Multivariate Seismic Calibration for the Novaya Zemlya Test Site

• Main Authors: Fisk, Mark D., Gray, Henry L., Alewine, III, Ralph W., McCartor, Gary D.
• Other Authors: MISSION RESEARCH CORP SANTA BARBARA CA
• Format: Text
• Language: English
• Published: 1992
• Subjects: Seismology; Nuclear Weapons; *YIELD(NUCLEAR EXPLOSIONS); *NUCLEAR EXPLOSION TESTING; *SEISMOLOGY; *SEISMIC DATA; USSR; NETWORKS; YIELD; CALIBRATION; SEISMIC ARRAYS; SEISMOLOGICAL STATIONS; AMPLITUDE; EPICENTERS; SEISMIC WAVES; EXPLOSION EFFECTS; UNDERGROUND EXPLOSIONS; EXPLOSIONS; NUCLEAR EXPLOSIONS; NORSAR(NORWEGIAN SEISMIC ARRAY); MULTIVARIATE CALIBRATION; CONFIDENCE INTERVALS; NOVAYA ZEMLYA TEST SITE; GRAFENBERG; TTBT(THRESHOLD TREATY BAN TEST); PE62714E; WUPL9A10DAAC; Novaya Zemlya
• Online Access: http://www.dtic.mil/docs/citations/ADA261725; http://oai.dtic.mil/oai/oai?&verb=getRecord&metadataPrefix=html&identifier=ADA261725

Within the last year, Soviet yield data have been acquired by DARPA for over 40 underground nuclear explosions at the Novaya Zemlya Test Site between 1964 and 1990. These yields are compared to previous estimates by other authors, based on observed seismic magnitudes and magnitude-log yield relations transported from other test sites. Several discrepancies in the yield data are noted. Seismic magnitude data, based on NORSAR Lg and P coda, Grafenberg Lg, and a world-wide m sub b, have been published by Ringdal and Fyen (1991) for 18 of these events. A similar set of Soviet network magnitudes have been published by Israelsson (1992). Using these data, estimates of the multivariate calibration parameters of the magnitude-log yield relations are computed. An outlier test is applied to the residuals to the lines of best fit. One of the two smallest events is identified as an outlier for every multivariate magnitude combination. A classical confidence interval is presented to estimate future yields, based on estimates of the unknown multivariate calibration parameters. A test of TTBT compliance and a definition of the F-number, based on the confidence interval, are also provided. F-number estimates are obtained for various magnitude combinations by jackknifing. The reliability of the results is discussed, in light of the fact that the data are tightly clustered for 16 of the 18 events. Prepared in cooperation with Southern Methodist University, Dallas, TX.