A study of near source effects in array-based (SPAC) microtremor surveys

Array-based methods for exploiting ambient seismic noise are receiving increasing attention in the literature, particularly for use in providing shear wave velocity profiles to assist with simulation of site-specific earthquake responses. Many of these microtremor methods-such as the spatial autocorrelation technique (SPAC)-operate under the fundamental assumption of plane wave propagation of surface waves. However, when there are significant microtremor sources in close proximity to an array, wave front curvature becomes significant, and the plane wave assumption becomes a tenuous one. This paper explains the use of a simplified geometry-based approach to examine the effect of source distance on SPAC spectra for hexagonal (six-station) and triangular (three-station) arrays. The results suggest that near source effects are generally minimal provided that most of the sources are located a distance equal to at least two array radii from the centre of the array, although this distance increases when attenuation is considered. Results of a field trial involving the deliberate use of near-sources in proximity to a hexagonal (six-station) array support the findings of the theoretical model. Near source effects generally result in the measured SPAC curve being offset to lower frequencies for wavenumbers less than the first secondary maximum of the SPAC curve (kr < 7.0; lambda > 0.86R), resulting in underestimates of shear velocity values in the inverted layered earth model. For the field example, shear velocity was underestimated by up to approximately 17 per cent. Studies of near source effects in a number of theoretical source distributions for hexagonal arrays revealed that cyclic variation in the imaginary SPAC coefficients are indicative of near source effects, although field examples indicate this effect is less significant than the geometric modelling suggests. Modelling of linear microtremor source distributions (intended to represent traffic travelling along a road) resulted in less severe near source effects than full (360 degrees) azimuthal source coverage-a result confirmed by field data. Model and field studies also indicated that in situations of a dominant, close microtremor source of limited azimuthal extent, use of a hexagonal (six-station) array significantly mitigates near source effects compared with a similarly aligned triangular (three-station) array.