Seismic Facies Classification Using Probabilistic Algorithms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current seismic facies classification methods are time-consuming, provide limited data resolution, and often generate unrealistic results due to reliance on well log data, which can distort subsurface feature representation, and fail to effectively utilize 3D seismic data for accurate geological feature mapping.
Innovation Solution
A method that generates attribute volumes and frequency decomposition colour blend volumes from 3D seismic data, using probabilistic algorithms to create a facies classification model dataset that optimizes seismic-driven facies classification, allowing for high-resolution geological feature representation even without well log data, and preserving the resolution of the original seismic data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If well log data is used for facies classification, then classification can be performed with limited data, but the resolution and accuracy of subsurface feature representation deteriorates
Solution Approach 1:
The patent replaces the traditional well log data-driven classification approach with a seismic attribute-based classification system. Instead of relying on mechanical drilling and logging operations to obtain subsurface data, the invention uses seismic wave propagation characteristics and derived attributes (frequency, amplitude, phase, impedance) to classify facies, thereby substituting a less invasive measurement system that provides higher spatial resolution across the entire survey area rather than only at well locations
Solution Approach 2:
The invention transforms the classification approach by changing from using well log parameters (limited to discrete well locations) to using seismic attribute parameters (available across continuous 3D space). The patent applies spectral decomposition to generate frequency-based attributes and uses these transformed parameters to create facies classification, thereby changing the parameter space from sparse well data to dense seismic data while improving measurement precision
2Productivity
If traditional facies classification methods are used, then the process can be completed with existing tools, but the time required and computational resources increase significantly
Solution Approach 1:
The patent performs preliminary spectral decomposition and attribute generation before the actual facies classification process. By pre-computing frequency-based attributes and organizing the 3D seismic data into structured attribute volumes, the invention prepares the data in advance to enable faster classification execution. This preliminary processing of seismic attributes allows the classification algorithm to work with pre-processed, optimized data structures rather than raw seismic traces
Solution Approach 2:
The invention replaces traditional iterative classification methods with a seismic attribute-based approach that leverages the inherent spatial and spectral information in 3D seismic data. By using frequency decomposition attributes and impedance variations as direct classification inputs, the system eliminates time-consuming intermediate processing steps and achieves faster classification while maintaining geological realism
3Reliability
If well log data is relied upon for facies classification, then the method can handle data scarcity, but unrealistic results are generated due to distortion of subsurface features
Solution Approach 1:
The patent substitutes the well log data acquisition system with a seismic-based classification system that preserves subsurface feature integrity. Instead of relying on discrete well measurements that may distort continuous subsurface structures, the invention uses continuous seismic attribute fields (frequency, amplitude, phase, impedance) that naturally represent the true spatial distribution of geological features without the sampling artifacts inherent in well log data
Solution Approach 2:
The invention applies local quality analysis by using frequency decomposition to identify subtle lateral variations in seismic attributes that correspond to different facies types. The spectral decomposition method examines frequency content at each location independently, allowing the classification to capture local geological heterogeneity and maintain realistic representation of subsurface features with varying spatial scales
Data Source
AI summary
The present invention describes a method for adaptively determining a plurality of sedimentary facies from 3D seismic data, comprising the steps of (a) generating an attribute volume comprising at least one attribute from said 3D seismic data; (b) generating at least one frequency decomposition color blend volume from said 3D seismic data; (c) generating a data volume comprising at least one geological object utilizing data from said attribute volume and said frequency decomposition color blend volume; (d) generating a facies classification model dataset for a predetermined region of interest of said 3D seismic data applying a probabilistic algorithm and utilizing data from said geobody volume and said frequency decomposition color blend volume; (e) selectively adjusting at least one first model parameter, so as to optimize said facies classification model dataset in accordance with a conceptual geological model; and (f) selectively providing said facies classification model dataset in a representative property model of said region of interest of said 3D seismic data.