First arrivals picking based on graph signal theory

摘要

Picking first arrivals is a fundamental but not easy task in the oil seismic exploration, especially in situations where the signal-to-noise ratios are very low. The paper proposed a new picking algorithm with the emerging new graph signal processing technology in the signal processing field. Fundamentally, first arrival picking is actually as a problem to recognize direct seismic waves among background noises. It is apparently that background noises and the direct waves are from different sources, and therefore, there is no connectivity among them, which formulates the fundamentals of the proposed method. With the help of graph signal processing theory, such non-connectivity is fully explored and cast into an optimization problem for picking the first arrivals in oil exploration. From simulations and real measurements results, the proposed method is validated for first arrival picking with good performances, especially in situations with low signal-to-noise ratios.