abstract
For a signal-to-noise ratio between 0.2 and 0.1 on the original single-component records, amplitudes for Rayleigh waves over oceanic paths of 155° at station MAT and 98° at station KIP have been determined as 12 mμ and 24 mμ peak-to-peak, respectively, with a standard error of less than 11 per cent. In each case the processed correlation signal is the highest in a half-hour record. The method makes use of preliminary high-pass filtering and normalized reference earthquake-matched filtering, and takes full advantage of the well-dispersed oceanic surface wave.
The method also provides high resolution of co-located events with short time separation, or of widely spaced events with Rayleigh waves arriving nearly simultaneously at a single station, when the summed vertical and radial matched filtered components are used. Examples include: (1) clear separation and amplitude determination at stations KIP and MAT of two MS = 6.5 earthquakes located 0.7° and 145 sec apart off the coast of central Chile; (2) clear separation at station KIP of a Novaya Zemlya mb = 4.8 event from interfering Rayleigh waves of an mb = 5.0 Kermadec Island earthquake arriving 120 to 140 sec prior to the searched event, with almost complete elimination of interference on the summed vertical and radial processed components; and (3) clear separation at station KIP of two co-located mb = 4.4 and 4.5 earthquakes 6 min apart off the coast of Chile, with determination of their amplitudes in the presence of interfering Rayleigh waves from two central Alaska earthquakes, the first (mb = 4.1) arriving 15 min prior to the first Chile Rayleigh wave and the second between the two Chile arrivals.
The single-station threshold reached (10 and 25 digital units, p-p) for stations MAT and KIP at 155° and 98°, respectively, corresponds to an MS = 3.3 and probably can be improved further.
Fast Determination of Earthquake Depth Using Seismic Records of a Single Station, Implementing Machine Learning Techniques
