Electronic Thesis and Dissertation Repository

Thesis Format

Integrated Article


Doctor of Philosophy




Gerhard Pratt


The imaging of structures in sedimentary basins has typically been carried out using conventional seismic reflection techniques. In complex geological architectures, such as those within the Superior Province in Canada, conventional seismic imaging yields suboptimal results due to factors stemming from complex energy scattering and crooked seismic geometries. Full-waveform inversion (FWI) is a nonlinear inverse technique capable of retrieving quantitative images of velocity structures. Judicious data preconditioning, and multiscale inversion strategies allow FWI to effectively estimate velocity variations in the first few kilometers of the subsurface. Non-conventional seismic reflection processing, based on azimuthal binning and enhanced migration velocity models, improves energy focusing and imaging quality of structures at depth.

I demonstrate that the serious nonlinearity of FWI in crystalline zones (in Larder Lake and in the northeastern portion of the Sudbury Structure), is alleviated by implementing a “multiscale layer-stripping” strategy. The strategy uses a i) combination of explosive and vibroseis sources to retrieve low-wavenumber features of the velocity background, ii) hierarchical minimization of logarithmic phase-only and conventional phase-amplitude residuals to mitigate large dynamic variations within the data, and iii) progressive inclusion of higher frequencies and late arrivals to obtain a natural transition between low- and high- wavenumber features. I illustrate that the implementation of optimum binning strategies effectively enhances signal alignment, generating in-phase stacked sections. The use of near-surface FWI P-wave velocity estimations in the construction of migration velocity models, improves the strength and continuity of reflections at depth.

With the application of these imaging techniques, I unravel the geometry of prominent structures in structurally-controlled mineralized zones. In Larder Lake, the Larder Lake-Cadillac Deformation Zone (LLCDZ), and the Lincoln-Nipissing Shear Zone (LNSZ), are imaged. The LNSZ is retrieved as a north-dipping fault, extending to depths not previously identified of 8 km. The Sudbury velocity model reveals the internal character of the structure, agreeing with known geology and borehole information. New estimates of thicknesses and dips are given for the northeastern portion of the Sudbury Structure.

Keywords: Full-waveform inversion, Finite-difference modeling, Seismic imaging, Azimuthal binning, Energy focusing, Reflection processing, Crystalline environments, Larder Lake, Sudbury.

Summary for Lay Audience

The Earth has finite resources, including mining commodities such as gold, copper and nickel that we use to build the world's infrastructure including roads, hospitals, electrical grids, and telecommunications. For many decades, the mining industry has been successfully able to mine shallow mineral deposits. However, these near-surface deposits are becoming increasingly difficult to find. There is a pressing need for exploration methods that can find deeper mining prospects more reliably. This requires mapping the distribution of ore deposits at depth. Subsurface imaging methods can provide solutions for these exploration campaigns. In this work, I employed seismic techniques to image both the structural near-surface configuration of the subsurface and the architecture of potential mineralized structural pathways at depth. I used full-waveform inversion to retrieve high-quality geophysical images of the first few kilometers of the subsurface. Full-waveform inversion is a highly accurate iterative tomographic technique that makes optimal use of the seismic wave information. I investigated the effect of crooked geometries in waveform inversion by not only considering a two-dimensional (2D) modeling and inversion approach, but also a two-and-one-half-dimensional (2.5D) approach. In addition, I performed advanced reflection processing by implementing geologically driven techniques in which "bins" used to group traces are re-orientated along geological strike, improving energy focusing. The final result is an optimal reflectivity image. Combining near-surface velocity images with reflectivity images at depth, I was able to fill the gap between surface geology and crustal-scale structures in two different hard rock environments in Canada: in the Larder Lake and Sudbury mining district.