Seismologic data are also essential for testing hypotheses regarding earthquake clustering, including foreshocks and aftershocks. Modeling seismicity on a fault network relies on accurate and complete seismic catalogs for the recognition of regional patterns in seismicity and for detailed studies of specific earthquake sequences. Upgrading the regional networks proposed as part of the ANSS will greatly facilitate seismological studies of fault systems in the United States. These instrumental improvements will enhance earthquake information and encourage the development of new seismological products, such as the cataloging of fault planes, rupture lengths, and slip propagation directions for moderate-size earthquakes.
Better structural data are needed on fault segmentation, along with an improved mechanical understanding of the role of segmentation in fault rupture. Work on fault-zone complexity suggests fundamental differences in behavior between mature and immature faults. If segment boundaries play a key role in the termination of ruptures, then highly segmented faults may tend to be more characteristic in their behavior or at least more predictable in the lateral extent of future ruptures. A long, smooth fault such as the San Andreas may not have any “hard” segment boundaries, making the size of ruptures more sensitive to time-dependent stress heterogeneities.
Changes in Coulomb stress parameters resulting from large earthquakes have been invoked as a quasi-static mechanism for modifying seismicity rates. If such dynamical interactions significantly affect the timing of earthquakes on nearby faults, they must be accounted for in any model of earthquake recurrence. An important issue is the role of time-dependent phenomena, such as transient fault creep, fault healing, poroelasticity, and viscoelasticity, in stress transfer. The best constraints on these processes come from near-fault deformations following large earthquakes, which can be measured using GPS and InSAR geodesy.
A true understanding of fault-system dynamics will be reached only by integrating the disparate observations of stress, strain, and rheology into self-consistent models that can be tested against observations not yet collected. Simulations of earthquake occurrence on fault networks are required to understand the behavior of natural fault systems and to address fundamental questions relating to earthquake occurrence, such as the effects of stress interactions and prior earthquakes in determining earthquake probability (Figure 6.3). The practical objective of this research is to develop procedures that can assimilate all information on fault-system behaviors into probabilistic forecasts and update these forecasts consistently based on seismic activity and other new information. A proper interpretation of fault systems must begin with a detailed representation of the active structural elements; in particular, it will be necessary to