Seismic Velocity Model Inversion via Monte Carlo Back Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing seismic imaging methods, such as beam tomography, are prone to cycle skipping and spurious alignments due to coherent noise, which degrade the accuracy of velocity model updates and seismic image alignment.
Innovation Solution
A Monte Carlo method combining simulated annealing techniques with back projection to iteratively select and refine velocity model corrections, reducing the impact of cycle skipping and spurious alignments by probabilistically choosing shift values and damping velocity updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If cross-correlation maximum is used to align beam pairs, then alignment speed is improved, but measurement precision deteriorates due to cycle skipping and spurious alignments
Solution Approach 1:
The patent applies preliminary action by using back-projection to compute expected time shifts before performing cross-correlation. This pre-computed information guides the cross-correlation process, allowing it to start from a more informed position and avoid local maxima caused by cycle skipping. The back-projection step prepares the data in advance, enabling faster and more accurate alignment without requiring exhaustive cross-correlation searches.
Solution Approach 2:
The patent introduces back-projection as an intermediary step between data acquisition and cross-correlation alignment. This intermediary computation provides expected time shift information that mediates the alignment process, helping to distinguish true alignments from spurious ones caused by coherent noise or cycle skipping. The back-projection acts as a filter that prepares the cross-correlation process with better initial information.
2Reliability
If reweighted least-squares inversion is used to handle cycle skips, then robustness is improved, but device complexity increases and can only handle small degrees of cycle skipping
Solution Approach 1:
The patent applies preliminary action by performing back-projection to compute expected time shifts before the inversion process. This pre-computation provides more reliable alignment information that is less susceptible to cycle skipping, reducing the need for complex reweighting schemes. The back-projection step prepares the data in advance, enabling simpler inversion methods to achieve better results.
Solution Approach 2:
The patent substitutes the mechanical least-squares inversion system with a back-projection-based approach. Instead of relying on iterative reweighting and matrix inversion to handle outliers, the method uses back-projection to compute expected time shifts that are inherently more robust to cycle skipping. This substitution replaces a complex mechanical inversion process with a more direct computational approach.
3Manufacturing precision
If more raypaths are processed to improve velocity model accuracy, then manufacturing precision is improved, but loss of time increases
Solution Approach 1:
The patent applies preliminary action by using back-projection to compute expected time shifts for all raypaths before the inversion process. This pre-computation step efficiently processes multiple raypaths simultaneously, providing accurate alignment information without requiring lengthy iterative cross-correlation searches for each raypath individually. The back-projection prepares the data in advance, enabling faster processing of large numbers of raypaths.
Solution Approach 2:
The patent uses back-projection to create synthetic or expected time shift values that copy the essential alignment information from the forward model. These copied expected values can be directly compared with observed time shifts, providing accurate velocity model updates without requiring extensive processing of actual cross-correlation data for every raypath.
Data Source
Figure 1(A)~1(B)
Figure 2(A)~2(B)
Figure 3
AI summary
The present invention is directed to a method and system for minimizing artifacts in a seismic image of a subsurface region of interest, wherein the image is determined a data beam set derived from recorded seismic data and a modeled beam set derived at least in part from a velocity model related to a subsurface region. The artifacts, which may result from cycle skipping and coherent noise, result in misalignment of the modeled and data beam sets. The present invention utilizes a Monte Carlo inversion technique to update the velocity model and thus minimize the artifact in the seismic image.