Tests of relative earthquake location techniques using synthetic data
Tests of relative earthquake location techniques using synthetic data
复制标题
DOI：
10.1029/2004jb003380
复制
发表时间：
2005-04
影响因子：
--
通讯作者：
G. Lin;P. Shearer
中科院分区：
文献类型：
--
作者：
G. Lin;P. Shearer
[1]我们比较了三个相对地震定位技术使用的合成数据，模拟了真实的走时数据的统计特性的测试。这些方法是（1）Jordan和Sverdrup（1981）的重心分解法，（2）Richards-Dinger和Shearer（2000）的源特定站项法（SSST），（3）Waldhauser和Ellsworth（2000）的修正双差法（DD）。我们在半空间速度模型中生成一组合成地震、台站和到达时间。我们模拟随机拾取误差，车站条款，和一般的三维速度结构所造成的走时变化的效果。我们使用常见的线性化方法实现这些算法，并使用共轭梯度法求解系统。我们约束的平均位置偏移为零的hypocentroid分解和双差位置。对于单个紧凑的事件集群，这三种方法在相对定位精度上产生非常相似的改进。对于分布式地震活动，DD和SSST算法都提供了改进的相对位置的可比精度。我们还提出了一种新的定位技术，称为收缩盒SSST方法，它提供了一些改进的绝对定位精度相比，SSST方法。在我们实现这些算法时，SSST方法运行速度明显快于DD方法。
[1] We compare three relative earthquake location techniques using tests on synthetic data that simulate many of the statistical properties of real travel time data. The methods are (1) the hypocentroidal decomposition method of Jordan and Sverdrup (1981), (2) the source-specific station term method (SSST) of Richards-Dinger and Shearer (2000), and (3) the modified double-difference method (DD) of Waldhauser and Ellsworth (2000). We generate a set of synthetic earthquakes, stations, and arrival time picks in half-space velocity models. We simulate the effect of travel time variations caused by random picking errors, station terms, and general three-dimensional velocity structure. We implement the algorithms with a common linearized approach and solve the systems using a conjugate gradient method. We constrain the mean location shift to be zero for the hypocentroidal decomposition and double-difference locations. For a single compact cluster of events, these three methods yield very similar improvements in relative location accuracy. For distributed seismicity, the DD and SSST algorithms both provide improved relative locations of comparable accuracy. We also present a new location technique, termed the shrinking box SSST method, which provides some improvement in absolute location accuracy compared to the SSST method. In our implementation of these algorithms, the SSST method runs significantly faster than the DD method.