Seismic Sediment Classification via Machine Learning Thickness Profiles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The interpretation of systems tracts in seismic and/or well log data is challenging due to incorrect or inconsistent application and lack of comprehensive analysis.
Innovation Solution
The method involves classifying sediment packages using seismic data by deriving thickness information, synthesizing it into normalized thickness profiles, and utilizing a learning machine to generate classifications based on sediment stacking patterns and accommodation space analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sequence stratigraphy is applied to interpret seismic data to identify systems tracts, then the ability to predict facies patterns and sediment properties improves, but the complexity of interpretation and risk of incorrect classification increases
Solution Approach 1:
The patent replaces manual geological interpretation with an automated machine learning system that uses neural networks to classify sediment packages. The system automatically processes seismic data, extracts features, and applies sequence stratigraphy rules without human intervention, thereby maintaining prediction accuracy while eliminating interpretation complexity and subjectivity.
Solution Approach 2:
The system enables the seismic data to classify itself through self-organizing neural networks that automatically learn patterns and relationships in the data. The machine learning model performs self-training and self-classification of sediment packages based on inherent geological patterns, reducing the need for expert manual analysis while improving consistency.
2Reliability
If comprehensive analysis of sediment packages is performed to improve subsurface models, then the reliability of geological interpretations increases, but the time and computational resources required increase
Solution Approach 1:
The patent implements iterative training cycles where the neural network is periodically retrained with new seismic data and feedback from classified results. This periodic refinement process continuously improves model reliability while maintaining efficient processing times through automated batch operations rather than continuous manual analysis.
Solution Approach 2:
The system performs preliminary automated classification of sediment packages before detailed analysis, using quick feature extraction to identify candidate regions. This preliminary action filters and prioritizes data that requires comprehensive analysis, reducing overall processing time while ensuring reliable classification of critical zones.
3Adaptability or versatility
If manual interpretation of systems tracts is performed, then flexibility in analysis is maintained, but consistency and accuracy decrease due to incorrect or inconsistent application
Solution Approach 1:
The patent creates a universal classification system that applies the same sequence stratigraphy rules and neural network algorithms to all seismic datasets regardless of location or characteristics. The system maintains adaptability by being configurable for different geological settings while ensuring consistent application of classification criteria through automated processing, eliminating human variability.
Data Source
AI summary
A method comprising obtaining a thickness for each of one or more sediment packages of a subsurface formation. The method comprises generating a thickness profile of each of the one or more sediment packages based on the thickness. The method comprises obtaining one or more properties of each of the one or more sediment packages based on the thickness profile. The method comprises generating, via a learning machine, one or more sediment package classifications based on the one or more properties. The method comprises and performing a subsurface operation based on the one or more sediment package classifications.