Classifying emergent and impulsive signals in continuous seismic waveforms
Abstract
Proper classification of different emergent and impulsive noise signals is critical for detection of microearthquakes and developing an improved understanding of ongoing weak ground motions. Tectonic events occupy about 1% of recorded seismic records and the remainder consists of various other natural and anthropogenic signals. Continuous waveforms recorded in 2014 by a dense array of 1,100 vertical geophones on the San Jacinto fault for 30 days provide detailed data to detect microearthquakes and other sources of impulsive and emergent signals. Recent studies have demonstrated that ongoing low-amplitude ground motion is dominated by various weak sources originating at the surface from anthropogenic and atmospheric interaction. Developing labels for new classes of waveforms originating from wind shaking obstacles above the surface, air-traffic, automobiles, and other non-tectonic signals provide important information for designing machine learning training data sets. We apply a new methodology that uses noise correlations to label continuous waveforms as random noise or non-random noise for training a convolutional neural network. The non-random noise is subdivided using unsupervised learning to develop a new training data set. Earthquakes, random noise, and multiple classes of non-random noise waveforms are used to train a convolutional neural network that labels the continuous records for the entire array. The results identify tectonic events and different classes of non-tectonic waveforms as coherent signals in the array. This novel approach to classify seismic waveforms provides insight to shallow deformation and surface generated ground motions and improves the detection of genuine microseismic events.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.S52A..07J
- Keywords:
-
- 0555 Neural networks;
- fuzzy logic;
- machine learning;
- COMPUTATIONAL GEOPHYSICS;
- 1910 Data assimilation;
- integration and fusion;
- INFORMATICS;
- 1914 Data mining;
- INFORMATICS;
- 1942 Machine learning;
- INFORMATICS