Abstract:

The modern fundamental theory of statistical signal processing is introduced to the analysis and processing of seismic signals. Based on one-dimensional slice of cross-fourth-order cumulants, a novel method of seismic signals detection and first arrival pickup is proposed with the principle and the algorithm. Using horizontal waveform similarities, as well as in allusion to the non-sensibility of  one-dimensional slice's cross-fourth-order cumulants to Gaussian noise, and thereby the method achieves first arrival automatic extracting. The result shows this method can effectively extract first arrival take-off time in the environment of strong Gaussian colored noise interference，and the signal-to-noise ratio reaches -3 dB. In the implementation of real data，this method can quickly and accurately determine the position of seismic signal first arrival, and the results are basically consistent with the manual extraction.

Key words: cross-fourth-order cumulants, one-dimensional slice, seismic wave, first arrival automatic extraction, colored Gaussian noise