根据非线性贝叶斯理论的界面波频散曲线反演  

Interface-wave dispersion curves inversion based on nonlinear Bayesian theory

作　　者：李翠琳[1,2] Stan E Dosso Hefeng Dong 

机构地区：[1]中国科学院海洋研究所海洋地质与环境重点实验室,青岛266071 [2]中国海洋大学工程学院,青岛266100 [3]University of Victoria Victoria V8W 3P6 Canada [4]Norwegian University of Science and Technology NO-7491 Trondheim Norway

出　　处：《声学学报》2012年第3期225-231,共7页Acta Acustica

基　　金：国家科技重大专项(2011ZX05026-004-06);国家高技术研究发展计划(2009AA093401)资助项目

摘　　要：通过时频分析法从海底环境噪声数据中提取界面波频散曲线,进而采用非线性贝叶斯反演方法估算海底沉积物厚度、剪切波速度、压缩波速度和密度等参数及其不确定性。参数的最大后验概率(MAP)估计值和边缘概率分布分别通过自适应单纯形模拟退火法和Metropolis-Hastings采样法在各参数先验区间内搜索获得,采用贝叶斯信息准则(BIC)从不同参数化模型中选择最优模型。界面波频散曲线反演结果表明:满足实测数据的最优海底模型结构为3层均匀分布剪切波速度剖面结构,海底深度的反演精度在800m以内,比起压缩波速度和密度,剪切波速度的不确定性更小,对界面波频散曲线更敏感。This paper applies a dataset of ocean ambient noise data to extract interface-wave dispersion curves using time-frequency analysis.The nonlinear Bayesian inversion is applied to estimate seabed sediment parameters such as thickness,shear-wave velocity,compression wave velocity and density,and their uncertainties from interface-wave dispersion curves.The maximum a posterior（MAP） model and marginal probability distributions of parameters are estimated using posterior probability densities computed by adaptive simplex simulated annealing and Metropolis-Hastings sampling methods.The Bayesian information criterion is applied to determine the optimal model that fully explains the observed data by the different parameterizations.The inversion results indicate that 3-uniform-layer model is chosen as the preferred parameterization.The resolution of inversion is up to 800 m-depth.The shear-wave velocity and layer thickness have fewer uncertainties and are more sensitive to the interface wave dispersion than the compression wave velocity and density.

关 键 词：贝叶斯理论 曲线反演 面波频散 非线性 剪切波速度 贝叶斯信息准则 边缘概率分布 最大后验概率 

分 类 号：P736.21[天文地球—海洋地质]