Locating events with a sparse network of regional arrays

Abstract An automated procedure for locating regional seismic events with a network including arrays and single element seismometers is described. The method incorporates backazimuth estimates, arrival-time data, and associated uncertainties into a least-squares-inverse location algorithm. The formulation is essentially that of Jordan and Sverdrup extended to incorporate azimuth data. This technique allows the use of both a priori and a posteriori information about data uncertainties to compute confidence ellipsoids for location estimates. This is important for obtaining realistic confidence ellipsoids for solutions based on few data. It also permits a refinement of the confidence ellipsoid calculations as experience accumulates for events in a particular area. Small arrays like NORESS in Norway provide accurate estimates for the backazimuth of regional phases. These azimuth data provide a strong constraint on the location of events detected by a small number of stations. The strength of the constraint depends on the geometry, and in some situations azimuth data are as important as arrival-time data. Synthetic examples illustrating this are given for a network including two arrays. Actual data from the NORESS array in southern Norway and the FINESA array near Helsinki, Finland, are presented to demonstrate the use of the location technique. Most of the events studied are mine blasts for which there are highly accurate, independent locations. Comparing one-array and two-array locations and their confidence ellipses with these independent locations provides a preliminary validation of our estimates of the (signal-dependent) a priori arrival-time and azimuth variances, and demonstrates the effectiveness of our location procedure for a sparse network of regional arrays.