Dictionary Learning for Seismic Deblending
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional deblending techniques struggle to effectively separate signal and noise from blended seismic data, often requiring manual parameter adjustments and failing to distinguish between signal and noise without additional classification algorithms, leading to inefficient data processing and representation.
Innovation Solution
The method employs dictionary learning to generate separate dictionaries for signal and noise atoms, allowing for the creation of a combined dictionary that can represent blended data, enabling automated separation of signal and noise through sparse reconstruction and atom usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional deblending techniques transform blended shot data into common receiver domain, then signal from one series of shots becomes coherent, but signals from other blended shots become blending noise requiring coherency filtering
Solution Approach 1:
The patent segments the blended seismic data into separate signal and noise components by transforming into both shot domain and receiver domain, then applying dictionary learning to identify and separate coherent signal atoms from incoherent noise atoms in each domain
Solution Approach 2:
The patent changes the domain parameters by performing transformations in both shot domain and receiver domain, and uses dictionary learning with adjustable atom parameters to adaptively separate signal from noise based on coherence characteristics
2Measurement precision
If conventional deblending techniques perform coherency filtering in transformed domain, then signal separation is achieved, but manual parameter adjustments are required to determine impact on results
Solution Approach 1:
The patent implements self-service by using dictionary learning algorithms that automatically learn optimal signal and noise dictionaries from the data itself, eliminating the need for manual parameter adjustments and providing automated signal-noise separation
3Extent of automation
If artificial intelligence techniques use dictionary learning on blended data, then automated processing is achieved, but the technique cannot distinguish between signal and noise without additional classification algorithms
Solution Approach 1:
The patent adds another dimension by performing dictionary learning in both shot domain and receiver domain, allowing the system to distinguish signal from noise by comparing coherence characteristics across multiple domains without requiring additional classification algorithms
4Productivity
If blended data acquisition is used, then cost-effective data collection is achieved, but deblending processing is required to separate signal from noise
Solution Approach 1:
The patent segments the deblending process into distinct steps: transformation to shot and receiver domains, dictionary learning in each domain, and combination of results, making the complex processing more manageable and efficient
Solution Approach 2:
The patent uses parameter changes through domain transformations and dictionary learning parameters to efficiently separate signal from noise in blended data, reducing processing complexity compared to traditional methods
Data Source
AI summary
Systems and methods include a method for deblending signal and noise data. A shot domain for actual sources, a receiver domain for virtual sources, and a receiver domain for actual sources are generated from blended shot data. A dictionary of signal atoms is generated. Each signal atom includes a small patch of seismic signal data gathered during a small time window using multiple neighboring traces. A dictionary of noise atoms is generated. Each noise atom includes a small patch of seismic noise data gathered during a small time window using multiple neighboring traces. A combined signal-and-noise dictionary is generated that contains the signal atoms and the noise atoms. A sparse reconstruction of receiver domain data is created from the combined signal-and-noise dictionary. The sparse reconstruction is split into deblended data and blending noise data based on atom usage to create deblended shot domain gathers for actual sources.