[S21P-07] You Only Search Once: An alternative framework to detect slow earthquakes.
The occurrence of slow earthquakes is a phenomenon of interest due to its direct relation to large megathrust earthquakes in subduction zones around the world. Since the discovery of this type of seismicity by Hirose et al. (1999) and Obara (2002) in Japan Trench and Nankai Trough respectively, active research has been done to understand the source, occurrence, and relation to “fast” seismicity. Specially, the detection of slow earthquakes is a challenging area of research due to the relatively low amplitude of these signals and the specific bandwidths in which they are found. For this reason, many detection and processing methodologies have been developed, and some of these are already being applied with successful results and high accuracy. In this work, we explore and propose an alternative workflow to detect low frequency earthquakes and tectonic tremor. In this methodology we create an initial catalog of events by applying an RMS amplitude-based detector and classify the initial detections using clustering techniques to discriminate between events and noise. The coherent families of events are selected as templates in a template-matching procedure using information from both spectral and temporal domains. At this point, we present the preliminar results of the initial stage of the framework using Morlet wavelets embedded in Gaussian noise and observed waveforms.