Discrimination of Seismic Sources Using Israel Seismic Network.
Abstract
Regional Densed Seismic Networks (RDSN) have additional as yet uninvestigated potential for discriminating weak local earthquakes and quarry blasts. Even conventional single station discriminants, such as P/S and spectral ratios are significantly improved, after averaging across the Israel Seismic Network, which consists of 36 short period stations. This report documents a study aimed at the development of new techniques specially designed for RDSN oriented discrimination: (1) subnet average of the seismic energy ratio between the low (1-6 Hz) and high (6-11 Hz) frequency ranges; (2) spectral semblance, measuring subnet spectral shapes coherency; (3) velogram analysis evaluating the different kinematic features of seismic waves for shallow and deep events. The algorithms were tested on 212 events: earthquakes, quarry ripple-fired and single blasts, and underwater explosions from some areas of the Middle East region with a 97-100% success rate. The study of this physically approved algorithms was complimented by testing of the multivariate procedures based on formal Integrative Approach: King's clustering procedure, Linear Discrimination Function and Artificial Neural Networks. When applied to a vector of spectral parameters derived from the Galilee data base, they provided 99-100% of true classification in a cross-validation test. All the procedures are applicable to routine processing of seismograms, thus significantly improving discrimination performance.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1996
- Accession Number
- ADA317385
Entities
People
- Avi Shapira
- Vladimir Pinsky
- Yefim Gitterman