Amplitude Independent Gradient for Seismic Velocity Inversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing seismic velocity inversion methods are either time-consuming or unreliable, particularly in complex subsurface structures, due to manual data picking, cycle skipping, and artifacts in gradient calculations.
Innovation Solution
A computer-implemented method for computing amplitude-independent gradients in the frequency domain, iteratively updating seismic velocities using a cross-correlation based objective function, which separates amplitude and traveltime information to quickly converge to true velocities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ray tomography is used for velocity inversion, then traveltime information can be utilized, but manual picking of arrival events is time-consuming and makes three-dimensional model building unaffordable
Solution Approach 1:
The patent replaces manual mechanical picking of arrival events with automated computational methods. The system uses cross-correlation based misfit functions and gradient computations to automatically determine traveltime information from seismic data, eliminating the need for manual event picking while maintaining velocity inversion accuracy
Solution Approach 2:
The velocity inversion system performs self-service by automatically processing seismic data without manual intervention. The algorithm autonomously computes gradients, updates velocity models, and converges to solutions through iterative optimization, making the process independent of human operators
2Loss of information
If full waveform inversion is used, then comprehensive waveform information is utilized, but cycle skipping problem causes convergence to local minimum
Solution Approach 1:
The patent extracts only the necessary traveltime information from full waveforms through cross-correlation, separating it from amplitude and other potentially misleading information. This extraction of essential traveltime data avoids the cycle skipping problem while maintaining effective use of waveform information for velocity inversion
Solution Approach 2:
The patent introduces cross-correlation as an intermediary between raw seismic waveforms and velocity inversion. This intermediary process transforms complex waveform data into reliable traveltime measurements, mediating the information flow to prevent direct waveform complexities from causing convergence issues
3Measurement precision
If cross-correlation based misfit function is used, then traveltime information is effectively utilized, but significant artifacts in gradients slow down convergence rate
Solution Approach 1:
The patent segments the gradient computation process into distinct components, separating the useful traveltime information from artifacts. By computing gradients specifically from traveltime residuals rather than full cross-correlation outputs, the method isolates and eliminates artifact-containing components while preserving convergence-accelerating elements
Data Source
AI summary
A method for computing amplitude independent gradient for seismic velocity inversion in a frequency domain includes receiving seismic data associated with a region. The region comprises one or more earth subsurface layers represented by a plurality of points, and each point is associated with a seismic velocity. Seismic velocities at the plurality of points are determined by iteratively updating the seismic velocities based on a plurality of gradient values, where each gradient value corresponds to a point and is determined by evaluating a gradient of an objective function at a location of the point. A seismic image of the one or more earth subsurface layers is displayed based on the determined seismic velocities.