The Influence of Instrument Response and Site Response on Moment-Tensor Inversions of Aftershocks of the 2011 Mw 5.8 Mineral, VA Earthquake

Monday, 15 December 2014
John Ebel and Stephen G. Hilfiker, Boston College, Chestnut Hill, MA, United States
There is a rich set of aftershock data available from portable seismic instruments that were deployed following the August 23, 2011 Mw 5.8 Mineral, VA earthquake. Amplitudes of the P and S waveforms for recorded aftershocks can be used to invert for the moment tensors of the individual aftershocks, allowing for determinations of their focal mechanisms and seismic moments. In testing the moment-tensor inversion method with some of the aftershocks, some problems were noted with the data. The most serious problem is the discovery of acausal filter responses of the instruments, which act to obscure the first arrivals of the P and S waves. An allpass filter to correct for this acausal response helps clarify the direction of the first P and S arrivals but at the expense of distorting the subsequent P and S pulse shape. Thus, a hybrid method where the first arrival is read from filtered waveforms but the amplitude is read from unfiltered waveforms must be employed. A second problem is that some stations appear to have anomalously high or low amplitudes for all events that were tested. This likely due to amplification of the ground motions by the surficial materials beneath the station. Average station corrections are computed to account for this site bias. A third problem appears when focal mechanisms computed using P and SH amplitudes are compared to those computed with only P wave amplitudes. The discrepancies between the two different solutions may reflect improper instrument orientations, scattered energy affecting the SH waveforms, or the use of an improper velocity model. This problem is addressed for each event analyzed by initially computing a moment tensor inversion using only P amplitudes and then later testing how the solution is affected by including SH amplitude data.