Extracting reflection with wavelet transform in vibroseis signal processing

Abstract

In conventional vibroseis signal processing, algorithms including cross correlation and deconvolution are applied to convert the raw trace data into a seismic section. However, their performance deteriorates when the trace data are corrupted by the ambient noise, so the mathematical tool for time–frequency analysis and wavelet transform is applied in this paper to overcome the difficulty. A time–frequency cross correlation (TFCC) algorithm based on wavelet transform is proposed to extract the reflection from the trace data by detecting the reflected sweeps and estimating their time delay. The source signal and the trace data are transformed into time–frequency domain with respect to a same wavelet basis function; then time–frequency cross correlation is performed between the source signal and the trace data. The reflected sweeps are converted into time–frequency correlation wavelets in the result; meanwhile, the trace data are converted into seismic section. In wavelet decomposition, the high-frequency noise can be suppressed automatically. In the time–frequency representation of the trace data, the ambient noise and the reflected sweeps can be separated from each other. So in the TFCC algorithm, the interference of the ambient noise can be decreased considerably, and the weak reflections can be extracted clearly. Real vibroseis trace data were processed with the TFCC algorithm and the conventional cross correlation. The results showed the superiority of the proposed new algorithm in vibroseis signal processing.