Problems with the standard vibroseis deconvolution: some practical solutions

In conventional vibroseis data processing, the recorded sweep is correlated with the recorded trace, assuming that the estimated groundforce is equal to the sweep, but it is demonstrated that due to the non-linearity associated with the vibrator system, the true groundforce and the recorded sweep are not the same. Further, while processing, it is assumed that the zero phase Klauder wavelet is convolved with the Earth's reflectivity, completely ignoring the fact that the far-field velocity is basically proportional to the time-derivative of the groundforce, hence on real data, it is not possible to achieve an ideal zero-phase wavelet. Additionally, the Earth's low-pass filtering generates a mixed-phase signal, which is not suitable for conventional deconvolution. We demonstrate that the recorded sweep contaminates the traces with harmonic noise and recommend a pre-determined sweep for cross-correlation. Further, to avoid minimum-phase violation issue, we endorse the frequency domain sweep deconvolution method (FDSD). Our results on synthetic as well as real seismic data, generated after FDSD, show significant improvement in resolution and noise suppression, as compared to the cross-correlated one. It was also shown that the predictive deconvolution combined with FDSD should be used to generate deconvolved seismic sections, and the spiking deconvolution should be avoided.