Analysis of three-component ambient vibration array measurements

Both synthetic and observed ambient vibration array data are analysed using high-resolution
beam-forming. In addition to a classical analysis of the vertical component, this paper presents
results derived from processing horizontal components.We analyse phase velocities of fundamental
and higher mode Rayleigh and Love waves, and particle motions (ellipticity) retrieved
from H/V spectral ratios. A combined inversion with a genetic algorithm and a strategy for
selecting possible model parameters allow us to define structural models explaining the data.
The results from synthetic data for simple models with one or two layers of sediments suggest
that, in most cases, the number of layers has to be reduced to a few sediment strata to find
the original structure. Generally, reducing the number of soft-sediment layers in the inversion
process with genetic algorithms leads to a class of models that are less smooth. They have a
stronger impedance contrast between sediments and bedrock.
Combining Love and Rayleigh wave dispersion curves with the ellipticity of the fundamental
mode Rayleigh waves has some advantages. Scatter is reduced when compared to using
structural models obtained only from Rayleigh wave phase velocity curves. By adding information
from Love waves some structures can be excluded. Another possibility for constraining
inversion results is to include supplementary geological or borehole information. Analysing
radial components also can provide segments of Rayleigh wave dispersion curves for modes
not seen on the vertical component. Finally, using ellipticity information allows us to confine
the total depth of the soft sediments.
For real sites, considerable variability in the measured phase velocity curves is observed. This
comes from lateral changes in the structure or seismic sources within the array. Constraining the
inversion by combining Love and Rayleigh wave information can help reduce such problems.
Frequency bands in which the Rayleigh wave dispersion curves show considerable scatter are
often better resolved by Love waves.
Information from the horizontal component can be used to correctly assign the mode number
to the different phase–velocity curve segments, especially when two modes seem to merge at
osculation points. Such merging of modes is usually observed for Rayleigh waves and thus can
be partly solved if additional information from the Love waves and the horizontal component
of Rayleigh waves is considered. Whenever a site presents a velocity inversion below the top
layer, Love wave data clearly helps to better constrain the solution.