QR-Based Coherence Analysis for Seismic Signal Detection in Dense Arrays

Summary Emerging sensing technologies across the physical sciences and engineering, including arrays of many sensors, have revolutionized the ability to collect high-resolution, large-scale datasets continuously. Although these advances have opened up new possibilities for scientific discovery, public data archives of some communities struggle to accommodate the sheer volume of larger datasets, and research teams face growing difficulties in efficiently exploring, transferring, visualizing, and interpreting the data at scale. These challenges are particularly pressing in the geophysics and seismology communities, which have encountered data that are several orders of magnitude larger in recent years. Array coherence, which relies on the similarity between sensor pairs within an array to analyze coherent signals, is effective in enhancing the detection of weak or emergent signals in noisy environments. In this study, we show how array coherence can be leveraged to effectively manage a surplus of data and serve as a robust method for signal detection and identification of regions of interest for further analysis. We present a novel algorithm that employs a QR decomposition to efficiently approximate array coherence and enhance the time resolution of detected signals, providing more precise signal detection. We demonstrate the utility and efficiency of this algorithm on data from a continuously recording, dense seismic array, allowing us to identify candidate seismic events and reduce the amount of data that require further transfer and analysis.