Seismic Attribute Aggregation for Reliable Low-Frequency Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in geological and geophysical exploration is identifying subsurface hydrocarbon reservoirs using seismic data, as low frequency anomalies do not consistently appear across different subsurface regions, making it difficult to reliably detect hydrocarbons.
Innovation Solution
An unsupervised machine learning clustering model is applied to seismic data to generate factor cubes for various attributes, learn weights specific to each region, and aggregate these cubes to create an aggregated cube that highlights hydrocarbon locations, using techniques like spectral decomposition and agglomerative clustering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional seismic data analysis methods are used to detect low frequency anomalies, then the detection process is simple, but the reliability of hydrocarbon identification is poor because low frequency anomalies do not consistently appear across different subsurface regions
Solution Approach 1:
The patent segments the seismic data analysis by dividing it into multiple factors (low frequency anomaly, flat spot, bright spot, fracture) and processing each factor separately through unsupervised machine learning clustering models to generate individual factor cubes, which are then aggregated to improve overall detection reliability
Solution Approach 2:
The patent changes the analysis parameters by applying unsupervised machine learning clustering models to learn region-specific weights for different seismic attributes, allowing the system to adapt to varying subsurface conditions and improve hydrocarbon identification reliability across different regions
2Measurement precision
If unsupervised machine learning clustering models are applied to learn region-specific weights for multiple seismic factors, then the detection accuracy is improved, but the computational complexity and processing time increase
Solution Approach 1:
The patent performs preliminary action by pre-processing seismic data into multiple factor-specific cubes (low frequency anomaly, flat spot, bright spot, fracture) before applying the unsupervised machine learning clustering models, which streamlines the subsequent weight learning and aggregation processes
Solution Approach 2:
The patent employs a universal unsupervised machine learning clustering model framework that can process multiple different seismic factors simultaneously, learning region-specific weights for each factor through the same clustering algorithm, thereby improving detection accuracy without proportionally increasing system complexity
Data Source
AI summary
Seismic data for a subsurface region is obtained. Individually, for each factor of multiple factors, a corresponding set of factor cubes specific to the factor is generated to obtain sets of factor cubes. Each factor cube includes cells having a value for the factor cube that is for a particular location in the subsurface region. An unsupervised machine learning clustering model is executed on the sets of factor cubes to determine a corresponding weight for each factor. According to the corresponding weight, the sets of factor cubes are aggregated to generate an aggregated cube, which is presented.