摘要
对实际两层各向异性介质,在一层假设条件下利用Silver和Chan(1991)的等效单层各向异性测量方法,以及在两层假设条件下利用剥去法和反演法,计算模型的分裂参数.首先利用理论雷克子波得到经过两层各向异性的观测数据,然后加入噪声之前和之后,分别计算不同反方位角覆盖条件下数据的各向异性,并与理论模型的分裂参数进行比较.结果发现在上层分裂特征已知的基础上,剥去法适用于任何模型,不管是否存在分裂,或者分裂特征在两层中如何分布;而反演法更适用于两层各向异性较强且不同的模型,其结果受噪声和方位角分布的影响.选取不同方位角分布的高信噪比数据,比较剥去法和反演法对下层分裂参数的计算结果,发现当入射方位角在两层快波方向之间时,反演结果比较可靠.最后利用这三种测量各向异性的方法,计算青藏高原东南缘CEArray台站观测到的SKS数据中记录的各向异性.除了云贵高原的台站YN.CUX表现出较明显的两层分裂特征,大部分台站下的岩石圈各向异性呈现一层或者较弱的各向异性.于是针对该台站,用反演法同时对这两层的分裂参数进行估计,上层的结果与接收函数计算的地壳分裂参数一致;下层的结果与剥去法计算的上地幔分裂参数一致,说明反演法同剥去法一样,都能有效的分析具有两层各向异性特征的介质.
For a two-layer anisotropic medium,we calculated the splitting parameters of the model,by using three different measurement methods:(1)the traditional one-layer estimating method of Silver and Chan(1991),(2)the top-layer stripping-off method,and(3)a simultaneous gridding search method.We generated a large waveform dataset for a total of 8 models using the Ricker wavelet as the source wavelet.We then investigated how well these three methods can recover the input models under different noise and limited backazimuthal coverage conditions.Our numerical tests indicate that the top-layer stripping-off method can recover the input models even when the data are from a narrow azimuthal range and have the level of noise in real data.In contrast,the grid-search based inversion method is more suitable for models with the fast directions of the two layers being a few tens of degrees away from each other.We also applied these three methods to the SKS waveform data recorded by 13 CEArray stations at the southeastern margin of Tibetan plateau.We found that the SKS data recorded by the YN.CUX station on the Yunnan-Guizhou plateau exhibit obvious two-layer splitting characteristics,while those from the other stations seem to be well explained by a one-layer anisotropic model.