3D Joint Inversion Dimensionality Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current geophysical data inversion methods, particularly 3D joint inversion, face significant computational challenges and inefficiencies due to the need for multiple forward simulations, leading to slow convergence and high computational costs, making it impractical for real-world applications.
Innovation Solution
The method reduces the 3D joint inversion problem to a series of 1D joint inversion problems by pre-processing 3D data to remove 3D effects, allowing for the creation of 1D models at selected (x, y) locations, which can be jointly inverted, thereby reducing the number of unknowns and computational intensity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If 3D joint inversion is performed using multiple geophysical data types, then the accuracy and reliability of subsurface property prediction is improved, but the computational cost and time required increase significantly
Solution Approach 1:
The patent segments the 3D joint inversion problem into multiple 1D joint inversion problems by dividing the 3D subsurface volume into a series of 1D vertical columns at discrete (x,y) locations. Each 1D inversion independently inverts multiple geophysical data types for that specific location, reducing the computational complexity while maintaining the joint inversion benefits of combining multiple data types.
Solution Approach 2:
The patent extracts 1D vertical profiles of subsurface properties from the 3D volume at selected (x,y) locations. By taking out these 1D slices, the method reduces the dimensionality of the inversion problem from 3D to 1D, making the computational task tractable while still capturing the vertical variation of properties needed for accurate hydrocarbon detection.
2Loss of information
If 3D joint inversion is performed with full 3D modeling, then the completeness of subsurface information is improved, but the computational complexity and resource requirements worsen
Solution Approach 1:
The patent changes the dimensionality of the inversion problem from 3D to 1D by performing inversions along vertical columns at discrete horizontal locations. This dimensionality reduction transforms the computationally intensive 3D joint inversion into a series of manageable 1D problems, while the combination of multiple 1D results reconstructs the 3D subsurface information.
Solution Approach 2:
The patent creates multiple copies of the 1D inversion process at different (x,y) locations across the survey area. Each location undergoes the same 1D joint inversion procedure, and the results are assembled to form the complete 3D subsurface model. This copying approach allows parallel processing and reduces overall computational complexity.
3Measurement precision
If multiple forward simulations are performed for joint inversion, then the accuracy of data prediction is improved, but the computational cost increases
Solution Approach 1:
The patent segments the forward simulation process into 1D vertical column simulations rather than performing full 3D forward modeling. By dividing the computational domain into discrete 1D columns at selected (x,y) locations, each forward simulation becomes computationally inexpensive while the ensemble of 1D results provides accurate predictions for the joint inversion.
Data Source
AI summary
Method for reducing a 3D joint inversion of at least two different types of geophysical data acquired by 3-D surveys to an equivalent set of 1D inversions. First, a 3D inversion is performed on each data type separately to the yield a 3-D model of a physical property corresponding to the data type. Next, a 1D model of the physical property is extracted at selected (x,y) locations. A 1D simulator and the 1D model of the physical property is then used at each of the selected locations to create a synthetic 1D data set at each location. Finally, the 1D synthetic data sets for each different type of geophysical data are jointly inverted at each of the selected locations, yielding improved values of the physical properties. Because the joint inversion is a 1D inversion, the method is computationally advantageous, while recognizing the impact of 3-D effects.