<p>Distributed Acoustic Sensing (DAS) is becoming a well-established technology in seismology. For historical and practical reasons, DAS manufacturers usually provide instruments that natively record strain (rate) as the principal measurement. While at first glance strain recordings appear similar to particle motion (displacement, velocity, acceleration) waveforms, not all of the seismological tools developed over the past century (e.g., magnitude estimation, seismic beamforming, etc.) can be readily applied to strain data. Notably, the directional sensitivity of DAS differs from conventional particle motion sensors, and DAS experiences an increased sensitivity to slow waves, often composed of highly scattered waves that are challenging to analyze. To address these issues, several strategies have been already proposed to convert strain rate measurements to particle velocity.</p> <p>Based on a previously proposed mathematical formalism, we stress some fundamental differences between path-integrated strain and conventional displacement measurements. DAS inherently records arc length variation of the cable which is a relative motion measurement along a curvilinear path. We show that if the geometry of the DAS deployment is adapted to the wavefield of interest, path-integrated strain can be used to closely approximate the displacement wavefield without the need of additional instruments. We validate this theoretical result using collocated seismometers, discuss the limitations of this approach, and show two benefits: enhancing direct P-wave arrivals and simplifying the magnitude estimation of seismic events. While using path integrated strain is in some aspects more challenging, it achieves flat (hence lower) noise levels both in frequency and wavenumber. It also provides better sensitivity to high velocity phases, and permits the direct application of conventional seismological tools that are less effective when applied to the original strain data.</p>

Over a period of less than a decade, Distributed Acoustic Sensing (DAS) has become a well-established technology in seismology. For historical and practical reasons, DAS manufacturers usually provide instruments that natively record strain (rate) as the principal measurement. While at first glance strain recordings seem related to ground motion waveforms (displacement, velocity, acceleration), not all the seismological tools developed over the past century (e.g., magnitude estimation, seismic beamforming, etc.) can be readily applied to strain data. Notably, the directional sensitivity of DAS differs from conventional particle motion sensors, and DAS experiences an increased sensitivity to slow waves, often highly scattered by the subsurface structure and challenging to analyse. To address these issues, several strategies have been already proposed to convert strain rate measurements to particle motion. In this study we focus on strategies based on a quantity we refer to as “deformation”. Deformation is defined as the change in length of the cable and is closely related to displacement, yet both quantities differ from one another: deformation is a relative displacement measurement along a curvilinear path. We show that if the geometry of the DAS deployment is made of sufficiently long rectilinear sections, deformation can be used to recover the displacement without the need of additional instruments. We validate this theoretical result using full-waveform simulations and by comparing, on a real dataset, the seismic velocity recovered from DAS with that recorded by collocated seismometers. The limitations of this approach are discussed, and two applications are shown: enhancing direct P-wave arrivals and simplifying the magnitude estimation of seismic events. While using deformation is in some respects more challenging, converted displacement provides better sensitivity to high velocity phases, and permits the direct application of conventional seismological tools that are less effective when applied to the strain (rate) data.

<p>Stromboli is an open-conduit volcano characterized by mild intermittent explosive activity that produces jets of gas and incandescent blocks. Explosions occur at a typical rate of 3-10 events per hour, VLP signals have dominant periods between 2 and 30 seconds. Seismic activity is also characterized by less energy short-period volcanic tremor related to the continuous out-bursting of small gas bubbles in the upper part of the magmatic column. The high rate of activity as well as the broadband frequency contents of emitted signals make Stromboli volcano an ideal site for testing new techniques of fibre-optic sensing.</p><p>In September 2020, approximately 1 km of fiber-optic cable was deployed on the Northeast flank of Stromboli volcano, together with several seismometers, to record the seismic signals radiated by the persistent Strombolian activity via both DAS and inertial-seismometers, and to compare their records.</p><p>The cable was buried manually about 30 cm deep over a relatively linear path at first and in a triangle-shaped array with 30-meters-long sides in the highest part of the deployment. The strain rate was recorded using a DAS interrogator Febus A1-R with a sampling frequency of 2000 Hz, a spatial interval of 2.4 m and a gauge length of 5m. Data were re-sampled at 200 Hz. A network of 22 nodes SmartSolo IGU-16HR 3C geophones (5 Hz) has been distributed over the fibre path. A Guralp digitizer equipped with a CMG CMG-40T 30 sec seismometer and an infrasound sensor were placed in the upper part of the path. The geolocation of the cable was obtained by performing kinematic GPS measurements with 2 Leica GR25 receivers. All equipment recorded simultaneously several hundreds of explosion quakes between September 20 and 23.</p><p>Data analysis provided the following main results:</p><ul><li>DAS interrogator clearly recorded the numerous explosion-quakes which occurred during the experiment, as well as lower amplitude tremor and LP events.</li> <li>DAS spectrum exhibits a lower resolution at long periods with a cut-off frequency of approximately 3 Hz.</li> <li>VLP seismic events generated by Strombolian activity are identified only at a few DAS channels belonging to a specific portion of the path, which seems affected by local amplification. At these channels, they display waveforms similar to those sensed by the Güralp CMG-40T.</li> <li>Comparison of DAS strain waveform to particle velocity recorded by co-located seismometers shows a perfect match in phase and a good agreement in amplitude.</li> <li>Beamforming methods have been applied to nodes data located on the upper triangle and to strain rate data, both in the 3-5 Hz frequency band. Slightly different back-azimuths were obtained, values estimated via DAS point more to the southwest with respect to the crater area. Apparent velocities obtained with DAS recordings have lower values compared to those obtained with nodes.</li> </ul>