Geophysical Inversion With ML Priors for Uncertainty Quantification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional geophysical inversion methods face challenges with incomplete information and non-unique solutions due to limited seismic data aperture and bandwidth, which affect the resolution and accuracy of petrophysical property estimation, particularly in multi-parameter inversions.
Innovation Solution
A machine-learning-augmented inversion method that incorporates prior geological knowledge and conditional models to initialize, solve, and test multiple plausible solutions, using machine-learned models to guide the inversion process and manage uncertainties.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If conventional two-step inversion process is used, then computational complexity is reduced, but inversion resolution and accuracy deteriorate due to incomplete information and non-unique solutions
Solution Approach 1:
The inversion process is segmented into multiple parallel plausible solutions rather than a single sequential solution. The system divides the inversion problem into multiple independent inversion runs, each producing a different plausible solution consistent with the observed data, thereby addressing the non-uniqueness issue while maintaining computational feasibility.
Solution Approach 2:
The system changes parameters by introducing prior geological knowledge and conditional models as additional constraints. By incorporating domain expert knowledge about rock types, facies, and fluid types, the inversion process transforms from a purely data-driven approach to one that integrates multiple parameter sets including geological priors, enhancing accuracy without excessive complexity increase.
2Measurement precision
If more seismic data aperture and bandwidth are acquired, then inversion resolution improves, but data acquisition cost and time increase
Solution Approach 1:
The system performs preliminary action by incorporating prior geological knowledge and conditional models before the actual inversion process. By pre-loading domain expert knowledge about rock types, facies, and fluid types, the system prepares constraint frameworks that guide the inversion, allowing high-resolution results to be achieved with existing limited seismic data rather than requiring additional data acquisition.
3Reliability
If multiple plausible solutions are generated and tested, then uncertainty quantification improves, but computational time increases
Solution Approach 1:
The system applies partial action by generating a limited but sufficient number of plausible solutions rather than exhaustively exploring all possible solutions. By producing multiple (but not all) plausible solutions consistent with the data and geological priors, the system achieves adequate uncertainty quantification without the computational burden of exhaustive enumeration, balancing reliability with efficiency.
Data Source
AI summary
A computer-implemented method for augmented inversion and uncertainty quantification for characterizing geophysical bodies is disclosed. The method includes machine-learning-augmented inversion that also facilitates the characterization of uncertainties in geophysical bodies. The method may further estimate wavelets without a well-log calibration, thereby enabling a pre-discovery exploration phase when well log data is unavailable. The machine learning component incorporates a priori knowledge about the subsurface and physics, such as distributions of expected rock types and rock properties, geological structures, and wavelets, through learning from examples. The methodology also allows for conditioning the characterization with the information extracted a priori about the geobodies, such as probabilities of rock types, using other analysis tools. Thus, the conditioning strategy may make the inversion more robust even when a priori distributions are not well balanced. Using the method, a scenario testing workflow may evaluate different candidate subsurface models, facilitating the management of uncertainty in decision-making processes.