HR: 0830h
AN: NG41C-0070    [PDF]
TI: Evaluation and Development of the Pattern Informatics Method of Earthquake Forecasting
AU: * Holliday, J
EM: jholliday@physics.ucdavis.edu
AF: Ctr. Comp. Sci. & Eng., University of California One Shields Ave., Davis, CA 95616 United States
AU: Rundle, J
EM: jbrundle@ucdavis.edu
AF: Ctr. Comp. Sci. & Eng., University of California One Shields Ave., Davis, CA 95616 United States
AU: Tiampo, K
EM: ktiampo@uwo.ca
AF: Dept. of Geology, University of Western Ontario, London, Ont N6A 5B8 Canada
AU: Donnellan, A
EM: andrea.donnellan@jpl.nasa.gov
AF: Earth and Space Science Division of Earth and Space Sciences, Jet Propulsion Laboratory 4800 Oak Grove Drive, Pasadena, CA 91109 United States
AU: Klein, W
EM: klein@buphy.bu.edu
AF: Dept. of Physics, Boston University 590 Commonwealth Ave., Boston, MA 02215 United States
AU: Turcotte, D
EM: turcotte@geology.ucdavis.edu
AF: Dept. of Geology, University of California One Shields Ave., Davis, CA 95616 United States
AB: We report results exploring the Pattern Informatics (PI) method of earthquake forecasting, in which state vectors are first constructed to characterize earthquake activity in seismically active regions, then used to compute a probability index forecasting future activity of large events. This method makes use of a description of the state of earthquake seismicity in terms of a phase-dynamical state vector in a Hilbert space. Using these state vectors, we compute the change in state over a period of time preceding the forecast interval. Probability change is then found by squaring the state vector difference, then subtracting off the average background rate. Here, we examine the influence of varying the amount of historic data included in formulating the forecast, and show that results are increasingly unreliable as the data set used is progressively truncated from below. We also report very preliminary results for other seismically active regions. Finally we discuss future work, including particularly the goal of using complex-valued state vectors constructed via a Hilbert transform instead of the real-valued state vectors currently used.
DE: 3220 Nonlinear dynamics
DE: 3240 Chaos
DE: 7209 Earthquake dynamics and mechanics
DE: 7223 Seismic hazard assessment and prediction
SC: Nonlinear Geophysics [NG]
MN: 2003 Fall Meeting