An approach to automatic location of regional events

Abstract

The conventional network location algorithms for automatic processing are based on P and S first arrivals. These procedures have the following shortcomings: (1) it is often difficult to identify the proper phase: (Pg or Pn, Sg or Sn); and (2) first arrivals are often masked by noise. Both factors may cause significant location errors. An alternative is to look for maximums of seismic energy time–distance distribution, which are less sensitive to these factors. We measure the time-of-maximum of P and S waves envelopes vs. distance for each available station of the Israel Seismic Network (ISN), thus providing a travel time curve (TTC). The record envelopes are obtained using 1D χ 2 optimal detector and 3–6 Hz `short-time-average' time-curves, having enhanced sensitivity for seismic signal arrivals. The corresponding P and S time-of-maximum vs. distance functions are approximated linearly by a least-square method for a set of local earthquakes and quarry blasts. Travel time inversion for location of small events comprises three main steps: (1) cleaning of the records from noise bursts and computation of the envelopes, (2) triggering and identification of the P and S phases and computation of their energy maximums, and (3) maximization of a sum of station residuals as a function of epicenter coordinates. The maximum of the functional is looked for on a 60×60 km grid, step of 2 km, covering different parts of Israel and Jordan. The algorithm is not sensitive to the source depth, thus providing epicenter determination only, but shows to be convenient and robust. As a result of the preliminary study for a set of relatively weak 74 local earthquakes and 58 quarry blasts, M L∼1.5–2.5, we have obtained the accuracy of epicenter estimation ±6 km for 80–90% of both types of events, which is satisfactory for automatic location. The accuracy is measured relative to the ISN bulletin locations for earthquakes and the ground truth information for quarries, respectively.