Abstracts Online
Calibrating the MKAR ARray Using Transfer Functions
RENWALD, M. D., Southern Arizona Seismic Observatory, Department of Geosciences, University of Arizona, Tucson, AZ 85721, mrenwald@geo.arizona.edu; TAYLOR, S. R., EES-11, Los Alamos National Laboratory, Los Alamos, NM 87545, taylor@lanl.gov; and WALLACE, T. C., Southern Arizona Seismic Observatory, Department of Geosciences, University of Arizona, Tucson, AZ 85721, wallace@geo.arizona.edu.
Developing regional discriminants at any given seismic station requires a ground-truth database of waveforms from both earthquakes and explosions. Recently installed stations lack a complete database, in terms of source type and source-station paths, which severely affects confidence in discrimination. To examine this problem, we investigated a procedure to develop surrogate recording histories for newly installed seismic stations using the recording histories at nearby long-operating stations through transfer functions. We developed a method to compute transfer functions simply using known effective regional discriminants (RD's) for a database of earthquakes and explosions located near the Lop Nor nuclear test site and recorded at the KNET array in Kyrgyzstan. For specific source-station paths, transfer functions work well. However, preliminary analysis of Indian and Pakistani nuclear tests indicate strong azimuthal dependence in the construction of reliable transfer functions. The success of the preliminary work suggests we can apply the same technique to calibrate the recently installed MKAR array using the GSN station MAKZ as a surrogate. Both MKAR, an 11-element array operational since 2000, and MAKZ (including its earlier counterpart MAK), operating very broadband instruments since 1994, are located in eastern Kazakhstan and separated by 25 km. To perform the calibration requires additional considerations not taken into account during the initial investigation: (1) utilizing amplitude spectra, rather than RD's, to calculate transfer functions; (2) computing transfer functions for a range of azimuths, as we believe the effectiveness of the transfer function is azimuthally dependent; and (3) determining whether working with each array element separately or developing a single-input/multiple-output model will provide more stable results and better error estimates.
[Back]