• CCS
• DAS
• denoising
• Distributed Acoustic Sensing
• microseismic monitoring
• seismology
• signal processing
• spectral subtraction

Distributed Acoustic Sensing (DAS) is a fast-developing technology that in recent years
has gained significant popularity in seismology, especially in microseismic monitoring operations. Its key advantage lies in its ability to convert fiber-optic cables, whether conventional telecommunication or purpose-built, into dense arrays of seismometers. This makes DAS uniquely effective for reservoir monitoring applications, especially in logistically challenging environments like offshore areas or in geothermal wells, where high temperature and pressure conditions do not allow the use of conventional seismic sensors. An important advantage of fiber-optic cables is their flexibility in deployment; they can be installed in existing production or injection wells, eliminating the need for dedicated
monitoring wells. Moreover, their intrinsic characteristics allow for them to be installed
close to the reservoir, enabling the detection of nearby very small seismic events. DAS systems can sample seismic wavefields quasi-continuously over several kilometers, at about 1 m spacing and sampling frequencies of 1 kHz or higher, making them ideal for applications such as Carbon Capture and Storage (CCS) and Enhanced Geothermal Systems (EGS) monitoring. The durability and longevity of optical fibers deployed in boreholes further ensure an effective monitoring of injection and post-injection activities
over extended periods. Despite these advantages, the high sampling rate in space and time of DAS data presents challenges for traditional seismological techniques, which often struggle to process such
data effectively. DAS recordings are usually influenced by various factors and may vary along segments of the fiber. The recorded seismic signals are often affected by multiple types of noise, including high-frequency random noise, common mode noise, coupling noise, and optical noise; these complicate the extraction of hidden signals, reducing the signal-to-noise ratio (SNR) of the overall recording and therefore limiting the detection of low-amplitude microseismic events. To ehnance the SNR of DAS data traditional frequency filtering methods, such as low- and high-pass filters or Butterworth bandpass
filters can be combined with other techniques (e.g., f-k filtering to remove optical noise
around k=0). However, DAS recordings generally exhibit higher noise levels compared to
conventional seismometers and such filtering procedures may be not sufficient to improve
the data quality.

The main goal of this thesis is to analyze microseismic data collected using DAS technology, installed for the CCS Ravenna Hub project in the offshore of Ravenna at the Porto
Corsini Mare Ovest (PCMW) platform. In this work we focus on the development of
an efficient de-noising workflow able to produce better results than traditional filtering
techniques. The de-noising approach developed in this Thesis is based on an adaptation
of spectral-subtractive algorithms, which are commonly used in speech enhancement for
audio signals. However, in this case, these algorithms have been re-purposed for signal
enhancement in DAS data.

To assess the performance of the proposed method, 2D synthetic DAS data were first
generated following the approach described by Rapagnani (M.Sc. Thesis, 2023). These
simulations modeled seismic events at varying depths and offsets relative to the installed
DAS fiber, using the actual velocity model and well geometry from the PCMW offshore
CCS field. Initially, Gaussian noise at different levels was added to the synthetic data to
test the algorithm’s capabilities with this type of noise. Later, real noise extracted from
both Eni’s proprietary DAS data from the Ravenna CCS project and publicly available
DAS data from the FORGE (Frontier Observatory for Research in Geothermal Energy)
EGS project was added to the synthetic waveforms to better replicate the characteristics
of recorded DAS signals across different scenarios and acquisition layouts.
The de-noising workflow was finally validated on real microseismic events recorded by the
DAS system at the Ravenna CCS project during the pre-CO2 injection phase, as well as
on DAS data from the onshore Enhanced Geothermal System (EGS) FORGE project.
All tests showed significant improvements in SNR across different microseismic DAS
datasets, acquired in various scenarios and environments.