This is the authors' abstract. We don't add key points for this paper.
Efficient and high-quality processing and visualization of Distributed Acoustic Sensing (DAS) data are essential for human and computer interpretation. Due to the large volume of data in its raw format, the information contained within DAS data are often extracted in predefined data features for practical analysis. Spectral features, those computed from data after a Discrete Fourier Transform (DFT), such as Frequency Band Energy (FBE), are commonly used in interpretation. Harmonizing feature data from different interrogator units (IUs) while maintaining information content is key for machine-learning applications and end-user interpretation. This work demonstrates that this can be achieved by using textbook time-series signal processing for spectral features - along with a novel data-driven method for estimating depth and time-varying noise in DAS data. Several data examples from the in-well environment are presented to illustrate how good practice impacts the result. It is shown that DAS data contains additive noise which have quasi-static Gaussian properties when looking at data in terms of strain (not strain rate). When subtracting the predicted noise from the power spectral densities (PSD) prior to estimating spectral features, we remove the variations in energy level over depth typically observed. We correlate these variations in noise to changes in temperature and mechanical strain of the fiber as observed in low-frequency (LF) DAS, and speculate that this is due to effectively changing the back scattering intensity pattern and thus the fiber as seen by the IU. Data from two different commercial DAS interrogators are compared, and it is shown that feature data from these two boxes are comparable after the suggested processing flow.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Haavik et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: