Earthquake backazimuth determination using a single three-component digital seismograph

Earthquake backazimuth estimation using a single three -component digital seismograph has been investigated. A three-dimensional principal components method formed the basic computational technique. Other details of the backazimuth estimation procedure were varied to determine the optimum approach. The data set comprised aftershocks of the 1988 Tennant Creek earthquakes, recorded on a number of stations in, or close to, the source zone. Epicentral distances (3.5 to 25 km) were unusually small compared to the data sets of others who have used similar techniques.

Complex geology challenged the accuracy of results. The most likely causes of error were believed to be the interruption of source -receiver wave travel paths by fault planes, and complex site geology. This included outcrops of highly deformed rocks, and proximity to a large intrusive body of anomalous character. Errors of 180° were common, possibly due to the low velocity surface layers.

Single station estimates were compared to reference backazimuths from network locations, determined by "EQLOCL" (SRC, RMIT), a program based on the least -squares travel -time approach. Approximately 75% of backazimuth discrepancies, projected into the range -90 to +90°, were less than or equal to 20°. This outcome indicates that the investigated technique can be successfully used to estimate earthquake backazimuth in at least some complex geologic receiver settings, and at short epicentral distances.