Decomposition of vibroseis data by the multiple filter technique

Abstract

A harmonically distorted vibroseis signal is decomposed using the “multiple filter technique” in a similar manner as the wavelet transform. The 1-D vibroseis signal is transformed into a 2-D function of time and frequency. The procedure consists of applying a family of narrow Gaussian filters whose center frequencies are close to the instantaneous frequencies of the harmonically distorted vibroseis signal. As a result of this decomposition, an envelope trace is obtained for every center frequency. For a linear fundamental (pilot) sweep, the fundamental sweep and its harmonic distortions have different arrival times in the decomposed envelope trace. An accurate analysis of the harmonic distortion can be carried out using the amplitudes along the slopes of the instantaneous frequency defined for the linear fundamental sweep and its harmonic distortions. Plotting the contoured amplitudes of each envelope trace on a frequency‐versus‐time scale yields the amplitude distribution between the fundamental sweep and its harmonic distortions. They represent the frequency content of the uncorrelated vibroseis signal and can be used to examine the interaction and leakage of the signal and its harmonics. Similarly, the correlation noise, the so‐called “ghost sweep” in the correlated vibroseis data, can be decomposed with this technique. The method is applied to synthetic, harmonically distorted, vibroseis signals and to a real vibroseis vertical seismic profile (VSP) data set.