Seismic Velocity Model Inversion Using Zero-Offset Wavefields
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for generating subsurface images from seismic data, such as full waveform inversion, face challenges due to high computational costs and slow convergence, often getting trapped in local solutions or failing to accurately represent deep geological features.
Innovation Solution
A method that iteratively calculates and optimizes a seismic velocity parameter model using a cost function dependent on zero-offset seismic wavefields and perturbation terms, employing linearized perturbations and migration operators to refine the model, allowing for efficient and accurate subsurface imaging.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If full waveform inversion is used to generate subsurface images, then imaging accuracy is improved, but computational cost increases significantly
Solution Approach 1:
The patent segments the velocity model updates into two distinct parts: a smooth background velocity model and a high-frequency reflectivity model. This segmentation allows the inversion to handle different aspects of subsurface imaging separately, reducing the computational burden while maintaining accuracy in capturing both shallow and deep geological features.
Solution Approach 2:
The patent introduces a migration operator as an intermediary tool that transforms the inverted velocity model into a subsurface image. This operator acts as a mediator between the seismic data and the final image, enabling efficient computation by leveraging existing seismic processing techniques to reduce the overall computational cost.
2Productivity
If local optimization techniques are used in full waveform inversion, then computational efficiency is improved, but the method gets trapped in local solutions
Solution Approach 1:
The patent employs a dynamic inversion strategy where the inversion process adapts to different depth regions. The cost function dynamically weights the contribution of shallow and deep structures, allowing the optimization to escape local minima by adjusting the search focus based on the current model state and data characteristics.
Solution Approach 2:
The patent changes the parameterization of the velocity model by separating it into smooth background and high-frequency components. This parameter change transforms the inversion problem into one where local optimization can effectively converge, as the separated parameters have different sensitivity characteristics that prevent trapping in local solutions.
3Measurement precision
If classical full waveform inversion is applied, then subsurface imaging is achieved, but convergence is slow
Solution Approach 1:
The patent performs preliminary actions by pre-processing the seismic data to extract zero-offset wavefields and preparing the cost function with pre-computed migration operators. These preliminary steps reduce the computational burden during the inversion process, accelerating convergence while maintaining imaging quality.
Solution Approach 2:
The patent uses copying of seismic wavefield information through the migration operator to efficiently transform the inverted velocity model into the final subsurface image. This copying process leverages existing wavefield data and processing techniques, significantly reducing the time required for convergence compared to traditional methods.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
The invention relates to a computer-implemented method for generating an image of a subsurface of an area of interest from seismic data. The method comprises: - providing (510) seismic wavefields, - providing (520) a zero-offset seismic wavefield dataset, - determining (530) a seismic velocity parameter model w(x) comprising an initial model w 0 (x), a low frequency perturbation term δm b (x) and a high frequency perturbation term δm r (x), - determining (540) an optimal seismic velocity parameter model w opt (x) by computing a plurality of iterations, each iteration comprising calculating and optimizing a cost function, said cost function being dependent on the zero-offset seismic wavefield and on the low frequency perturbation term δm b (x) as a parameter in the optimization of the cost function, the high frequency perturbation term δm r (x) being related to the velocity parameter model w(x) to keep the provided zero-offset seismic wavefield data invariant with respect to the low frequency perturbation term δm b (x).