Denoising DAS data in urban volcanic areas through a Deep Learning Approach
- 1Istituto Nazionale di Geofisica e Vulcanologia, Osservatorio Etneo, Piazza Roma 2, Catania, Italy (martina.allegra@ingv.it)
- 2Department of Electrical, Electronic and Computer Engineering, University of Catania, Viale 9 Andrea Doria, 6, Catania, Italy
- 3GFZ German Research Centre for Geosciences, Einsteinstrasse 42-46, Potsdam 14473, Germany
The notable benefits of Distributed Acoustic Sensing (DAS) technology—high coverage, high resolution, low cost—have led to its widespread application in the geophysical domain for high-quality data recording. Among possible applications, the ability to interrogate telecommunication cables has enabled the detection of a variety of seismic-volcanic events in poorly instrumented environments, such as densely populated urban areas.
Nevertheless, the sensing of commercial fiber optic cables has to deal with the presence of anthropogenic noise that frequently corrupts the seismic signal. Indeed, vibrations induced directly or indirectly by anthropogenic activities significantly reduce the signal-to-noise ratio by masking target events.
Taking advantage of the high spatiotemporal resolution of the DAS data, a deep learning approach has been adopted for noise removal. The architecture of the neural network together with the training strategy have enabled the extraction and preservation of salient information while neglecting anthropogenic noise.
The validation on real low-frequency seismic events recorded during the 2021 Vulcano Island unrest has provided encouraging results, demonstrating the potential of the proposed approach as a pre-processing step to facilitate subsequent DAS signal analysis.
How to cite: Allegra, M., Cannavo, F., Corsaro, M., Currenti, G., Jousset, P., Palazzo, S., Prestifilippo, M., and Spampinato, C.: Denoising DAS data in urban volcanic areas through a Deep Learning Approach, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-10925, https://doi.org/10.5194/egusphere-egu24-10925, 2024.