Characterization of Underwater Explosions by Spectral/Cepstral Analysis, Modeling and Inversion
Abstract
This report describes the development and evaluation of a cepstral analysis, modeling and inversion program for the characterization of underwater explosions recorded by seismic stations. The algorithm matches synthetic against observed cepstrums for suspected underwater blasts. The observed cepstrums are computed for regional phases (e.g., Pn, Pg, Sn, and Lg) by taking the logarithm of the trend-corrected spectrums for each phase, stacked across a regional array if array data is available, and then taking the inverse Fourier transform of log amplitude spectrum. The signed cepstrums for regional phases from underwater explosions have negative peaks caused by the reflections of the acoustic wave from the surface and positive peaks from the bubble pulse. The depth and yield of the underwater blast are determined by finding a synthetic cepstrum that most closely matches the observed cepstrum. The best matching synthetic cepstrum is the one with the highest match, either by cross correlation coefficient L1 Norm, or L2 Norm, with the observed cepstrum. This algorithm has been tested on the November 1999 DTRA sponsored calibration blasts in the Dead Sea, and the method gives explosion yields consistent with the known yields and depths of the events. Results are also given for analysis of two events in the Barents Sea, an explosion near Murmansk and the Kursk submarine event. We conclude that the Kursk event was an underwater explosion with a yield of about 4300kg, or 4.73 tons, at a depth of 90 m.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 2005
- Accession Number
- ADA443931
Entities
People
- Angelina Freeman
- Douglas R. Baumgardt