Seismic Data Coherent Noise Attenuation via Radon Transform
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for attenuating coherent noise in marine seismic data acquisition, such as those generated by man-made devices, often result in unsatisfactory outcomes due to the attenuation of valuable seismic information and the introduction of artifacts, particularly when the noise position is known.
Innovation Solution
A method involving the generation of a model in a second domain to describe seismic data, allowing for the simultaneous derivation of models for both coherent noise and pure seismic signals, using a linear least squares Radon formulation, which enables the separation and removal of coherent noise without affecting the seismic data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If traditional frequency-wavenumber (FK) dip filtering or tau-p muting is used to remove coherent noise, then noise attenuation is achieved, but valuable seismic information is attenuated and artifacts are introduced
Solution Approach 1:
The patent segments the seismic data into distinct components (noise and signal) by generating separate models for each in the Radon domain. The coherent noise is modeled with a specific Radon transform that captures its linear chevron pattern, while the seismic signal is modeled separately. This segmentation allows selective removal of noise without affecting the signal.
Solution Approach 2:
The patent introduces the Radon transform as an intermediary domain to facilitate noise separation. By transforming the data into the Radon domain, the noise and signal become distinguishable based on their different Radon characteristics. The linear Radon model serves as an intermediary representation that enables precise noise identification and removal while preserving signal integrity.
2Object-affected harmful factors
If traditional FK dip filtering or tau-p muting is used to remove coherent noise, then noise attenuation is achieved, but artifacts are generated
Solution Approach 1:
The patent changes the domain parameters by using the linear Radon transform with specific parameters (slope and intercept) that characterize the coherent noise's chevron pattern. By adjusting the Radon parameters to match the noise characteristics, the method achieves precise noise modeling and removal without introducing artifacts that would compromise image accuracy.
3Ease of operation
If coherent noise is removed using methods that do not account for known noise position, then processing is simpler, but noise attenuation effectiveness is reduced
Solution Approach 1:
The patent performs preliminary action by incorporating the known noise position information into the Radon model generation step. The linear Radon model is configured with parameters that reflect the known position and characteristics of the coherent noise source. This preliminary incorporation of position information enhances noise attenuation effectiveness while maintaining the overall simplicity of the processing workflow.
Data Source
AI summary
Computing device, computer instructions and method for denoising input seismic data d. The method includes receiving the input seismic data d recorded in a first domain by seismic receivers, wherein the input seismic data d includes pure seismic data ss relating to an exploration source and coherent noise data n generated by a man-made device; generating a model m in a second domain to describe the input seismic data d; and processing the model m to obtain an output seismic dataset d′ indicative of seismic data substantially free of the coherent noise data n generated by the man-made device.