 |
Center for Nonlinear Studies |
The proper classification of emergent and impulsive noise signals is critical for detection of microearthquakes and developing a complete understanding of ongoing ground motions. Tectonic events occupy a small percentage of each day and seismic records contain numerous natural and anthropogenic signals. In 2014, a dense array of 1,100 vertical geophones recorded ground motions on the San Jacinto fault for 30 days. The data provides detailed waveforms to detect microearthquakes and observe surface/atmospheric processes that manifest as impulsive and emergent signals. Efforts to detect seismic events using a shallow architecture Random Forest model show a 72% increase from the regional catalog. Recent studies have demonstrated that ongoing low-amplitude seismic motion is dominated by various weak sources originating at the surface from anthropogenic and atmosphere interaction. Labeling new classes of waveforms from wind generated ground motions, air-traffic, automobiles, and other non-tectonic signals can provide insightful information when designing a machine learning training data set. We apply a new methodology that uses waveform noise correlations to label continuous waveforms as random noise or non-random noise for training a convolutional neural network. The effort is focused on identifying different classes of non-tectonic signals in the noise using unsupervised learning techniques to subdivide the signals into a new training data set. A classification model is applied to continuous waveforms to develop a time series of additional classes of noise signals. Results of coherent signals across the array provide insight to shallow crustal deformation and surface generated ground motions that can mask microseismic events. Future efforts to monitor temporal variations in crustal properties and modulation of seismic activity in southern California are underway. Four multi-year dense seismic arrays will probe the San Jacinto fault for noise sources, microseismicity, and the response to transient loading. The models being developed now will allow rapid analysis of temporally changing seismic waveforms and provide new insights from a novel data set.
Host: Paul Johnson