Studies of Fault Fabrics and Earthquake Mechanics from the Precise Relative Locations of Microearthquakes

具有相似震源（位置）和震源机制（断层面和滑动方向）的微地震（小于3级的地震）在给定的地震台产生的地面运动看起来非常相似。通过比较这些地震记录的地震波波形，可以非常准确地估计地震波的相对到达时间，并最终非常准确地估计地震的相对位置。对于相距数百米的事件，用这种方法得到的相对定位误差为几十米；对于相距几十米或更短的事件，经过时间相关台站变化校正后，得到的相对定位误差为米。这些误差比美国地质调查局北加州地震台网（NCSN）公布的目录中的误差小10-100倍。这种分辨率的提高在断层研究的两个不同领域带来了重要的新见解。首先，以前无法看到的断层带内部结构已经被成像。其次，由于最近的微地震之间的相对定位误差远远小于破裂尺寸（几十到几百米），这种分辨率允许使用数千个事件的数据集来研究地震相互作用。在断层构造区，微震重新定位表明，沿许多爬行断层的地震活动性被组织成滑动平行的“条纹”，高数十至数百米，长达公里。在破裂相互作用区域，已经观察到沿圣安德烈亚斯断层中央的微地震余震分布是非常不对称的，最近的余震（在空间和时间上）发生在先前主震的西北而不是东南。这是由于断层上物质特性的对比，以及这种对比如何影响主震的动力学（优先向东南传播）。如果这种解释是正确的，那么它与地震危险有关，因为大地震的指向性将地面震动集中在传播方向上。目前的研究致力于在地质感兴趣的地区开发数千个精确定位的微地震目录，根据断层力学的含义解释这些目录，并改进重新定位方法。正在为NCSN台站确定更多的随时间变化的校正，并提供给地震学界。通过台站校正获得的更精确的位置被用于搜索(1)与观测到的微地震线相关联的断层倾角的细微变化；(2)被认为有小滑动的断层上的短线，以区分关于线的起源的不同观点。在非对称余震分布区域，使用Parkfield HRSN目录将数据集扩展到较小的主震规模，正在修改重新定位代码（具有时间相关的站点更正），以常规确定局部跨断层p波和s波速度对比，并且正在进行不同材料界面上破裂的数值模拟，以测试对对称性起源的解释。最后，将不同断层微地震余震序列的统计数据与速率-状态摩擦解析模型的预测结果以及平面断层破裂相互作用的数值模型进行了比较。目的是测试速率和状态故障摩擦模型，并确定各种故障的加载速率和/或表面特性的差异。由于在触发事件之后的网络“盲”时间（可能是几十秒），对于一些断层，余震序列的最早部分可能已经“丢失”，因此正在处理所有已编目事件的近100 s存档波形，以识别和定位通常会触发网络的事件。这些研究将导致对地震机制的进一步了解。

Microearthquakes (earthquakes smaller than magnitude ~3) that have similar hypocenters (locations) and focal mechanisms (fault planes and slip directions) produce ground motions at a given seismic station that appear very similar. By comparing the recorded seismic waveforms of these earthquakes, it is possible to obtain very accurate estimates of the relative arrival times of the seismic waves, and ultimately very accurate estimates of the relative locations of the earthquakes. Errors in relative location obtained using this method are tens of meters for events separated by hundreds of meter, and, after correcting for time-dependent station changes, meters for events separated by tens of meters or less. These errors are a factor of 10-100 smaller than those in the published US Geological Survey Northern California Seismic Network (NCSN) catalog. This increase in resolution has led to significant new insights in two different areas of fault studies. First, structures internal to the fault zone that previously could not be seen have been imaged. Second, because the relative location errors between the nearest microearthquakes are much smaller than the rupture dimensions (tens to hundreds of meters), this resolution allows studies of earthquake interaction using datasets of many thousands of events. In the area of fault structure, microearthquake relocation shows that the seismicity along many creeping faults is organized into slip-parallel "streaks", tens to a few hundred meters tall and up to kilometers in length. In the area of rupture interaction, it has been observed that the distribution of aftershocks of microearthquakes along the central San Andreas fault is very asymmetric, with many more of the nearest aftershocks (in space and time) occurring to the northwest of a prior mainshock than to the southeast. This is attributed to the contrast in material properties across the fault and how this contrast affects the dynamics of the mainshock (preferential propagation to the southeast). If this explanation is correct, it is relevant to seismic hazards because directivity of large earthquakes concentrates ground shaking in the direction of propagation.Current research is devoted to developing catalogs of thousands of precisely-located microearthquakes in regions of geologic interest, interpreting these catalogs in terms of their implications for fault mechanics, and improving the relocation method. More time-dependent corrections are being determined for NCSN stations and made available to the seismological community. The more precise locations obtained with the station corrections are being used to search for (1) subtle changes in fault dip associated with the observed microseismic lineations and (2) short lineations on faults believed to have little slip, to distinguish between competing ideas for the origins of the lineations. In the area of asymmetric aftershock distributions, the dataset is being extended to smaller mainshock sizes using the Parkfield HRSN catalog, the relocation code (with time-dependent station corrections) is being modified to routinely determine the local across-fault P-wave and S-wave velocity contrasts, and numerical modeling of rupture on the interface between dissimilar materials is being undertaken to test the explanation proposed for the origin of the symmetry. Finally, the statistics of aftershock sequences of microearthquakes on different faults are being compared to the predictions of analytic models of rate-and-state friction, and to numerical models of rupture interaction on planar faults. The goals are to test models of rate-and-state fault friction and to determine differences in loading rates and/or surface properties of the various faults. Because for some faults the earliest portion of the aftershock sequences might have been "lost" because of the network "blind" time following a triggering event (perhaps a few tens of seconds), the nearly 100 s archived waveforms of all cataloged events are being processed to identify and locate events that would ordinarily have triggered the network. These studies will lead to further insights into the mechanics of earthquakes.