论文部分内容阅读
提出了基于方差分量估计的地震同震滑动分布反演方法,该方法不仅可以确定不同数据集的相对权重,而且还能得到平滑因子的值,突破了方差分量估计方法仅用于各类数据定权的实际应用范围。模拟算例验证了此方法的有效性,从解的概率后验密度分布分析来看,方差分量估计方法得到的结果是线性和非线性混合方法反演结果的一个解。用方差分量估计方法反演了Bam地震同震滑动分布,并与线性非线性混合方法反演的结果进行了对比分析。两种方法得到的Bam地震断层滑动分布之间的空间互相关系数为0.999 9,地震同震的最大滑动量相同,都为3.04m;其深度仅有微小区别,前者为4.50km,后者为4.47km;平均滑动量前者为0.714m,后者为0.718m。但前者反演计算用时227s,后者却达到1.5×106 s,约17d,前者仅为后者的1/6 608。结果表明,本文方法相对于线性和非线性混合方法具有计算简单、计算量小和计算效率高的优点。
A seismic inversion method for seismic coseismic slip distribution based on variance component estimation is proposed. This method can not only determine the relative weights of different data sets, but also obtain the values of smoothing factors. It breaks through that the method of variance component estimation is only used for various data sets The practical scope of the right. Simulation results show the effectiveness of this method. From the point of view of the probabilistic posterior density distribution analysis, the result of the variance component estimation method is a solution to the inversion results of the linear and nonlinear hybrid methods. The covariance slip of the Bam earthquake is retrieved by the method of variance component estimation, and compared with the linear non-linear hybrid inversion method. The spatial cross correlation coefficient between the slip distributions of the Bam seismic fault obtained by the two methods is 0.999 9, and the maximum amount of slippage of the earthquake coseismicity is the same, both of which are 3.04 m; the depth is only slightly different with the former being 4.50 km and the latter being 4.47km; the average slip is 0.714m for the former and 0.718m for the latter. However, the former calculation time of 227s, the latter reached 1.5 × 106 s, about 17d, the former is only the latter 1/608. The results show that the proposed method is simpler, less computationally expensive and more computationally efficient than the linear and nonlinear hybrid methods.