Subsurface Data Retrieval Using Seismic-Text Intelligence Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing subsurface exploration systems face challenges in accurately interpreting vast amounts of data for hydrocarbon exploration and extraction, particularly in domains like energy development and geoscience, leading to inefficiencies in modeling and decision-making.
Innovation Solution
A method and system utilizing intelligence models trained on seismic data-text pairs to facilitate search and retrieval of subsurface data, enabling interactions through vision-language models for enhanced data manipulation, analysis, and extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If sophisticated visual models are used to model subsurface measurements, then understanding of underground content is enhanced, but accuracy of interpretation of sensed data deteriorates
Solution Approach 1:
The patent introduces text as an intermediary modality that bridges visual seismic data and interpretive understanding. The system generates text descriptions from seismic images and uses text-to-image generation to create visual representations from text queries, enabling more accurate data interpretation while maintaining visual model comprehensibility.
Solution Approach 2:
The patent adds a textual dimension to the traditional visual-only subsurface analysis system. By incorporating text generation, text-to-image translation, and multi-modal search capabilities, the system creates a new dimension of interaction that improves both understanding and interpretation accuracy simultaneously.
2Quantity of substance
If multiple different modalities are used to gather information about hydrocarbon exploration, then volumes of data increase, but efficiency of evaluation and practical improvement in yields deteriorates
Solution Approach 1:
The patent creates a unified multi-modal system that handles seismic images, text descriptions, and generated content through a single intelligence model. This universal system can perform multiple functions including data generation, search, retrieval, and analysis across different modalities, improving evaluation efficiency while working with voluminous multi-modal data.
Solution Approach 2:
The system generates synthetic text descriptions from real seismic data and synthetic seismic images from text descriptions, creating copies that enable efficient evaluation without requiring direct analysis of all original voluminous data. This copying approach maintains information fidelity while improving processing efficiency.
3Ease of operation
If sophisticated visual models are used for subsurface analysis, then elaborate understanding is achieved, but time and effort required for data analysis increases
Solution Approach 1:
The system pre-generates text descriptions for seismic images and stores them in a database indexed for efficient retrieval. This preliminary action eliminates the need to generate descriptions on-demand, significantly reducing the time and effort required for data analysis while maintaining elaborate understanding capabilities.
Solution Approach 2:
The system implements feedback loops where text descriptions are generated from seismic data, used to generate new seismic images, and then re-analyzed. This feedback mechanism refines the understanding of subsurface features while automating the analysis process, reducing manual time and effort investment.
Data Source
AI summary
A method for search and retrieval of subsurface data of a geological region includes receiving input data related to the geological region. The method also includes generating a plurality of seismic data-text pairs based on the input data. The method also includes training an intelligence model based on the plurality of seismic data-text pairs. The method also includes generating a database using the intelligence model. The method also includes receiving an input query including a seismic data query, a text query, an image query, or a combination thereof. The method also includes generating an output from the database using the intelligence model and based on the input query.