3-COMPONENT DATA AS AN AID TO SEISMIC EVENT DETECTION AND ASSOCIATION - A CASE-STUDY USING DATA FROM THE SIL (ICELAND) NETWORK

A case study of the analysis of single-station three-component (3C) local earthquake seismological data from the eight station ''SIL'' array on Iceland is presented. The analysis applied gives estimates of arrival time and the direction of propagation of P-waves. The data are interesting because automatic 3C analysis at each site already has been conducted as part of the (newly set up) data-collection system. Reported results from this automatic analysis are of variable quality. Here we examine if this represents an absolute limitation of 3C analysis, or whether results can be improved by improving or ''tuning'' the in-field 3C algorithm. It is concluded that for almost all of the data examined the reported problems were the result of minor inadequacies in the SIL software, primarily inappropriate choice of various processing parameters. Some more fundamental problems also are encountered; primarily when the P-wave arrives near the vertical (giving little energy on the horizontal components), and when scattering effects resulting from complex structures near the recording site are severe. However, these effects are identified readily and their implications are minor for the practical use of 3C analysis algorithms in event detection and association. After exclusion of all data from one station, it was determined that using a broad, fixed frequency band (3-25 Hz) and 31 azimuths, no estimate was in error relative to the ''true'' geometrical azimuth by more than 55-degrees, and only three were in error by more than 35-degrees. The mean and r.m.s. misfits were 17 and 22-degrees respectively. It is recommended that the implementation of routine in-field 3C analysis be considered for all new and existing local and regional networks, especially where the network consists of a larger number of stations and the association of a large number of events detected by individual stations becomes a major logistical problem.