Time Domain Reconstructed Full Wavefield Inversion for Seismic Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional full waveform inversion (FWI) techniques face challenges in converging to accurate subsurface models due to local minima issues, lack of low frequencies, and cycle skipping, which affect the fidelity of seismic imaging and subsurface structure characterization.
Innovation Solution
The implementation of time domain reconstructed full waveform inversion (TDRFWI) methods, which relax the constraint of exact wave equation solutions, allowing for an L2 approximate approach to jointly minimize data misfit and wave equation error, thereby expanding the search space and avoiding local minima.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional full waveform inversion techniques are used, then the method can process seismic data to generate subsurface models, but the convergence is slow and prone to local minima due to cycle skipping effects
Solution Approach 1:
The patent applies preliminary wavefield reconstruction to obtain an accurate source wavelet before performing full waveform inversion. This preliminary action provides a reliable starting point for the inversion process, preventing cycle skipping and local minima issues that plague conventional methods. By preparing the source wavelet in advance through wavefield reconstruction, the inversion converges faster and more reliably.
Solution Approach 2:
The patent introduces wavefield reconstruction as an intermediary step between data acquisition and full waveform inversion. This intermediary process generates a reconstructed source wavelet that serves as a bridge, enabling the subsequent inversion to proceed without the cycle skipping problems that occur when using conventional direct inversion methods. The intermediary wavefield reconstruction ensures accurate phase and amplitude information is available for reliable convergence.
2Measurement precision
If exact wave equation solutions are enforced, then the forward modeling is physically accurate, but the search space is constrained leading to local minima
Solution Approach 1:
The patent applies partial enforcement of the wave equation by using wavefield reconstruction to obtain source wavelets without requiring exact satisfaction of the wave equation during the inversion process. This partial action approach maintains sufficient physical accuracy while freeing the inversion from strict wave equation constraints, allowing the search to explore a broader parameter space and avoid local minima. The method uses enough wave equation compliance to ensure physical realism but not so much as to restrict the search.
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
Computer systems and methods are provided for seismic wavefield processing, including time domain reconstructed full wavefield inversion (TDRFWI) of seismic survey data. Suitable methods include forward propagating a source signal based on a model of a subsurface region, generating a residual based on the forward- propagated wavefield in comparison to field data, back propagating the residual to generate a reconstructed source perturbation, and adding the reconstructed source perturbation to the source signal to generate a reconstructed source wavefield. The reconstructed source perturbation is forward propagated to generate a reconstructed wavefield perturbation, and added to the reconstructed wavefield perturbation to reconstruct the wavefield. The propagations are performed in the time domain, and the model can be updated based on the reconstructed wavefield and reconstructed source perturbation to generate high fidelity subsurface images.