Predictive semblance estimation in seismic inversion

A method for estimating shear wave velocity from well log seismograms is described. The results obtained by joining together the ARMA (autoregressive moving average) modeling procedure, which allows accurate estimates but can have estimation problems in the case of noisy data, and the proposed semblance technique, which is particularly robust in the presence of noise, are described. A relevant reduction of the overlap among the different propagating modes is obtained by performing an appropriate preprocessing and by applying the semblance method to the residuals at the outputs of the inverse ARMA filter. This results in a sharper behavior of the semblance function and, consequently, in a more precise estimate of arrival times. The semblance method's low sensitivity to noise is preserved.<>