A P waves' automatic picking by detecting the changes of seismic signals' stationary random process through similarity analysis. Issue 115 (December 2018)

• Record Type: Journal Article
• Authors: Wang, Xu; Fu, Jihua; Tang, Chengpei; Li, Zhitao; Wang, Jianjun
• Is Part Of: Soil dynamics and earthquake engineering. Issue 115(2018)
• Page Start: 225
• Page End: 231
• Publication Date: 2018-12
• Subjects: Seismic events detection -- P waves' picking -- Similarity analysis method -- Stationary random process; Soil dynamics -- Periodicals; Earthquake engineering -- Periodicals; Sols -- Dynamique -- Périodiques; Génie parasismique -- Périodiques; 624.176205
• DOI: 10.1016/j.soildyn.2018.08.006
• Languages: English
• ISSNs: 0267-7261

Abstract

Abstract: Detection of P waves in the seismic phrases is essential for studying the earthquake ground motion. A P waves' automatic picking by detecting the changes of seismic signals' stationary random process is put forward through the similarity analysis in the paper. Firstly a known template of stationary random process is composed or chosen. And the similarity coefficients between the chosen template and the seismic signal are calculated, and they present the stability of the seismic signal. Then a given threshold of the similarity coefficients' changes is applied to detect the significant changes of the similarity coefficients, which represents the P waves' arrivals. By the similarity coefficient method, the detecting results of P waves accompanied confidence level are obtained. This similarity analysis method was tested by both the main-shocks and the aftershocks of the Ms 8.0 Wenchuan Earthquake. Comparing the picking results, the similarity coefficient method performed better than the short-term and long-term average ratio (STA/LTA) algorithm. For given examples, the mean error of the P waves' picking time by the similarity analysis method (−0.13 s) is less than the mean error by the STA/LTA method (−0.33 s). When noise is added, the mean error of the P waves' picking time by the similarity analysis method increased to 0.25 s, that is still less than the mean error by the STA/LTA method (4.07 s). Highlights: The similarity analysis method could detect the state changes. A dimensionless process can eliminate the influences by different types of instruments on the similarity coefficients calculated. Confidence level is obtained in the similarity analysis method.