Wavefield reconstruction, inversion and imaging using local solvers

Jaimes Osorio, Ligia Elena (2021) Wavefield reconstruction, inversion and imaging using local solvers. Doctoral (PhD) thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (12MB)

Abstract

In exploration geophysics, we usually implement non-invasive methods to image the interior of the Earth. Seismic methods, for example, are ideal for this purpose because they provide high-resolution images at large depths. During the construction of these images, understanding how seismic waves propagate through the subsurface is vital. Lacking analytical solutions for realistically complex models, numerical modeling becomes essential to comprehend the recorded data. Conventional modeling algorithms are designed to solve the wave equation throughout the entire model, even though we are often interested in only much smaller areas. To solve for the whole model is computationally expensive, particularly when the algorithm repeatedly simulates wavefields, as is often the case in common imaging techniques such as reverse time migration and full waveform inversion. The main goal of this thesis is to explore and understand the potential of local solvers under different scenarios, for instance, in the presence of large impedance contrast re ectors and complex geometries in different domains. We first demonstrate that a local acoustic-elastic solver can model the amplitude with sufficient accuracy between the surface and the reflector of interest and then can be used as a constraint in Amplitude Versus Offset (AVO) inversion. We show that the amplitude combined with the phase improves the convergence of this algorithm. However, the amplitude is not in perfect agreement with that obtained with full elastic modelling. This motivates our interest in building a local elastic solver that avoids the interchange between domains but keeps the computational, time, and resources cost feasible. We implement the Multiple Point Sources (MPS) method in a standard Finite Difference (FD) solver, computing the global wavefield in the local domain successfully using only the ability to record and inject data at the boundary, while being memory efficient. We test the robustness of the local elastic solver on different elastic models and uses in AVO and Phase Versus Offset (PVO) inversions, retrieving elastic parameters with high accuracy. To take the elastic solver to the next level we assess the possibility of using converted wave modes during the inversion. To this end we explore different imaging conditions based on different wave modes that allow us to measure the focusing of the image to be ultimately used in conjunction with the local elastic solver. Notice that all implementations and results are shown in 2D.

Actions (login required)

View Item