HDFD Seismic Trace Segmentation for High-Resolution Visualization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current seismic data interpretation techniques, such as Fourier Transform and Matching Pursuit, face limitations in time-frequency resolution, leading to loss of detail and increased complexity in models, especially in non-stationary signals and seismic data with interference effects.
Innovation Solution
The High Definition Frequency Decomposition (HDFD) method subdivides seismic traces into characteristic segments, generates analytical model functions using adapted wavelets, and optimizes these functions to match seismic data, preserving time domain resolution and reducing residual energy, allowing for improved frequency decomposition without low-pass filtering.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Fourier Transform or Matching Pursuit is used for frequency decomposition, then frequency analysis capability is improved, but time-frequency resolution is lost and computational complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the seismic trace into multiple characteristic segments based on envelope peaks and troughs. Each segment is then independently analyzed using analytical model functions, allowing preservation of both time and frequency information locally rather than losing temporal resolution in favor of global frequency analysis.
Solution Approach 2:
The patent changes the parameter representation by using analytical model functions with adjustable parameters (amplitude, frequency, phase, duration) to fit each seismic segment. This allows the model to adapt to local characteristics while maintaining computational efficiency, resolving the trade-off between detailed frequency analysis and time-frequency resolution.
2Loss of information
If traditional frequency decomposition techniques are applied, then frequency information is extracted, but original time domain resolution is degraded due to low-pass filtering
Solution Approach 1:
The patent performs preliminary segmentation of the seismic trace into characteristic segments before applying frequency decomposition. By identifying envelope peaks and troughs beforehand, the method prepares the data structure to maintain temporal boundaries throughout the decomposition process, preventing the time domain resolution degradation that occurs with traditional low-pass filtering approaches.
Solution Approach 2:
The patent employs dynamic analytical model functions that adapt to each seismic segment's characteristics. The model parameters (frequency, amplitude, phase) are optimized for each segment rather than applying a static low-pass filter, allowing frequency information extraction while preserving the dynamic temporal structure of the original signal.
3Measurement precision
If detailed frequency decomposition is performed, then geological feature visualization is improved, but computational cost increases
Solution Approach 1:
The patent segments the seismic trace into characteristic segments bounded by envelope peaks and troughs. This segmentation reduces the computational burden by limiting the application of complex analytical model fitting to manageable segments rather than processing the entire trace uniformly, thereby improving computational efficiency while maintaining visualization quality.
Solution Approach 2:
The patent applies frequency decomposition selectively to characteristic segments identified by envelope analysis rather than processing every portion of the seismic data with equal detail. This partial action approach focuses computational resources on segments containing significant geological information, achieving good visualization quality without the excessive computational cost of uniform detailed decomposition.
Data Source
AI summary
Visually enhancing a geological feature in 3D seismic survey data may include selecting a first seismic trace from a 3D seismic survey dataset. Said first seismic trace is subdivided into a plurality of identified characteristic segments. At least one first analytical model function is generated for each of said plurality of identified characteristic segments. At least one adapted wavelet from an existing dictionary is utilized. A matching characteristic is determined between said first seismic trace and said at least one first analytical model function. Said at least one first analytical model function is optimized with respect to said matching characteristic. Both determining a matching characteristic, and optimizing said at least one first analytical model function, are repeated until a predetermined condition is met. A model dataset is generated from said optimized at least one first analytical model function for at least part of said first seismic trace for visual representation.