Models for Broad Area Event Identification and Yield Estimation: Multiple Coda Types.

摘要

Models of laterally varying attenuation and site effects of regional phases and their codas are necessary to correct regional amplitudes to reveal source effects of importance to identification and yield estimation procedures. We accomplish this using 2-D amplitude tomography, incorporating absolute spectral estimates based on precise ratio techniques and moment tensor results. The spectral constraints allow calibration to absolute levels, and can be used to validate amplitude prediction models. Constraints may also be used to improve resolution in the margins of ray path coverage where source-path tradeoffs would otherwise dominate. Our recent efforts have focused on improving coda analysis in areas of low Q and blockage. We must be able to use whatever coda type appears in our records if we are to estimate Mw for MDAC purposes, or use in yield estimation studies. The Lg coda has been used the most in such studies; however low crustal Q can strip the Lg and its coda from the record, especially at long distances and high frequencies. We often observe high quality Pn and Sn coda in such cases. Pg coda is observed for close in recordings, and behaves similarly to the Lg, with strength in lower bands, but generally has limited application outside local distance ranges. We have examined coda envelopes in plate boundary regions of Asia that are known to severely attenuate Lg, picking Pn, Pg, Sn, and Lg coda manually. We then processed coda using standard methods before turning amplitude data over to a simultaneous inversion of multi-coda type and multi-band data for 2-D path effects, site effects, and the absolute transfer function. The multiple coda types improved the bandwidths of observed seismic spectra as Lg covered low frequencies, and Pn and Sn coda covered overlapping high frequency bands, with the Pn band slightly higher than the Sn band. Distances were too great to include Pg coda in this study.