GEM Plate Boundary Simulations for the Plate Boundary Observatory: A Program for Understanding the Physics of Earthquakes on Complex Fault Networks via Observations, Theory and Numerical Simulation

Abstract

The last five years have seen unprecedented growth in the amount and quality of geodetic data collected to characterize crustal deformation in earthquake-prone areas such as California and Japan. The installation of the Southern California Integrated Geodetic Network (SCIGN) and the Bay Area Regional Deformation (BARD) network are two examples. As part of the recently proposed Earthscope NSF/GEO/EAR/MRE initiative, the Plate Boundary Observatory (PBO) plans to place more than a thousand GPS, strainmeters, and deformation sensors along the active plate boundary of the western coast of the United States, Mexico and Canada (http://www.earthscope.org/pbo.com.html). The scientific goals of PBO include understanding how tectonic plates interact, together with an emphasis on understanding the physics of earthquakes. However, the problem of understanding the physics of earthquakes on complex fault networks through observations alone is complicated by our inability to study the problem in a manner familiar to laboratory scientists, by means of controlled, fully reproducible experiments. We have therefore been motivated to construct a numerical simulation technology that will allow us to study earthquake physics via numerical experiments. To be considered successful, the simulations must not only produce observables that are maximally similar to those seen by the PBO and other observing programs, but in addition the simulations must provide dynamical predictions that can be falsified by means of observations on the real fault networks. In general, the dynamical behavior of earthquakes on complex fault networks is a result of the interplay between the geometric structure of the fault network and the physics of the frictional sliding process. In constructing numerical simulations of a complex fault network, we will need to solve a variety of problems, including the development of analysis techniques (also called data mining), data assimilation, space-time pattern definition and analysis, and visualization needs. Using simulations of the network of the major strike-slip faults in southern California, we present a preliminary description of our methods and results, and comment upon the relative roles of fault network geometry and frictional sliding in determining the important dynamical modes of the system.