Importance of covariance components in inversion analyses of densely sampled observed data: an application to waveform data inversion for seismic source processes
Importance of covariance components in inversion analyses of densely sampled observed data: an application to waveform data inversion for seismic source processes
复制标题
DOI：
10.1111/j.1365-246x.2008.03884.x
复制
发表时间：
2008-10
影响因子：
2.8
通讯作者：
Y. Yagi;Y. Fukahata
中科院分区：
文献类型：
--
作者：
Y. Yagi;Y. Fukahata
在空间和（或）时间上名义上连续的数据是在天体物理学的各种观测中获得的。由于计算机技术的提高，我们现在可以以非常高的采样率反演这样的观测数据。密集采样的观测数据通常不是完全相互独立的，因此我们必须考虑这种影响。对于地震波形数据，由于地球非弹性衰减的影响，它们至少具有时间相关性。考虑到数据的协方差，我们已经开发了一种方法，地震震源反演，并将其应用于2003年阿尔及利亚Boumerdes-Zemmouri地震的地震P波数据。通过比较有和没有协方差分量的最终滑移分布，我们发现协方差分量的影响对于较高采样率（≥5 Hz）的数据集至关重要。如果我们忽略协方差分量，反演结果变得不稳定，由于高估的信息从观测数据。到目前为止，人们普遍认为，我们可以得到一个更好的震源过程的图像，通过反演波形数据与更高的采样率。然而，观测数据的协方差分量来源于地球的非弹性效应，这限制了反演震源模型的分辨率。
SUMMARY Nominally continuous data in space and/or time is obtained in various observations in geophysics. Due to an enhanced technology of computers, we can now invert such observed data with a very high sampling rate. Densely sampled observed data are usually not completely independent of each other, and so we must take this effect into account. As for seismic waveform data, they have at least temporal correlation due to the effect of inelastic attenuation of the Earth. Taking the data covariance into account, we have developed a method of seismic source inversion and applied it to teleseismicP-wave data of the 2003 Boumerdes-Zemmouri, Algeria earthquake. From the comparison of the final slip distributions inverted with and without the covariance components, we found that the effect of covariance components is crucial for a data set of higher sampling rates (≥5 Hz). If we neglect the covariance components, the inverted results become unstable due to overestimation of the information from observed data. So far, it has been widely believed that we can obtain a finer image of seismic source processes, by inverting waveform data with a higher sampling rate. However, the covariance components of observed data originated from inelastic effect of the Earth give a limitation on the resolution of inverted seismic source models.