Automatic P-wave picking using undecimated wavelet transform

From the seismologists' point of view, it is extremely important to accurately detect the first P wave arrival time. The P wave arrivals have considerable information about events such as location, magnitude, mechanism, and source parameters. In the classic methods, P wave pickings have been accomplished manually in a visual way. But in the era of information and communication technology, it can be done by computer programs. Seismologists have developed many methods for the picking of the first arrival time of P wave. The wavelet transform is one of the methods to analyze the arrival times and useful for picking up the singularities of any function. Decomposing signals by wavelet transform is a master key to the study of time-frequency varying signals such as earthquake seismograms. This paper presents P phase picking without any prior information using undecimated wavelet transform. For undertaking this study, a simple envelope characteristic function is used for P phase picking. The proposed method is tested on 5 earthquakes recorded by the Fnet network in Japan that have varying signal-to-noise ratio levels for calibrating. Then the method is applied on 50 earthquakes. The observed results are compared with manual phase picking and standard STA/LTA method. The wavelet base method shows the higher accuracy of phase picking in event detection and time picking, respect to the standard STA/LTA method, when compared to manual picking.