震源最佳反演结果误差的随机加权估计
|
摘要
提出利用随机加权方法估计地震矩张量最佳反演结果的误差．这种方法不对误差来源作任何分布假设，只需重复产生一定数量的服从Ｄｉｒｉｃｈｌｅｔ分布的权向量．对于每个权向量，将其各分量分配到每条地震记录上，相应地确定一个震源参数解．对这些震源参数反演解进行基本的统计分析，便能估计出最佳解的方差、协方差和置信区间．还讨论了１９８６年台湾地震（Ｍｓ＝７．８）的震源矩张量中存在各向同性分量这一解答的可靠性．
In this paper, the random weighting method is introduced to estimate the error in the best solution of seismic source parameters. This method needs no distribution assumption for the error source. It just needs generating a number of weigthing vectors which follow Dirichlet distribution. For each vector, its components are assigned to the observed seismograms successively, and a source solution iS determined. Analyzing these solutions statistically, one can derives the variance, covariance, and confidence interval of the best solution. The probability that a compressional isotropic component existed in the 14 November 1986 Taiwan earthquake is discussed also.
In this paper, the random weighting method is introduced to estimate the error in the best solution of seismic source parameters. This method needs no distribution assumption for the error source. It just needs generating a number of weigthing vectors which follow Dirichlet distribution. For each vector, its components are assigned to the observed seismograms successively, and a source solution iS determined. Analyzing these solutions statistically, one can derives the variance, covariance, and confidence interval of the best solution. The probability that a compressional isotropic component existed in the 14 November 1986 Taiwan earthquake is discussed also.
引文
［1］Tarantola，A．，Valette B．，InverseProblems－questforinformation，J．Geophys，50，159－170，1982．［2］Efron，B．，Gong，G．，Aleisurelylookatmebootstrap，thejackknife，andcrossvalidation，TheAmericanStatistician，37，36一48，1987．［3］Tichelaar，B．W．，Ruff，L．J．，Howgoodareourbestmodels？jackknifing，bootstrapping，andearthquakedepth，EOS，70，605－606，1989．［4］Sandvol，E．，Hearn，T．，Bootstrapingshear－wavesplittingerrors，Bull．Seism．Soc．Am．，84，1971－1977，1994．［5］郑忠国，随机加权法，应用数学学报，10，247－253，1987．［6］Zheng，T．Y．，Yao，Z．X．，Liu，P．C．，The14November1986Taiwanearthquake─－anevent Withisotropiccomponent，Phys．EarthPlanet．Interi．，91，285一298，1995．［7］刘鹏程、郑天愉，用全局反演方法确定地震震源的最佳和极端模型，地震学报，18，11－18，1996．［8］姚振兴、郑天愉、温联星，用P波波形资料及演中强地震地震矩张量的方法，地球物理学报，37，36─44，1994．［9］Sipkin，S．A．，Interpretationofnon－double－coupleearthquakemechanismsderivedfrommomenttensorinversion，J．Geophys．Res．，91，531－547，1986．［10］Hogg，R．V．，Craig，A．T．，IntroductiontoMathematicalStatistics，FourthEdition，MacmillanPulishingCO．，Inc，NewYork，148一149，1989．［11］Fisz，M．，概率论及数理统计，王福保译，上海：上海科学技术出版社，343－348，1978．［12］Vasco，D．W．，Momenttensorinvariants：searchingfornon－double－coupleearthquakes，Bull．Seism．Soc．Am，80，354－371，1990．［13］Vasco，D．W．，Johnson，L．R．，Inversionofwaveformsforextremesourcemodelswithanapplicationtotheisotropicmomenttensorcomponent，Geophys．J．，97，1－18，1989．
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心