Geophysical Data Inversion Using Adjoint Method Gradients
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Iterative geophysical inversion methods are computationally expensive due to the need for numerous simulations, making them impractical for many applications, especially when dealing with large numbers of geophysical sources.
Innovation Solution
The method employs the adjoint method to efficiently calculate the gradients of the cost function for local optimization, using encoded gathers of simultaneous source data to reduce computational time and improve inversion efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If iterative inversion methods are used to achieve accurate subsurface models, then measurement precision is improved, but computational expense increases significantly
Solution Approach 1:
The patent divides the inversion problem into smaller sub-problems by processing data from individual sources separately rather than simultaneously. Each source's data is inverted independently using the same computational model, which reduces the complexity of each individual computation while maintaining overall accuracy through aggregation of results from multiple sources.
Solution Approach 2:
The patent performs inversion for a subset of sources at a time rather than attempting to invert all sources simultaneously. This partial action approach allows the computational model to handle manageable portions of the data, reducing the computational burden per iteration while still achieving convergence through multiple passes over the complete data set.
2Measurement precision
If the number of simulations is increased to improve model accuracy, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent segments the inversion process by handling each source independently, which allows for more efficient use of computational resources. By dividing the total computation into smaller, manageable tasks that can be processed in parallel or sequentially without requiring simultaneous inversion of all sources, the method reduces total computational time while maintaining the number of iterations needed for accuracy.
Solution Approach 2:
The patent performs preliminary computations by pre-processing source data and preparing the computational model before the main inversion process begins. This includes pre-calculating source waveforms, preparing initial models, and setting up computational parameters, which reduces the time required during actual inversion iterations.
3Productivity
If simultaneous source data is processed to reduce computational effort, then productivity is improved, but measurement precision may deteriorate
Solution Approach 1:
The patent segments the simultaneous source data into individual source components, processing each source's contribution separately through the inversion model. This segmentation maintains the precision of individual source measurements while achieving the computational efficiency of simultaneous processing, as each segmented source can be inverted using the same optimized computational model without requiring simultaneous inversion of all sources.
Solution Approach 2:
The patent creates a simplified copy of the inversion process that can be applied to each source independently. By using the same computational model and inversion parameters for each source, the method maintains measurement precision while reducing computational complexity through repeated application of a standardized, optimized process rather than complex simultaneous inversion.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for efficient inversion of measured geophysical data from a subsurface region to prospect for hydrocarbons. Gathers of measured data (40) are encoded (60) using a set of non-equivalent encoding functions (30). Then all data records in each encoded gather that correspond to a single receiver are summed (60), repeating for each receiver to generate a simultaneous encoded gather (80). The method employs iterative, local optimization of a cost function to invert the encoded gathers of simultaneous source data. An adjoint method is used to calculate the gradients of the cost function needed for the local optimization process (100). The inverted data yields a physical properties model (110) of the subsurface region that, after iterative updating, can indicate presence of accumulations of hydrocarbons.