Investigating the effects of random data errors on the waveform-based moment tensor inversion

SUMMARY The linear Gauss–Markov model for waveform-based moment tensor inversion often relies on the overdetermined least-squares method. It needs a proper stochastic of observables accurate and precise estimates unknown parameters. Furthermore, estimating level distribution random errors in observed waveforms is challenging due to assessing minimum-variance unbiased estimator (MVUE). Hence, according considerable effects data uncertainty components, this paper aims describe an MVUE covariance matrix its application quantification tensor. used mathematical prescription allows us use three-component noise records at every station all possible cross-correlations among recorded wavefield. To illustrate proposed method’s performance, we conducted tests with synthetic using configuration 2018 Mw 6.8 Zakynthos (Ionian Sea, Greece) earthquake. Both uncorrelated correlated traces were added waveform amounts between 5 20 per cent maximum amplitude. In order test efficiency method, considered three different structures matrix: (i) diagonal (contains variance individual measurements seismic stations), (ii) block-diagonal (considering cross-covariance components each station), (iii) full matrix. Test results are presented by comparison outcomes known levels generated focal mechanisms, ability estimated illustrating minimum parameters (namely, posterior uncertainties), parameters, values also stations. Finally, applied method real earthquake having inferred mechanism strike/dip/rake angles 13/40/171 (deg) 33 double couple (DC) −61 compensated vector dipole component (CLVD). solution has 19/34/177 69 DC −23 CLVD when our