Iterative First-Break Picking for Seismic Velocity Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Seismic data processing is hindered by geometry errors and noisy data due to incorrect source or receiver placement and near-surface structures, leading to distorted images of subsurface formations, which existing methods struggle to accurately correct and model.
Innovation Solution
An iterative approach using AI, ML, CNNs, and RL for quality control and geometry correction, combined with machine learning for identifying and correcting geometry errors, trace editing, and statics optimization, to improve first-break picking and seismic imaging.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional seismic data processing methods are used, then processing speed is maintained, but measurement precision and reliability of first-break picking deteriorate due to geometry errors and noisy data
Solution Approach 1:
The patent introduces an intermediary quality control system that mediates between raw seismic data and the final velocity model. This intermediary layer performs automated geometry error detection, noise filtering, and first-break picking validation using machine learning algorithms, thereby improving measurement precision without requiring complete redesign of the processing system.
Solution Approach 2:
The quality control system employs self-service mechanisms through automated algorithms that independently identify and correct geometry errors, filter noisy traces, and validate first-break picks without continuous human intervention. The system uses self-learning capabilities to improve its performance over time, reducing the need for manual calibration while maintaining high accuracy.
2Reliability
If iterative quality control and geometry correction are applied, then reliability of subsurface imaging is improved, but processing time increases
Solution Approach 1:
The patent applies preliminary action by performing quality control checks and geometry corrections in advance before the main velocity model building process. Automated algorithms pre-identify and correct geometry errors, pre-filter noisy data, and pre-validate first-break picks, ensuring that the subsequent imaging process operates on cleaned data and thereby improving reliability without adding significant time to the overall workflow.
Solution Approach 2:
The system implements feedback mechanisms where the results of velocity model building are fed back into the quality control process. This iterative feedback loop allows the system to automatically adjust and refine its quality control parameters based on actual imaging results, progressively improving subsurface imaging reliability while optimizing processing efficiency through learned patterns.
3Productivity
If machine learning algorithms are used for automated first-break picking, then productivity is improved, but measurement precision may deteriorate due to algorithm errors
Solution Approach 1:
The patent incorporates feedback loops where machine learning algorithms for automated first-break picking are continuously validated against quality control metrics. The system monitors picking results, identifies outliers and errors, and uses this feedback to refine algorithm parameters and improve accuracy over time, ensuring that productivity gains do not compromise measurement precision.
Solution Approach 2:
An intermediary validation layer is introduced between the machine learning picking algorithm and the final velocity model. This intermediary performs automated quality checks, compares picks against geometric constraints, and flags suspicious results for review, thereby maintaining high productivity while safeguarding against algorithmic errors that could reduce measurement precision.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This method enhances the accuracy of first-break picking and seismic imaging by iteratively refining picks and geometry corrections, reducing noise and distortion, and providing a more reliable subsurface velocity model, effectively addressing the challenges of near-surface complexities and data quality issues.
Implementation Method 1
Seismic waves travel through the subsurface of the earth at different speeds based on the material they are traveling through. As the seismic waves encounter different types of rock or soil, the waves can be partially refracted and partially reflected.
Data Source
AI summary
A new method for iteratively picking the seismic first breaks and conducting imaging of the near-surface velocity structures in an iterative fashion is provided that the first-break picks of the input seismic data are applied to image the near-surface velocity structures and the calculated travel times associated with the updated velocity structures are applied to help refine the first-break picks in the first break picking process until first-break picks satisfy a number of quality control criteria, statics solutions are optimized, and the near surface imaging reaches an acceptable data misfit. This invention produces a velocity model that can be used for near surface statics corrections or for the prestack depth migration.