Читать книгу Statistical Methods and Modeling of Seismogenesis - Eleftheria Papadimitriou - Страница 21

Earthquake simulators were initially based on the elastic rebound theory of earthquakes, introduced by Reid (1910) in his study of the 1906 San Francisco earthquake. Starting from the pioneering work of Burridge and Knopoff (1967), the earthquake process has been simulated by a slider-block model in which each spatially coarse grained site on a fault is represented by a block sliding on a frictional surface. Physically, the blocks are intended to represent the sticking points, or asperities, on the fault surface (Rundle et al. 2002). A slider-block model becomes more and more detailed as the number of blocks is increased (Figure 2.1). While the original slider block model (Burridge and Knopoff 1967) specified massive blocks with inertia, more recent models are commonly based on the stochastic cellular automaton type (Rundle and Jackson 1977; Rundle and Brown 1991). In the last 20 years, the increase in computer power has allowed the development of physics-based earthquake simulators based on very complex and realistic fault models, containing thousands of blocks (also named “patches” or “cells” by different authors). They can generate synthetic earthquake catalogs containing hundreds of thousands of events on a time span of hundreds of thousands of years and in a wide magnitude range. The earthquake simulators differ in the type of methodology developed within them and for the geometry type of the patch used in the definition of the topology of a complex fault.

Of course, even if simulators have become more and more complex, they remain models approximating the infinitely more complex physical reality of the earthquake process with many limitations. As a result, not all conclusions based on this kind of models can be tested in any detail against actual earthquakes. Nevertheless, models can be useful in developing hypotheses to explain earthquake observations, such as well-known statistical relationships like magnitude-frequency distributions, temporal relationships like the Omori law and some properties of earthquake clustering. In this respect, there is a general agreement on the usefulness of physics-based earthquake simulators for improving overall testing procedures of earthquake forecasting (for example Schultz et al. 2015; Christophersen et al. 2017; Field 2019).