Automatic P-Wave Arrival Picking Based on Inaction Method

Abstract

Accurately picking the arrival of the P-wave is of great significance for earthquake relief and emergency rescue. The picking accuracy of some previous methods was easily disturbed by noise. We propose an automatic picking algorithm for seismic P-wave first arrivals based on the inaction method (IM) and the Akaike information criterion (AIC), called the IMAIC method, to alleviate the above problems. We use the IMAIC method to directly extract the seismic phase characteristics from the normal time-frequency transform (NTFT) spectrum to pick the P-wave arrival, which does not require inverse transformation of the NTFT. The proposed method essentially implements bandpass filtering to avoid noise interference to the greatest extent possible while optimally preserving the important features of the signal, such as the sharp direct P-wave arrival. We applied the IMAIC method to 9909 aftershock records from Wenchuan in 2008 to verify its reliability and robustness, taking the manually picked P-wave arrival as a reference. The average deviation is -0.010 ± 0.170 s when using the proposed IMAIC method. Furthermore, a comparative study of P-wave arrivals that are automatically derived using multiscale wavelet analysis and singular value decomposition (SVD)-like picking algorithms with the same data set demonstrates the effectiveness of the IMAIC method. The proposed method offers much promise for picking seismic phase onsets, even when using data sets with a low signal-to-noise ratio (SNR).