are combined with single station HVSR microtremor observations recorded along two profiles transverse to the valley axis to evaluate the SV and SH frequencies of resonance and to complete the site resonance of the Tamar Valley in Launceston. Numerical simulations of the propagation of surface waves in a 2D model representation of the Tamar Valley agree well with SPAC and HVSR microtremor observations in Launceston, and demonstrate the potential of axial-COH and HVSR methods to conduct site resonance studies in a valley environment.