Seismic Noise Attenuation via Local Window Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional noise suppression methods in seismic data struggle to effectively estimate and remove spatially and temporally varying noise, often leading to erroneous suppression of signals in areas with strong signals and weak noise.
Innovation Solution
A computer-implemented method that transforms seismic data into a sparse or compressible domain, divides it into windows representing spatio-temporal locations, determines statistics for each window, applies a filter based on those statistics, and performs an inverse transform to create noise-attenuated data, allowing for variable noise suppression across space and time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If conventional noise suppression methods normalize amplitudes across seismic data using Automatic Gain Control (AGC), then noise attenuation is applied uniformly, but signal is erroneously suppressed in areas with strong signal and weak noise
Solution Approach 1:
The patent applies different noise attenuation filters to different spatial windows across the seismic data. Each window receives a filter customized to its local noise characteristics, allowing strong signals in low-noise areas to be preserved while still attenuating noise in high-noise regions. This local adaptation prevents the uniform suppression caused by conventional AGC methods.
Solution Approach 2:
The seismic data is divided into multiple spatial windows, and noise attenuation is performed independently on each window. This segmentation allows the system to adapt to spatially varying noise levels and prevent erroneous signal suppression in regions where noise is weak but signals are strong.
2Adaptability or versatility
If conventional methods apply uniform noise suppression across the entire seismic dataset, then processing is simple, but they cannot accommodate spatial and temporal variation in noise levels
Solution Approach 1:
The patent segments the seismic data into multiple spatial windows and applies different attenuation filters to each window based on local noise statistics. This segmentation enables adaptation to spatial and temporal noise variations while maintaining a systematic processing framework that manages complexity through modular operation on smaller data segments.
Solution Approach 2:
The noise attenuation filter is made dynamic by computing noise statistics separately for each spatial window and temporal segment. This allows the filter characteristics to adapt dynamically to changing noise conditions across different locations and times, rather than applying a static uniform filter throughout the entire dataset.
3Object-affected harmful factors
If noise attenuation is strengthened to remove more noise, then noise levels decrease, but signal loss increases in areas with strong signals
Solution Approach 1:
The patent computes noise statistics and applies attenuation filters locally to each spatial window rather than uniformly across the entire dataset. This allows strong signals in low-noise areas to be preserved with minimal attenuation, while still applying stronger noise suppression in regions where noise levels are high, thereby reducing overall signal loss.
Solution Approach 2:
The patent applies noise attenuation selectively and partially - only to the extent necessary in each local region. By computing noise statistics for each window and applying appropriate filter strength locally, the system avoids excessive attenuation that would cause signal loss, while still achieving effective noise removal where needed.
Data Source
AI summary
A system and method for attenuating noise in seismic data representative of a subsurface region of interest including receiving the seismic data; transforming the seismic data into a domain wherein the seismic data have a sparse or compressible representation to create transformed seismic data; dividing the domain into windows wherein the windows represent known spatio-temporal locations in the seismic data; determining statistics of the transformed seismic data in each window; determining a filter for each window based on the statistics of the transformed data; applying the filter for each window to the transformed seismic data in each window to create filtered seismic data; and performing an inverse transform of the filtered seismic data to create noise-attenuated seismic data.