Seismic detection with distributed acoustic sensors using a convolutional neural network in the frequency wavenumber spectrum

Abstract

With the development of optical fiber distributed acoustic sensors (DAS), their application to seismic observation has become popular. We conducted DAS measurements from November 19 to December 2, 2019, using dark fiber of an ocean bottom cable seismic and tsunami observation system off the Sanriku coast in northeastern Japan and investigated seismic detection methods from the obtained strain rate data. We examined a new seismic detection method using a convolutional neural network, to the best of our knowledge, treating a frequency wavenumber spectrum of strain rate as an image. This method effectively captured a characteristic wave described as the T-phase in a sound fixing and ranging channel even with low signal-to-noise ratio data.