HR: 0800h
AN: G11A-1196    [Abstracts]
TI: Rapid Event Detection and Discrimination From High Rate GPS Data
AU: * Genrich, J F
EM: jeff@gpsmail.ucsd.edu
AF: Institute of Geophysics and Planetary Physics, Scripps Institution of Oceanography, MS 0225 9500 Gilman Dr., La Jolla, CA 92093-0225 United States
AU: Bock, Y
EM: ybock@ucsd.edu
AF: Institute of Geophysics and Planetary Physics, Scripps Institution of Oceanography, MS 0225 9500 Gilman Dr., La Jolla, CA 92093-0225 United States
AB: Building on existing concepts for seismic data we are developing algorithms for rapid detection and discrimination of potential natural hazard events captured by high rate GPS data streams. As a first step, a time-domain based characteristic function detects single station singularities based on continuously updated noise properties. Detection is treated as a stochastic concept, with an associated probability computed from signal-to-noise ratios and event history parameters. A second step temporospatially correlates singularities within a network, while single station time series analysis classifies the identified singularity. A descriptive (e.g., offset, spike, ramp, or oscillation) single stream classification is predominately based on computed velocity patterns. It provides the basis for a station specific event source discrimination (processing software-, hardware-, communications link-, atmospheric/satellite configuration-, or monument-related) with the help of isotropy considerations and ancillary data (receiver monitoring parameters, satellite visibility and health information, status of the communication link, etc.). Assigned probabilities permit multiple source selection. A third step gathers station specific results to reanalyze them in a network context. This step makes use of a matrix approach to compare the near real time derived event properties with pattern and temporospatial correlation parameters predefined for a set of event types (e.g., near field strong motion, far field seismicity, volcanic activity, or slope failure) in order to arrive at an event identification. A final step computes pertinent quantities (e.g., station distance to epicenter in case of an earthquake) of the event. Initially, we apply the concept to 1 Hz network recordings of the September 28, 2004, Parkfield earthquake.
DE: 1207 Transient deformation (6924, 7230, 7240)
DE: 1241 Satellite geodesy: technical issues (6994, 7969)
DE: 3270 Time series analysis (1872, 4277, 4475)
DE: 7212 Earthquake ground motions and engineering seismology
DE: 8194 Instruments and techniques
SC: Geodesy [G]
MN: Fall Meeting 2005