A Seismic Shift in Scalable Acquisition Demands New Processing: Fiber-Optic Seismic Signal Retrieval in Urban Areas with Unsupervised Learning for Coherent Noise Removal

With the development of fiber-optic seismic acquisition systems, dense seismic monitoring of the near surface in urban areas is quickly becoming much easier than ever before. We provide a case study illustrating the use of data from a new type of deployment, a fiber-optic array in existing telecommu...

Full description

Saved in:
Bibliographic Details
Published in:IEEE signal processing magazine 2018-03, Vol.35 (2), p.31-40
Main Authors: Martin, Eileen R., Huot, Fantine, Yinbin Ma, Cieplicki, Robert, Cole, Steve, Karrenbach, Martin, Biondi, Biondo L.
Format: Magazinearticle
Language:English
Subjects:
Cluster analysis
Colleges & universities
Correlation analysis
Data analysis
Data mining
Fiber optics
Filtration
Monitoring
Noise
Optical fiber cables
Optical fiber sensors
Optical fiber theory
Optical fibers
Seismic measurements
Signal processing
Strain rate
Telecommunications
Unsupervised learning
Urban areas
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:With the development of fiber-optic seismic acquisition systems, dense seismic monitoring of the near surface in urban areas is quickly becoming much easier than ever before. We provide a case study illustrating the use of data from a new type of deployment, a fiber-optic array in existing telecommunications conduits underneath the Stanford University campus in California. We perform cross correlations of strain-rate measurements of the ambient wavefield to extract signals that mimic the response of the array to virtual active seismic sources at any receiver location. Now that we can so easily use our telecommunications infrastructure for continuous, dense, urban seismic acquisition, data collected in such a manner will go to waste unless we significantly automate ambient noise processing. We analyze 37 days of ambient noise; introduce an exploratory data analysis tool that clusters a week of noise to help us quickly find coherent, repeating noises that inhibit reliable extraction of useful signals; and design filters to remove nearby car recordings from the next 30 days of noise. We review a metric to quantify convergence of the cross correlations and use that metric extended throughout the array to support filtering out recordings of cars near the array.
ISSN:1053-5888
1558-0792
DOI:10.1109/MSP.2017.2783381