Automatic Onset Time Picking Algorithms For Multiple Seismic Arrivals

The need for accurate, dependable onset time picking is of significant importance toseismological community for a variety of additional applications. For example, the onset-timepicking speed and accuracy is directly related to the efficiency and precise in earthquakerelocation, seismic phase recognition and focus dynamic process. In the early day, the onsettime picking was only done by manually, which is extremely time-consuming and subjective,as similarly qualified person will pick onsets at different times. With the fast development ofcomputational technique, data acquisition and data processing technologies, as well as thecoming quantitative or digital seismology, the onset time picking has been transformed fromthe early-stage manually analysis to the middle-stage human-computer interaction, and thento the current-stage automatic picking. At present, there are many algorithms for onset timepicking using single-or three-component recordings, but none of them is able to make aconsistent picking for different signal/noise levels, source environments, travel paths andreceiver locations, especially when the noise and signal has nearly the same frequencycontent, and the later arrivals are buried in the coda of early arrivals.In this paper we give a briefly review, analyze and summarize the current methodsconcentrated ourselves in time domain, frequency domain, time-frequency domain andcomprehensive methods, and an analysis and compare to the methods of the amplitude ratiomethod, improved Coppens’s method, improved STA/LTA method, the curve length ratiomethod, the fractal dimension method, autocorrelation method and VAR-AIC method hasbeen made by using the practical examples of primary wave automatic detecting in the hopethat we can do some improvements of these methods.Finally, we developed four methods forthe first arrival automatic detecting and identification: a modified correlation method, twomodified AIC method, and a comprehensive method by combining the improved amplituderatio method, improved Coppens’s method, fractal dimension method and the modifiedcorrelation method, and we test their effectiveness through applying these methods todifferent signal to noise ratio of the different synthetic seismic records and the actual seismicdata, achieving a good detection. The results show that it is constructive to look for a comprehensive method, which were based on the differences between the signal and noise,and between different seismic phases in terms of kinematic, dynamic to form a simplealgorithm, high picking accuracy, multi-channel functional and suitable for real timecomprehensive picking method and technology.