Exploring the application of Characteristic Functions on DAS data and their influence in event detection performance.

Distributed Acoustic Sensing (DAS) has emerged as a powerful tool in seismological applications, transforming fiber-optic cables into dense arrays of geophones that can continuously sample seismic wavefields across several kilometers. DAS data acquisition presents a versatile approach, utilizing either ad hoc installations with specific cables or leveraging existing telecommunication optical fiber-network infrastructure. Its adaptability makes DAS particularly advantageous for seismic monitoring in logistically challenging environments like volcanoes or offshore areas, where traditional seismometers may face limitations.

 

Conventional seismological techniques struggle to effectively process DAS data due to its unique characteristics—typically, wavefields are sampled at 1 m spacing with frequencies exceeding 1 kHz. As a result, this technology provides a detailed mapping of the seismic wavefield across the length of the fiber, and it also generates a significant amount of data compared to the sparse seismometer installations. In order to efficiently analyze these data, we introduced HECTOR, a waveform-based detection method designed specifically for DAS data (Porras et al. 2024).

 

In this study, we investigate the capabilities of HECTOR following preprocessing of DAS data using various characteristic functions (CF). We explore non-negative functions, including Short Term Average to Long Term Average (STALTA), Energy, and Envelope, whose peculiarity is to preserve noise. Conversely, zero-mean characteristic functions such as Short Term Average to Long Term Average derivative (STALTA derivative), Kurtosis, and Kurtosis derivative enhance signals and mitigate noise. Our objective is to assess HECTOR's performance when analyzing preprocessed data compared to raw data.

 

To validate our findings, we initially test the detector on synthetic data. These simulations encompass diverse optical fiber geometries, source configurations, and locations. Subsequently, we apply the algorithm to real data collected in two distinct scenarios. The first scenario involves the FORGE experiment situated in Utah, US, which entails a borehole installation of 1 km optical fiber deployed above a geothermal reservoir characterized by induced seismic activity. The second scenario involves a 90 km horizontal optical fiber deployed in the Pyrenees region. The area is characterized by natural earthquake activity with magnitudes (2.01≥ML≥0.02), alongside anthropic events due to quarry blasts. 

Our evaluation focuses on quantifying the enhancement in HECTOR's performance following the application of CFs compared to analyzing raw data.

Through this comprehensive exploration, we aim to advance the understanding of DAS data processing, demonstrating the efficacy of HECTOR across diverse scenarios. 

We would like to thank TotalEnergies for sharing this data set with us as well as Febus Optic for providing the DAS interrogator used for the data acquisition.

 

References: 

A Semblance-based Microseismic Event Detector for DAS Data.

• Porras, D. Pecci, G. Bocchini, S. Gaviano, M. De Solda, K. Tuinstra, F. Lanza, A. Tognarelli, E. Stucchi, F. Grigoli. Geophysical Journal International (GJI) 2024 (Accepted)