One uses seismic interferometry (SI) to recover Green's functions (i.e.
impulse response) from ambient seismic recordings and estimate surface-wave phase
velocities to investigate subsurface structure. This method has been commonly used
in the last 20 years because this method only utilizes ambient seismic recordings
from seismic stations/sensors and does not rely on traditional seismic sources (e.g.
earthquakes or active sources). SI assumes that the ambient seismic wavefield is
isotropic, but this assumption is rarely met in practice. We demonstrate that, with
linear-array spatial sampling of an anisotropic ambient seismic wavefield, SI provides
a better estimate of Rayleigh-wave phase velocities than another commonly used ambient
seismic method, the refraction microtremor (ReMi) method. However, even SI does not work
in some extreme cases, such as when the out-of-line sources are stronger than the inline
sources. This is because the recovered Green's functions and surface-wave phase
velocity estimations from SI are biased due to the anisotropic wavefield. Thus, we propose
to use multicomponent data to mitigate this bias. The multicomponent data are vertical
(Z) and radial (R) components, where the R direction is parallel to a line or great circle
path between two sensors. The multicomponent data can deal with the extreme anisotropic
source cases, because the R component is more sensitive to the in-line sources than the
out-of-line sources, while the Z component possesses a constant sensitivity to sources in
all directions.
Estimation of source distributions (i.e. locations and strengths) can aid correction of
the bias in SI results, as well as enable the study of natural ambient seismic sources
(e.g. microseism). We use multicomponent seismic data to estimate ambient seismic source
distributions using full-waveform inversion. We demonstrate that the multicomponent data
can better constrain the inversion than only the Z component data, due to the different
source sensitivities between the Z and R components. When applying the inversion to field
data, we propose a general workflow which is applicable for different field scales and
includes vertical and multicomponent data. We demonstrate the workflow with a field data
example from the CO2 degassing in Harstouˇsov, Czech Republic. We also apply the
workflow to the seismic recordings in Antarctica during February 2010 and estimate the
primary microseism source distributions.
The SI results include both direct and coda waves. While using the direct waves in
investigating subsurface structure and estimating source distributions, one can utilize
the coda waves to monitor small changes in the subsurface. The coda waves include
multiply-scattered body and surface waves. The two types of waves possess different spatial
sensitivities to subsurface changes and interact each other through scattering. We present a
Monte Carlo simulation to demonstrate the interaction in an elastic homogeneous media. In the
simulation, we incorporate the scattering process between body and Rayleigh waves and the
eigenfunctions of Rayleigh waves. This is a first step towards a complete modelling of
multiply-scattered body and surface waves in elastic media.