RTM Seismic Imaging Memory-Based Storage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Reverse Time Migration (RTM) seismic imaging processes are bottlenecked by disk I/O and require significant computational resources, especially when dealing with large models and extensive data storage needs, which limits the efficiency and accuracy of subsurface imaging.
Innovation Solution
Implementing a synchronized, communication-intensive, massive domain-partitioned approach on a distributed processor/memory supercomputer, where the velocity model is partitioned into blocks and processed in parallel across thousands of nodes, reducing the need for disk storage and alleviating disk I/O bottlenecks by using local memory for intermediate data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional RTM seismic imaging processes are used with disk storage for intermediate data, then data can be persisted, but disk I/O bottlenecks occur and computational efficiency decreases
Solution Approach 1:
The patent replaces the mechanical disk I/O system with a memory-based storage system. Intermediate wavefield data that traditionally would be written to and read from disk is instead kept in high-speed memory throughout the RTM process, eliminating disk I/O bottlenecks and significantly improving computational efficiency.
Solution Approach 2:
The patent divides the velocity model and shot data into multiple partitions or blocks that can be processed independently in parallel. This segmentation allows different portions of the computation to occur simultaneously in memory, reducing overall computation time and eliminating the need for sequential disk I/O operations.
2Measurement precision
If more computational resources are allocated to RTM processing, then imaging accuracy improves, but system cost and complexity increase
Solution Approach 1:
The patent segments the large-scale RTM computation into smaller, independent partitioned computations that can be distributed across multiple processing units. Each partition processes a subset of the velocity model and shot data, allowing parallel execution that improves imaging accuracy through more comprehensive sampling while managing system complexity through modular design.
Solution Approach 2:
The patent introduces a new dimension of parallel processing by partitioning the computational domain across multiple processors or computing nodes. This dimensional expansion allows the system to achieve higher imaging accuracy through increased computational power without proportionally increasing per-node complexity.
3Quantity of substance
If disk storage is used for intermediate wavefield data, then data can be saved, but storage requirements and I/O overhead increase
Solution Approach 1:
The patent substitutes memory-based storage for disk-based storage of intermediate wavefield data. By keeping all necessary intermediate data in high-speed memory throughout the RTM computation, the system eliminates the slow disk I/O operations that limit processing speed while maintaining adequate storage capacity through efficient memory management.
Solution Approach 2:
The patent loads all necessary velocity model and shot data into memory before processing begins, and maintains this data in memory throughout the computation. This preliminary action of pre-loading data eliminates the need for repeated disk I/O operations during processing, significantly improving processing speed.
Data Source
AI summary
A system, method and computer program product for seismic imaging implements a seismic imaging algorithm utilizing Reverse Time Migration technique requiring large communication bandwidth and low latency to convert a parallel problem into one solved using massive domain partitioning. Several aspects of the imaging problem, including very regular and local communication patterns, balanced compute and communication requirements, scratch data handling and multiple-pass approaches. The partitioning of the velocity model into processing blocks allows each sub-problem to fit in a local cache, increasing locality and bandwidth and reducing latency.