**Title of research project:** Dynamics of seismicity

**Reference number of the grant:** EAR-9804859

**Period of performance:** 3 year; from 1 June 1998 to 31 May 2001

**Project Coordinator:** Prof. D. Turcotte (Cornell University)

**Principal Investigator**: Prof. V. Keilis-Borok

**Summary: **

The goals of proposed research are: (i) to expand fundamental base of earthquake prediction research; (ii) to develop and test a new generation of earthquake prediction algorithms which combine intermediate-term and short-term prediction; and (iii) to improve seismic risk assessment. In addition to the earthquake catalogs, vast volumes of data on geotectonics and on source mechanisms will be incorporated in the analysis. The project will result in important know-how transfer. Feasibility of the project is demonstrated by pilot studies. We will take advantage of 15-year test of phenomenological intermediate-term prediction algorithms for which high statistical significance has been established, of new mathematical models of dynamics of seismicity, and of newly available data bases. Major parts of the project are the following.

Lattice models of dynamics of seismicity. We propose: (i) To model different possible types of chaotic behavior, besides self-organized criticality, and to explore earthquake precursors for each type. (ii) To analyze the recently established dramatic changes in predictability for simulated seismicity and to search for the way to predict the changes of predictability in a seismic region; to include adaptation to such changes in prediction algorithms. (iii) Using forest fire models, to develop an algorithm for prediction of "safe periods", when a large earthquake will not occur (this problem is not quite equivalent to prediction of earthquake occurrence).

Modeling the dynamics of seismicity by a system of blocks-and-faults with realistic geometry. We propose to use these models as follows: (i) To reproduce geometric incompatibility G as an integral measure of instability of the system. and to incorporate G into earthquake prediction algorithms. (ii) To explore the possibility to use estimations of G for prediction of the birth of new faults. (iii) To explore premonitory phenomena specific to geometry of a fault system.

New earthquake prediction algorithms. We will develop a next generation of prediction algorithms and start testing them by advance prediction. Algorithms will be developed by analysis of synthetic and observed data, taking advantage of unlimited time periods reproducible in models. On the basis of pilot studies we propose to explore at first the following types of premonitory phenomena: (i) Variation of stress release and strain fields generated by seismicity. We will use for this a new methodology which allows to analyze about 5 times more earthquakes than the usual methods do. (ii) Premonitory changes in the spatial distribution of seismicity. (iii) Variation in the correlation distance in the spatial distribution of seismicity. (iv) Region- specific variations of premonitory phenomena. The necessary data have been accumulated in worldwide application of prediction algorithms. Transition to short-term prediction. We propose to search for short-term precursors within the limits of intermediate-term alarms. As potential precursors we will consider at first earthquake clusters and log-periodic behavior. As an alternative to the latter we will consider other patterns of frequency increase, which may better fit the spiked variations of observed seismicity.

Catastrophic intraplate earthquakes in the North American and East European platforms. We propose: (i) To develop a comprehensive geodynamic reconstruction of hierarchical systems of blocks and fault zones for the platforms. (ii) To model the dynamics of these systems, interaction with the mantle included. (iii) To explore the possibility to identify unstable zones prone to large earthquakes; this will be done by pattern recognition applied to models and observations. (iv) To explore phenomena premonitory to large earthquakes.