Supervisors: Morten Jakobsen, Inga Berre, Einar Iversen, Florin A. Radu
Short description of the project:
Activities such as conventional and unconventional hydrocarbon production, underground mining, geothermal energy production, and carbon sequestration change the stress-state of the subsurface, often triggering microseismicity. These events occur when critically stressed faults slip, producing weak seismic signals with high dominant frequencies. Monitoring such events is essential to mitigate damage to geological formations. The damaged formation may create a zone for the leakage of gases, such as methane, into overburden rocks and ultimately into the ecosystem. Effective monitoring enables real-time operational adjustments, such as modifying fluid injection rates or altering target locations, to reduce induced seismic hazards.
Microseismic monitoring relies on seismic data recorded by borehole or surface sensors. However, surface data are more contaminated by noise than borehole data and require extensive processing. The complete characterization of the microseismic source includes the determination of the event location, origin time, moment tensor, and source-time function. This is crucial to understand the local stress regime and propagation trends of a rupture in geological formations. Most geological formations exhibit anisotropy caused by layering and aligned fractures. Using linear slip theory, the elastic parameters of a fractured anisotropic medium can be related to those of the unfractured background and the fracture properties. Since fractures represent material weaknesses, they play a key role in determining where critically stressed regions are and therefore where microseismic events can occur. The thesis develops methodologies for building fractured anisotropic velocity models and performing microseismic waveform inversion within such models.
The thesis work applies scattering theory to model the seismic wavefield and invert seismic waveforms in anisotropic elastic media. The medium is decomposed into a background and a contrast component. The scattering approach requires discretization of the contrast region only, making it inherently target-oriented. Wave scattering is formulated through a frequency-domain integral equation of Lippmann–Schwinger type. This representation combines the background wavefield with the scattered wavefield arising from medium contrasts. The fast Fourier transform–accelerated Krylov subspace method is used to solve the integral equation efficiently. The forward solver models the wavefield from both moment tensor and force vector point sources and is integrated into the frequency-domain full waveform inversion (FWI) framework. Inversion is performed using the distorted Born iterative method, which linearizes the sought model around a known heterogeneous background and computes Fréchet derivatives in a matrix-free manner. This enables efficient recovery of both elastic and fracture parameters. The distorted Born iterative method supports the time-lapse monitoring of velocity model changes due to ongoing fluid injection or extraction by reusing precomputed Green’s functions for cost-effective updates in localized regions.
Building on the recovered fractured anisotropic velocity model, the study then addresses microseismic source characterization. Fractures influence both the moment tensor and the radiation patterns of microseismic events. The focus is on double-couple mechanisms—the most common source type in both natural and induced seismicity. It represents a shear slip on planar faults. The double-couple moment tensor is parameterized in terms of fracture properties and geometrical parameters, including fault dip, azimuth of the fault normal, slip magnitude, and slip angle. Microseismic waveform inversion is performed to jointly estimate the source location and these geometrical parameters.
By incorporating fracture-induced anisotropy into seismic modelling and inversion, this work facilitates more accurate subsurface characterization. An accurate characterization of microseismic events improves their utility for passive seismic imaging in areas where active seismic coverage is limited.