In the processing of P-wave data one important product of computing static corrections is a P-wave near-surface velocity model. This model is usually inverted from the traveltimes of the refracted waves. In the case of converted-wave data, for which S-wave near-surface corrections must be used on the receiver locations, refracted S-waves are usually not available or hard to identify. Here, we propose an inversion approach based on the τ differences obtained by crosscorrelating τ-p receiver gathers from different locations. For this, the structure of the near-surface at a given location must be know. Then, the τ-p receiver gathers are crosscorrelated with the gather obtained at the reference location, and the time lag of the maximum of the crosscorrelation function is picked. These picks along with an initial guess of the depth of the near-surface layer, its S-wave velocity, local dip and velocity of the medium underneath (replacement velocity) are the input to initialize the inversion. An iterative quasi-Newton inversion approach is used in this study. Different data conditions are used to study the robustness of the inversion. Results show that the inversion of the depth of the near-surface layer is very sensitive to the presence of noise in the picks and the lack of large rayparameter values (p > 0.5ms/m). Although to a lower degree, inverted velocities were also affected by these conditions. However, inverted dips displayed very stable results under different data conditions. Alternative inversion approaches able to exploit the robustness of the inverted dips must be considered to improve the results provided in this study.
View full article as PDF (0.91 Mb)