| 英文摘要: | Non-technical Summary
Earthquakes are caused by the sudden release of stresses along faults. Plate tectonics describes the process of long-term strain accumulation, but the specifics of stress release -- what ultimately leads to fault failure and whether an earthquake (small or large), tremor, or slow slip results -- are still not well understood. The time-varying stress/strain field at the depths where earthquakes begin is one of the most important properties controlling the sequencing and nucleation of these seismic/aseismic events. The importance of deformation transients has been illustrated by the success of predicted stress changes in accounting for the spatial distribution of aftershocks, as well as providing one explanation for the clustering of large seismic events, such as the nearby 1992 Landers and 1999 Hector Mine earthquakes.
The measurement of stress, however, is notoriously difficult, particularly at depth. Geodesy provides important constraints on the surface deformation field, which can be related to stress through an assumed rheology. Yet, the constraints on the depth distribution of stress and strain from geodesy are limited. These surface constraints need to be combined with other techniques that, while not as directly related to stress and strain, have superior depth resolution. The most promising techniques at present appear to be seismic. For example, patterns of seismicity have long been used to make inferences about the stress state before and after seismic events. Yet, a basic requirement for this approach to work is that changes in stress over time generate seismic waves that can be observed at the surface, which means that aseismic changes in stress and/or strain cannot be observed in this manner. One way of assessing this fully aseismic component is through observations of temporal changes in the elastic properties of the crust. Indeed, such earthquake-related changes have long been predicted and sometimes observed, due to stress-induced changes in the characteristics and or distribution of fluid-filled cracks. Yet, it has been difficult to conclusively observe temporal variations in the medium, due both to the small signal level, and to the problem of accounting for other systematic effects that may produce apparent changes, such as variations in source location or shallow environmental influences.
Technical Description
We are conducting a continuous cross-well active-source seismic experiment utilizing the SAFOD (San Andreas Fault Observatory at Depth) pilot and main holes. The broad, long term goal of this experiment is to develop a tool to monitor the time-varying stress field associated with earthquakes and other stress-dependent earth processes, such as aseismic slips and non-volcanic tremors through the detection of temporal changes in the crustal velocity structure at seismogenic depths. This technique would be a type of "stress meter". Such a monitoring system would perhaps be the single most important means of understanding the triggering processes of seismic and aseismic events. The fundamental physics behind a seismic stress meter is well established. Numerous laboratory studies over several decades have shown that seismic velocities clearly exhibit stress dependence, usually attributed to changes in the physical characteristics of cracks (e.g. crack density, crack orientation).
Our current experiment is built on a previous experiment we conducted at the SAFOD site in 2005 and 2006. Over a two-month period, we observed a negative correlation between changes in the time required for a shear wave to travel through the rock between the pilot and main hole (a few microseconds) and variations in barometric pressure (about 1 kilopascal). This result is a "calibration" of the stress sensitivity of seismic velocity at our experiment site. We also observed two large excursions in the travel-time data that are coincident with two earthquakes, a magnitude 3 and a magnitude 1 earthquake, that occurred sufficiently close to produce large coseismic stress changes at the SAFOD site. The two excursions started approximately 10 and 2 hours before the magnitude 3 and 1 earthquakes, respectively, suggesting that they may be related to pre-rupture changes in crack properties, as observed in the early laboratory studies.
In the current experiment, we are using a similar equipment configuration to collect data that sample 10-15 magnitude 2 to 3 local earthquakes. We are using a cross-correlation based method and the coda wave interferometry technique to image systematic changes in medium near the SAFOD site, and use them to monitor temporal changes in fault zone processes near Parkfield, CA. This project has important implications for the study of earthquakes and for the EarthScope program. It will demonstrate whether there are measurable changes in seismic velocity structure near the source region of an impending earthquake immediately preceding the rupture. As such, this work could lead to significant improvement in our understanding on physical processes prior to earthquakes. It will represent significant progress toward measuring stress transients associated with earthquakes and other processes, demonstrating that tectonic stress could be continuously monitored with continuous active source borehole observations. |