Computation of Virtual Far-offset Refractions Using Near-offset Refractions from Deep Land Seismic Data

A recurrent difficulty encountered when picking first arrivals on deep seismic data is the very low signal-to-noise (SNR) ratio that characterizes the far-offset traces. This has a direct impact on the accuracy of the applied static corrections and on the results of the seismic tomography. In order to remedy this problem, further enhancement of these traces has to be done. This effect can be achieved using the Refraction Interferometry (RI) method. According to Bharadwaj et al. (2011) , long-offset virtual refractions can be obtained from near-offset refractions in two steps: cross-correlation and summation of the traces for a position x of the receiver and two different sources (A and B) and then convolution of the result with the real far-offset refractions. The obtained traces have virtual head-wave arrivals characterized by higher amplitudes and continuity than the raw far-offset refractions. This method proves to be more efficient if the CSGs are previously filtered. Also, for crooked profiles, the pairs of common-source gathers used for RI have to be carefully chosen in order to obtain reliable results.