Cross spectral analysis of Swabian Jura (SW Germany) three-component microearthquake recordings
Cross spectral analysis of Swabian Jura (SW Germany) three-component microearthquake recordings
复制标题
施瓦本汝拉（德国西南部）三分量微地震记录的交叉谱分析
DOI：
--
复制
发表时间：
1986
期刊：
影响因子：
--
通讯作者：
J. Wendler
中科院分区：
文献类型：
--
作者：
F. Scherbaum;J. Wendler
在斯瓦比亚汝拉（德国西南部）地震带，对于不同的震源-接收器几何形状，也观察到了类似的三分量微地震记录。该数据集用于研究互谱分析技术的分辨率，以估计相对差分时间以及对速度监测的适用性。差分时间在频域中通过假设线性相位互谱来估计，其中斜率指示个体时间差。所有的地震已经重新定位相对于一个主事件，使用相对P和S延迟时间从交叉频谱分析作为源错位的措施。总体定位误差强烈依赖于主事件和所研究事件之间的初始距离。对于地震最初位于远离约1.5公里，重新定位的结果在总的位置误差较差，而对于事件最初位于接近1公里的重新定位的精度提高。剩余的残差为10 ms的数量级，这大约是数字化间隔的3倍。为了检验互谱分析在速度监测中的适用性，利用合成数据模拟了噪声和震源时间函数差异的影响。在信噪比足够高的情况下，加性白色噪声的影响似乎是可以接受的。然而，在源时间函数的形状的小的变化，被发现有很大的影响，差分时间估计。在相干谱中可以忽略的上升、维持和衰减时间的变化虚假地引入了相位差，就延迟时间而言，相位差很容易达到数字化间隔的大小。因此，速度监测使用交叉频谱分析技术似乎强烈依赖于平等的对比，相似性的源时间函数的事件进行比较。相干谱不是检测所有显著差异的充分量度。
Similar three-component microearthquake records have been observed in the Swabian Jura (SW Germany) seismic zone for different source-receiver geometries. This data set is used to study the resolution power of cross spectral analysis techniques for the estimation of relative differential times as well as the applicability to velocity monitoring. The differential times are estimated in the frequency domain by assuming a linear-phase cross spectrum with the slope indicating the individual time difference. All earthquakes have been relocated with respect to a master event, using the relative P and S delay times from the cross spectral analysis as a measure of source mislocation. The overall location error is strongly dependent on the inital distance between master and studied event. For earthquakes initially located farther apart than approximately 1.5 km, the relocalization result in terms of total location error was poorer, whereas for events initially located closer than 1 km the precision of the relocalization was improved. The remaining residuals are of the order of 10 ms, which is approximately 3 times the digitization interval. In order to test the applicability of cross spectral analysis to velocity monitoring, synthetic data were used to model the influences of noise and source time function differences. The effect of additive white noise seems to be acceptable in cases where the S/N ratio is sufficiently high. Small changes in the shape of the source time function, however, were found to be of great influence to the differential time estimates. Variation of rise, sustain and decay times, which were negligible in the coherence spectrum, spuriously introduced phase differences which, in terms of delay times, easily reach the magnitude of the digitization interval. Thus, velocity monitoring using cross spectral analysis techniques seems to depend strongly on the equality in contrast to similarity of the source time functions of the events which are compared. The coherence spectrum is not a sufficient measure to detect all the significant differences.