Prediction of large events on a dynamical model of a fault

Abstract

We present results for long term and intermediate term prediction algorithms applied to a simple mechanical model of a fault. We use long term prediction methods based, for example, on the distribution of repeat times between large events to establish a benchmark for predictability in the model. In comparison, intermediate term prediction techniques, analogous to the pattern recognition algorithms CN and M8 introduced and studied by Keilis-Borok et al., are more effective at predicting coming large events. We consider the implications of several different quality functions Q which can be used to optimize the algorithms with respect to features such as space, time, and magnitude windows, and find that our results are not overly sensitive to variations in these algorithm parameters. We also study the intrinsic uncertainities which are associated with seismicity catalogs of restricted lengths.

Predictability of self-organizing systems

“…With A we are able to predict 90% of the large events, with alarms which occupy 15% of the total space-time volume (Fig. 3), which corresponds to alarms which occupy significantly smaller time intervals than the average duration of small scale seismicity prior to large events 20 in Fig. 2.…”

“…[ 20] a detailed study of predictability of the UBK model revealed that among a set of precursors, the two most effective are activity A= # earthquakes, and a new measure which is a better measure of the development of spatial correlations. This new measure, which we call active zone size AZS, is defined to be: AZS = # blocks which have slipped (independent of the number of times) in each space-time window.…”

“…For the purposes of qualitative comparisons, it is sufficient to roughly estimates as we do here. We take: N = 8192, σ = .01, α = 3, and ξ/a = 10 for the UBK model, 20 system sizes 32 × 32 for the other models, and α = .2 for the OFC model. 2 (2).…”

Boolean delay equations: A simple way of looking at complex systems

“…These authors used the simulated catalogs to study in greater detail the performance of pattern recognition methods tested already on observed catalogs and other models [96,107,108,109,110,111,112,113,114], devised new methods, and experimented with combination of different individual premonitory patterns into a collective prediction algorithm.…”

Initiation propagation and termination of elastodynamic ruptures associated with segmentation of faults and shaking hazard

“…While these models do not have the complete set of earthquake behaviors, they are nevertheless rich enough to use in a variety of ways. We can, for example, compare and hone our techniques and pose new questions in a context where limited catalogues and statistics are not an issue [Pepke et al, 1994] before tackling real catalogues [Kossobokov and Carlson, 1995].…”