Robust Bayesian estimator for S-wave spectra, using a combined empirical Green’s function

Summary We propose a new fully automatic and robust Bayesian method to estimate precise reliable model parameters describing the observed S-wave spectra. All spectra associated with each event are modelled jointly, using t-distribution as likelihood function together informative prior distributions for increased robustness against outliers extreme values. The includes noise combined empirical Green’s function. It captures source-, receiver-, path-dependent terms in description of by combining physical source attenuation spatially event-size dependent compensation. proposed propagates estimation uncertainties along entire processing chain starting from hypocentre location delivers uncertainty estimands. objective is automatically provide valid descriptions generated an earthquake noisy heterogeneous environment. efficiency tested synthetic seismograms, calibrated cross-validated 31 640 mining induced seismic events iron ore mine (in north Sweden) comprehensive network. evaluated both posterior predictive checks residual analysis we found no evidence that indicates any deficiencies respect central tendency, dispersion, trends.