In this study, the inversion modeling method
variability of S-wave velocity structure results by
inverting/forward-modeling waveforms and HVSR
curves is investigated. For this purpose, eight KiK-net
seismic stations using 636 foreshocks, mainshock of the
2011 Tohoku earthquake, and 4983 aftershocks obtained from KiK-net are used. The most reliable and representative inverted S-wave velocity structures are based
on close fitting of the theoretical to the observed data of
waveforms and HVSR curves. The inverted S-wave
velocity structures derived from the total stress analyses
are showing high degrees of similarity and consistency
using both EW and NS components and evidenced by
the isotropic HVSRs directionality related to the linear
and nonlinear site effects. Conversely, stiffness variations due to nonlinearity, inferred from inverted S-wave
velocity structures derived from the diffuse field assumption, significantly correspond with stiffness variations evaluated from the observed fundamental resonance frequencies. The corresponding percent of reduction and recovery reflect low variability in S-wave velocities that result from inversion modeling based on
diffuse field assumption. Qualitatively, the presence of
the stiffness degradation process due to linear and nonlinear site responses could be obviously detected using
both inversion modeling methods based on total stress
analyses and diffuse field assumption. Quantitatively,
the precision of the nonlinearity evaluation is questionable due to significant uncertainties in the inverted Swave velocity structures inferred from both inversion
modeling methods. These uncertainties are strongly dependent on the theory of each inversion modeling method. The use of complementary methods with more prior
information is suggested to resolve the nonuniqueness
of the inverted S-wave velocities and hence overconfidence in the precision of the nonlinearity evaluation.