3D Seismic Image Formation Using Dataset Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional seismic data processing methods face a significant bottleneck in loading and processing large 3D travel time cubes, which limits the efficiency of forming three-dimensional seismic images of subterranean regions, particularly in hydrocarbon reservoir characterization.
Innovation Solution
The method involves dividing the seismic dataset into subsets with approximately equal numbers of source and receiver locations, allowing for efficient transmission and processing of seismic partial images using a computer processing unit, leveraging GPUs attached to CPUs for pre-stack Kirchhoff migration, thereby overcoming the computation constraint of loading two travel time cubes for each seismic trace.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the entire seismic dataset with large 3D travel time cubes is loaded for processing, then the completeness of seismic image formation is improved, but the processing efficiency and memory utilization deteriorate due to the bottleneck in loading and processing large datasets
Solution Approach 1:
The patent divides the seismic dataset into multiple subsets, each containing a portion of source locations, receiver locations, and corresponding 3D travel time cubes. This segmentation allows the system to process manageable portions of data sequentially or in parallel, reducing memory requirements and improving processing efficiency while maintaining the completeness of the final seismic image through combination of all subset results
2Measurement precision
If the 3D travel time cubes are loaded in full for each seismic trace processing, then the accuracy of depth migration is improved, but the memory consumption and computation time worsen significantly
Solution Approach 1:
The patent segments the large 3D travel time cubes into smaller portions corresponding to different subsets of source and receiver locations. By loading and processing only the relevant portions of travel time cubes needed for each subset, the system maintains the accuracy required for proper depth migration while significantly reducing peak memory consumption
Solution Approach 2:
The patent implements partial action by loading only the necessary portions of 3D travel time cubes required for processing each seismic subset, rather than loading the entire cube. This partial loading approach maintains sufficient accuracy for depth migration while avoiding the excessive memory consumption that would result from loading complete cubes
3Ease of operation
If conventional processing methods are used without dataset division, then the simplicity of the processing workflow is maintained, but the scalability and ability to handle large datasets worsen
Solution Approach 1:
The patent introduces dataset division into subsets as a scalable solution that maintains workflow simplicity. The segmented approach allows the system to handle increasingly large datasets by simply increasing the number of subsets, while the underlying processing logic remains consistent and manageable through standardized subset handling procedures
Data Source
AI summary
Methods and systems for forming a three-dimensional (“3D”) seismic image of a subterranean region of interest is disclosed. The method includes obtaining a seismic dataset a seismic trace for each of a plurality of pairs of one source and one receiver location and obtaining a 3D travel-time cube for each source location and each receiver location. The method further includes dividing the seismic dataset into a plurality of seismic subsets composed of set of source locations, set of receiver locations a seismic trace for each pair of source and receiver location and the 3D travel-time cube for each source for each receiver location. The method still further includes transmitting, to a random-access memory block of a computer processing unit the seismic subset, and forming a seismic partial image based on the seismic subset, and determining the 3D seismic image based on a combination of the seismic partial images.