A Neural Network Based Multi-Component Earthquake Detection Method

Date: 4/24/2019

Time: 06:00 PM

Room: Grand Ballroom

We present a method to detect earthquake events from continuous 1-D seismic data. Many traditional earthquake detection methods detect events by the use of amplitude threshold and similarity in data. Currently, machine learning based earthquake detection methods have shown promising detection results. Among all those machine learning techniques, neural networks have been a used as popular method. However, most of the existing neural network frameworks for earthquake detection are based on one component seismic data. In our work, we propose an end-to-end framework which take the full advantage of the three components from the seismic data. A series of experiments are conducted to validate the effectiveness of our proposed model. The classification results and visualization results validate that our framework can significantly improve the detection accuracy.

Presenting Author: Youzuo Lin

Authors