Theoretical earthquake nucleation, with applications to creep fronts, tremor, and slow slip

摘要本奖项由2009年美国复苏与再投资法案（公法111-5）资助。传统上，断层被认为以两种方式中的一种来适应滑动：粘滑运动，在这种运动中，断层基本上被长期锁定，被短暂的快速滑动（地震）打断，以及以板块构造速度（每年厘米）或更少的稳定滑动。在过去的十年里，在许多俯冲带中发现了断断续续的缓慢滑动和相关的震颤，以及沿着圣安德烈亚斯断层的震颤，表明断层的行为更加多样化。在幕式缓慢滑动中，断层的滑动速度比板块构造大1到2个数量级，每隔1年左右的准规则间隔。由于滑动事件持续数周，并延伸到数十到数百公里的区域，它们释放的能量相当于6级或以上的地震。震颤是一种准连续的地震信号，被认为是由无数类似于1.5级或更小的常规地震的小事件组成，只是更缓慢。由于慢滑事件增加了在9级地震中可能滑动的断层的浅锁部分的应力率，因此了解慢滑事件可能有助于减少地震危害。此外，从这项工作中产生的对滑动定位的进一步了解可以用于解释正在进行的监测加利福尼亚和南非几公里深度地震成核的实验。在过去的几年里，PI与Jean-Paul Ampuero合作，了解可变形断层地震成核的标准速率和状态摩擦方程的含义。他们推导出的解析解对于理解复杂的数值模拟非常有用。这个建议是将这项工作扩展到对俯冲带和沿着圣安德烈亚斯断层的慢滑和震颤的解释，同时使用数值和分析技术。慢滑和震颤只是滑移在非均质界面上的不同表现，这一共识正在形成。扩展的蠕变锋面似乎是缓慢滑动非均质断层的普遍特征，并且在它们碰撞时非常有效地产生地震滑动速度。这两种性质使得蠕变锋作为震源具有吸引力。本研究的第一个目标是解决这样的蠕变锋在孤立的情况下可能传播多远，以及两个碰撞锋产生地震滑动的必要条件。这既适用于标准速率和状态摩擦，也适用于其他针对慢滑移提出的本构律。一旦爬行锋发生碰撞，它们就会产生具有矩率历史的地震事件，这取决于潜在的本构律，与地震学家所期望的大不相同。这将首先使用？辐射阻尼吗?近似弹性动力学和最终完全弹性动力学。由于目前许多信息来自于对潮汐和远震触发的地震的观测，因此将对这两种触发形式进行建模。

ABSTRACTThis award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law111-5).Traditionally, faults were thought to accommodate slip in one of two ways: Stick-slip motion,in which long periods where the fault is essentially locked are punctuated by brief episodes ofrapid slip (earthquakes), and steady creep at plate tectonic rates (centimeters per year) or less. Thediscoveries within the last decade of episodic slow slip and associated tremor within numeroussubduction zones, and tremor along the San Andreas fault, show that fault behavior is much morevaried. In episodic slow slip, faults slip at rates 1 to 2 orders of magnitude larger than plate tectonicrates at quasi-regular intervals on the order of 1 year. Because the slip episodes last for severalweeks and extend over areas tens to hundreds of kilometers across, they release energy equivalentto earthquakes of magnitude 6 or more. Tremor is a quasi-continuous seismic signal thought tobe made up of myriad small events similar to regular earthquakes of magnitude 1.5 or less exceptin being more sluggish. Because slow slip events increase the stressing rate on the shallowerlocked portion of faults that can slip in magnitude 9 earthquakes, understanding them may proveto be useful for earthquake hazards reduction. In addition, the increased understanding of sliplocalization that arises from this work can be used to interpret ongoing experiments monitoringearthquake nucleation at several kilometers depth in California and South Africa.Over the past several years the PI has worked with Jean-Paul Ampuero to understand the implicationsof the standard rate-and-state friction equations for earthquake nucleation on deformablefaults. They derived analytic solutions that have been very useful for understanding complex numericalsimulations. This proposal is to extend this work to the interpretation of slow slip andtremor in subduction zones and along the San Andreas fault, using both numerical and analyticaltechniques.A consensus is emerging that slow slip and tremor are just different manifestations of slip ona heterogeneous interface. Propagating creep fronts seem to be a ubiquitous feature of slow slipon heterogeneous faults, and are very effective at generating seismic slip speeds when they collide.Both properties make creep fronts attractive as tremor sources. The first goal of this study is toaddress how far such creep fronts might propagate in isolation, and the necessary conditions fortwo colliding fronts to generate seismic slip. This will be carried out both for standard rate-andstatefriction and for other constitutive laws that have been proposed for slow slip. Once the creepfronts collide, they can generate seismic events with moment rate histories that, depending upon theunderlying constitutive law, are quite different from those that seismologists have come to expect.This will be studied initially using the ?radiation damping? approximation to elastodynamics andultimately full elastodynamics. Because much information is currently coming from observationsof tidally- and teleseismically-triggered tremor, both forms of triggering will be modeled.