With the development of fiber optic seismic acquisition, continuous dense seismic monitoring is quickly becoming more accessible than ever before. By deploying fibers in existing telecommunication conduits, it becomes possible to run seismic studies at low cost, even in locations where traditional geophones are not easily deployed, such as in urban areas. However, due to the large volume of continuous dense streaming data, data collected in such a manner will go to waste unless we significantly automate the processing workflow. Using data acquired over three years by a fiber-optic array deployed in the telecommunication conduits under Stanford University campus, we train a convolutional neural network for earthquake detection. We demonstrate that fiber optic systems can effectively complement sparse geophone data to detect small local earthquakes. The convolutional neural network allows for reliable earthquake detection despite low signal-to-noise ratio and varying coupling with the ground.
Presentation Date: Monday, October 12, 2020
Session Start Time: 1:50 PM
Presentation Time: 4:20 PM
Location: 351D
Presentation Type: Oral
