Earthquake source mechanism analysis using broadband data

Abstract

Information on stresses within the earth can be gained by examining the direction and amount of faulting due to earthquakes in a region, and recorded seismograms can be used to determine this faulting. Techniques which use the entire earthquake waveform to model the faulting have in some overseas cases successfully determined the direction of faulting with only a single seismic station. However, there has been little analysis of single station solutions using New Zealand data. The fewer the number of stations needed for robust solutions, the more earthquakes than can be modelled, and the more that can be learned about the state of stress in the earth.

This study tested the usefulness of a waveform inversion technique using broadband regional data from a limited number of seismic stations in New Zealand. The inversion was tested with a series of scenarios involving synthetic data and artificial error sources.  Earthquakes with known mechanisms were then inverted using the waveform inversion and an amplitude ratio technique, and the results were compared to those determined using a global distribution of stations. The results show that single station solutions are not reliable in the New Zealand seismic environment due to the high level of ocean noise, but solutions using two or three stations are far more likely to agree with solutions determined using a large number of stations.

---

Technical Abstract

Data from the permanent broadband seismic station SNZO, and other temporary broadband deployments have been used to test the operation of the regional, full waveform inversion source mechanism determination technique of Dreger and Helmberger (1991) in New Zealand. An amplitude ratio technique similar to the AMPRAT technique (Robinson and Webb, 1996) which is widely used in New Zealand, was also tested with the same data.

Tests using synthetic data showed that the waveform inversion is not reliable if data from only one station are available, but that the reliability of the technique improves when a second station is added, and the reliability is directly proportional to the azimuthal separation of the stations. Addition of a third station further improves the results.

To test the applicability of these synthetic results to actual studies, all regional earthquakes with published moment tensors since the installation of SNZO were investigated using the waveform inversion technique. Where possible, additional data from temporary broadband deployments were added to the data from SNZO.  The results of these tests on the twelve selected events support the synthetic results. While 33% of the events returned source mechanisms consistent with the published solutions, there was no priori way to separate the consistent from the inconsistent mechanisms, making it impossible to reliably use the technique with data from only one station. Addition of further stations improved the results, and the technique was considered reliable with data from more than one station.

The same events were tested using an amplitude ratio technique plus first motion polarities from the NZ short period network, and similar results were found. The amplitude ratio technique was found to be more reliable than the waveform inversion when only one station (plus first motions) was used, with 58% of the events returning source mechanisms consistent with the published solutions. This was still not sufficiently reliable to guarantee good results.  Addition of further stations again improved the reliability of the results.

In addition to the tests of the techniques, new source mechanisms are presented for the 18 June 1994 Arthur’s Pass and 24 November 1995 Cass earthquakes, and evidence is presented for a preferential amplification of low frequency seismic waves travelling south-west – north-east within New Zealand. This amplification may be caused by the high Q core of the subducted Pacific Plate.