Slowness vector estimation over large-aperture sparse arrays with the Continuous Wavelet Transform (CWT): application to Ocean Bottom Seismometers

Abstract

This work presents a new methodology designed to estimate the slowness vector in large-aperture sparse Ocean Bottom Seismometer (OBS) arrays. The Continuous Wavelet Transform (CWT) is used to convert the original incoherent traces that span a large array, into coherent impulse functions adapted to the array aperture. Subsequently, these impulse functions are beamformed in the frequency domain to estimate the slowness vector. We compare the performance of this new method with that of an alternative solution, based on the Short-/Long-Term Average algorithm and with a method based on the trace envelope, with the ability to derive a very fast detection and slowness vector estimation of seismic signal arrivals. The new array methodology has been applied to data from an OBS deployment with an aperture of 80 km and an interstation distance of about 40 km, in the vicinity of Cape Saint Vincent (SW Iberia). A set of 17 regional earthquakes with magnitudes 2 < mbLg < 5, has been selected to test the capabilities of detecting and locating regional seismic events with the Cape Saint Vincent OBS Array. We have found that there is a good agreement between the epicentral locations obtained previously by direct search methods and those calculated using the slowness vector estimations resulting from application of the CWT technique. We show that the proposed CWT method can detect seismic signals and estimate the slowness vector from regional earthquakes with high accuracy and robustness under low signal-to-noise ratio conditions. Differences in epicentral distances applying direct search methods and the CWT technique are between 1 and 21 km with an average value of 12 km. The backazimuth differences range from 1° to 7° with an average of 1.5° for the Pwave and ranging from 1° to 10° with an average of 3° for the Swave