OPTIMIZED REGULARIZATION OF INTERVAL S-WAVE VELOCITY ESTIMATION FROM SEISMIC SURFACE WAVE DATA

Surface waves can be used to estimate near-surface shear wave velocity (VS) models due to their dispersive nature: local dispersion curves (DCs) are subjected to a significant dependency to time-average velocity (Vsz) and the relationship between the wavelength of surface waves and the investigation depth is used to directly transform DCs into Vsz, avoiding the inversion procedure. Such a technique, known as W/D method, has been recently introduced but the estimation of the interval VS from noisy Vsz using a DIX type equation requires regularization of the data. A synthetic dataset of 1D VS models has been generated with a stochastic approach based on a Monte Carlo procedure. A well-structured MATLAB code was developed to simulate the interval VS under different noise levels to test and compare the regularization parameter that best fits the simulated VS with its original profile, in an attempt to correlate noise levels and regularization parameter. Such a correlation is then tested in a real dataset to validate its effectiveness.