TY - THES

T1 - Modeling and analysis of microseismic data with Distributed Acoustic Sensing (DAS)

AU - Rapagnani, Giacomo

N1 - no embargo

PY - 2023

Y1 - 2023

N2 - Distributed Acoustic Sensing (DAS) is an emergent data acquisition technology allowing to measure the dynamic strain along the axis of an optical fiber. A DAS system consists of an optical fiber and an optoelectronic device, known as an interrogator unit (IU), that emits laser pulses into an optical fiber. The IU is able detect subtle phase shifts present in the backscattered light along the fiber and convert it in strain (or strain rate). In this way it is possible to record the dynamic strain caused by an external source (e.g. seismic waves) and affecting the fiber and, for this reason, this technology is becoming popular in seismology since it turns the optical fiber into a seismic array. In the last few years, the interest of the seismological community in DAS technology has grown exponentially, especially for microseismic monitoring operations in borehole installations, where its high spatial and temporal sampling, if compared to standard seismological acquisition technologies (e.g. geophones or seismometers), provides more detailed information on the seismic wavefield. Furthermore, DAS is particularly useful for data acquisition in logistic challenging scenarios (e.g. offshore areas, borehole installations, glaciers or volcanic environments) where the deployment of conventional seismometers can be difficult. It’s thus clear that, due to their high spatial and temporal sampling, Distributed Acoustic Sensing (DAS) systems can generate massive amounts of data, especially for long data acquisition periods. For example, a sample of 1 day of data collected with a 1 km long fiber with inter channel distances of about 1m and temporal sampling of 0.5 ms can easily reach 2 TB. This highlight the need of specific data analysis procedures that, at the same time, are fast enough and capable to exploit the massive amount of information contained in such data. DAS data acquisition experiments are not yet a standard practice, and for this reason it is useful to use synthetics to evaluate the performance of different DAS acquisition geometries and to test new data analysis methods. Despite the popularity of DAS systems, there is a lack of standard modeling and analysis tools that can be used within routine procedures.In this thesis, I have developed a versatile workflow for synthetic DAS data generation based on the convolutional model. For this purpose, I have developed a travel-time calculator that solves the Eikonal equation, which is capable to manage various data acquisition geometries, including scenarios where an optical fiber is deployed in deep boreholes, either vertical or deviated ones. Synthetic DAS seismograms are then generated by using the computed travel-times both for P and S phases and the convolutional model. Although DAS synthetics calculated with the convolutional model are less realistic than the ones calculated with other methods, such as the reflectivity or the spectral element method, their computation is much faster and efficient. This aspect is particularly important in the generation of large DAS synthetics datasets. Our synthetic generation workflow can be used to 1) test newly developed seismic event detection and location methods for DAS data 2) train machine learning models. The first part of the thesis covers the basic concepts about the physics and the instruments needed to understand how Distributed Acoustic Sensing works. The second part of the thesis focuses on the description of the synthetic DAS data generator workflow I developed. I first describe the implementation of the Eikonal solver for travel-time calculation, followed by a detailed description on the synthetic DAS data generation process, based on the convolutional model. In the last part of this thesis, I show a comparison of the synthetics obtained using our workflow with the ones obtained using the spectral element method and I will show an application with waveform-based DAS data analysis methods.

AB - Distributed Acoustic Sensing (DAS) is an emergent data acquisition technology allowing to measure the dynamic strain along the axis of an optical fiber. A DAS system consists of an optical fiber and an optoelectronic device, known as an interrogator unit (IU), that emits laser pulses into an optical fiber. The IU is able detect subtle phase shifts present in the backscattered light along the fiber and convert it in strain (or strain rate). In this way it is possible to record the dynamic strain caused by an external source (e.g. seismic waves) and affecting the fiber and, for this reason, this technology is becoming popular in seismology since it turns the optical fiber into a seismic array. In the last few years, the interest of the seismological community in DAS technology has grown exponentially, especially for microseismic monitoring operations in borehole installations, where its high spatial and temporal sampling, if compared to standard seismological acquisition technologies (e.g. geophones or seismometers), provides more detailed information on the seismic wavefield. Furthermore, DAS is particularly useful for data acquisition in logistic challenging scenarios (e.g. offshore areas, borehole installations, glaciers or volcanic environments) where the deployment of conventional seismometers can be difficult. It’s thus clear that, due to their high spatial and temporal sampling, Distributed Acoustic Sensing (DAS) systems can generate massive amounts of data, especially for long data acquisition periods. For example, a sample of 1 day of data collected with a 1 km long fiber with inter channel distances of about 1m and temporal sampling of 0.5 ms can easily reach 2 TB. This highlight the need of specific data analysis procedures that, at the same time, are fast enough and capable to exploit the massive amount of information contained in such data. DAS data acquisition experiments are not yet a standard practice, and for this reason it is useful to use synthetics to evaluate the performance of different DAS acquisition geometries and to test new data analysis methods. Despite the popularity of DAS systems, there is a lack of standard modeling and analysis tools that can be used within routine procedures.In this thesis, I have developed a versatile workflow for synthetic DAS data generation based on the convolutional model. For this purpose, I have developed a travel-time calculator that solves the Eikonal equation, which is capable to manage various data acquisition geometries, including scenarios where an optical fiber is deployed in deep boreholes, either vertical or deviated ones. Synthetic DAS seismograms are then generated by using the computed travel-times both for P and S phases and the convolutional model. Although DAS synthetics calculated with the convolutional model are less realistic than the ones calculated with other methods, such as the reflectivity or the spectral element method, their computation is much faster and efficient. This aspect is particularly important in the generation of large DAS synthetics datasets. Our synthetic generation workflow can be used to 1) test newly developed seismic event detection and location methods for DAS data 2) train machine learning models. The first part of the thesis covers the basic concepts about the physics and the instruments needed to understand how Distributed Acoustic Sensing works. The second part of the thesis focuses on the description of the synthetic DAS data generator workflow I developed. I first describe the implementation of the Eikonal solver for travel-time calculation, followed by a detailed description on the synthetic DAS data generation process, based on the convolutional model. In the last part of this thesis, I show a comparison of the synthetics obtained using our workflow with the ones obtained using the spectral element method and I will show an application with waveform-based DAS data analysis methods.

KW - seismology

KW - das

KW - microseismic

KW - modeling

KW - seismology

KW - das

KW - microseismic

KW - modeling

U2 - 10.34901/mul.pub.2024.023

DO - 10.34901/mul.pub.2024.023

M3 - Master's Thesis

ER -