Multicomponent distributed acoustic sensing: concept, theory, and applications

Abstract

Conventionally, ground motion due to seismic waves is monitored using seismometers either as single point measurements or as arrays. Such acquisition systems are used both in exploration seismology and in seismic monitoring. In recent years, the use of optical fiber sensing has been steadily on the rise especially in the borehole environment for information, including, but not limited to, temperature, pressure, and strain. Optical fibers are enticing because of their dense spatial sampling and their low operation cost, given that the optical fiber is readily available at the time boreholes are constructed. In exploration seismology, distributed acoustic sensing (DAS) provides the means to record seismic strain projections along the optical fiber. Although this empowers us to perform dense spatial sampling measurements along the optical fiber, the single component data are a poor approximation to the total strain tensor. In my thesis, I propose solutions to obtain the full strain tensor at any location along the purposefully designed optical fiber through geometrical concepts and inversion theory. I also present a mechanism to recover multicomponent DAS data using only a single optical fiber in an averaging sense suitable for long wavelength applications. The ability to obtain multicomponent DAS data paves the way to better representation of the elastic wavefield leading to better tomography, imaging, and ultimately reservoir characterization. Given the capacity to record multicomponent DAS data, I introduce an imaging method for seismic source mechanism that takes advantage of conventional displacement vector and the novel strain tensor measurements. The technique makes it possible to estimate source parameters in near real-time with approximate subsurface models that adequately describe the direct P- and S-waves kinematics. According to wave propagation theory, accurate seismic wavefield extrapolation requires both the seismic stress tensor and displacement data. Typical practice using only displacement data introduces nonphysical wave modes during wavefield extrapolation resulting in elastic seismic images contaminated by artifacts. The availability of multicomponent strain data makes stress observations possible given that the material properties in the vicinity of the recording array are known to facilitate accurate wavefield reconstruction based on multicomponent DAS data. I demonstrate accurate elastic seismic wavefield extrapolation, leading to seismic image improvements that better represent the subsurface. Using the same general wavefield reconstruction theory, I also propose to accurately reconstruct wavefields from a single layer boundary on the computation domain to reduce memory and computational requirements for seismic applications that require wavefield storage, especially for large-scale 3D experiments. I validate all proposed techniques with realistic numerical experiments using models with arbitrary anisotropy and heterogeneity. I test my methods using field datasets such as a DAS vertical seismic profile (VSP) data from the Eagle Ford shale formation. The thought-provoking development of multicomponent DAS acquisition, which is not possible with currently available technology, has the potential to revolutionize imaging and inversion using elastic wavefields. The work in this thesis can inspire new applications outside of seismic exploration challenges, including, but not limited to structural health monitoring and seismological investigation at global or regional scales.