Automatic Classification with an Autoencoder of Seismic Signals by Using Distributed Acoustic Sensing
Abstract
This study aims to probe the association between enhanced geothermal systems (EGS) and certain kinds of seismicity that may result from hydraulic fracturing occurring at depth using unsupervised machine learning (ML). In April and May 2019, a distributed acoustic sensing (DAS) borehole array at the Frontier Observatory for Research in Geothermal Energy site near Milford, Utah recorded seismic data during hydraulic injection stimulation of a nearby well. Using an autoencoder, a type of deep neural network, we reduce the dimensionality of spectrograms of the detected signals to a lower-dimensional latent feature space. Next, Gaussian mixture model clustering is performed on the latent feature space for improved clustering efficiency and performance, assigning each detected signal to one of 11 classes. For each signal class, we examine spatiotemporal distributions of the clustering results and find that most occur between 450 and 850 m, suggesting that the borehole measurements may be negatively impacted by surface noise. In the temporal distribution, clustering results show how certain kinds of signals may be associated with injection-related activities. Using clustering, we demonstrate the ability to discern not just when and where a signal is detected, but what kind, thus enabling more rapid and targeted data exploration.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFM.S11B..04C