Real Time Forecasts through an Earthquake Clustering Model Constrained by the Rate-and-State Constitutive Law: Comparison with a Purely Stochastic ETAS Model

• DOI: 10.1785/gssrl.78.1.49
• Publication Date: 2007-12-13T22:24:39Z
• Authors: R. Console; M. Murru; F. Catalli; G. Falcone
• Keywords: 16. Peace & justice; 01 natural sciences; 0105 earth and related environmental sciences

We propose an earthquake clustering model based on the popular concept of epidemic models. In these models every earthquake can be regarded as both triggered by previous events and as a potential triggering event for subsequent earthquakes (Ogata 1988, 1998; Ogata and Zhuang 2006 and reference therein; Console and Murru 2001; Console et al. 2003; Console et al. 2006a, 2006b; Helmstetter and Sornette 2002a, 2002b, 2003 for reviews; and Vere-Jones 2006 for review on the use of stochastic models for earthquake occurrence). The occurrence-rate density at any time and geographical location is computed by the contribution of every previous event using a kernel function that takes into proper account: (a) the magnitude of the triggering earthquake, (b) the spatial distance from the triggering event, and (c) the time interval between the triggering event and the instant considered for the computation. The magnitude distribution adopted here is the Gutenberg-Richter law (Gutenberg and Richter 1944). The above-mentioned criteria are implemented through the introduction of the rate-and-state constitutive law in a previously existing epidemic algorithm. The validity of the model can be tested in an exercise of real-time forecast. Stochastic short-term models describing the phenomenon of earthquake clustering are achieving increasing success in the seismological community ( e.g., Helmstetter et al. 2006). Progress is also being made with models that link stress changes to seismicity rate changes using the Dieterich rate-and-state model (Ruina 1983; Dieterich 1986, 1992, 1994; Console et al. 2006a). In any such model, earthquakes are regarded as the realization of a point process modeled by a generalized Poisson distribution. Each event is characterized by its location-time-magnitude parameters ( x, y, z, t, m ). Depth is not used in our analysis considering its limited range (<30 km) in comparison with the …