Near-surface P-velocity Estimation via Predictive Deconvolution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for building a near-surface P-velocity model are challenging due to masking by noise, contamination by multiple reflections, and missing near-offset traces, often requiring an initial velocity model that may not be available.
Innovation Solution
A method using multidimensional predictive deconvolution on seismic data to generate synthetic gathers, which are then used to create a velocity model that maps near-surface velocities to subsurface layers without relying on an initial velocity model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If classical methods (first break picking, dispersion curves picking, multi-wave inversion) are used to build near-surface velocity model, then S-velocity can be obtained, but P-velocity cannot be obtained and the process is challenging due to velocity inversions
Solution Approach 1:
The patent replaces classical mechanical picking and inversion methods with a seismic wavefield-based approach using recorded seismic data and synthetic gathers. Instead of manually or algorithmically picking first breaks and performing complex inversions, the method uses seismic wave propagation characteristics to directly estimate P-velocity, substituting a more effective physical approach for the conventional mechanical process.
Solution Approach 2:
The patent changes the fundamental parameter being measured from S-velocity (shear wave velocity) to P-velocity (compressional wave velocity). By using seismic data that contains P-wave energy and processing it through predictive deconvolution to generate synthetic gathers, the method directly targets P-velocity estimation, avoiding the limitation of classical methods that only yield S-velocity.
2Measurement precision
If near-surface primary reflections are used for velocity analysis, then direct identification is possible, but they are masked by strong noises and contaminated by strong multiple reflections
Solution Approach 1:
The patent extracts the useful signal from the noisy seismic data by applying predictive deconvolution to generate synthetic gathers. This process separates the primary reflection information from the harmful noises and multiple reflections, isolating the clean signal needed for velocity analysis without requiring manual filtering or picking.
Solution Approach 2:
The patent introduces synthetic gathers as an intermediary between the raw noisy seismic data and the final velocity model. The synthetic gathers, generated through predictive deconvolution, serve as a clean intermediate representation that contains the necessary velocity information without the contamination of surface waves, guided waves, and multiple reflections present in the original data.
3Measurement precision
If elastic Full Waveform Inversion is used to estimate P-velocity model, then P-velocity can be obtained, but it needs a reasonably correct initial velocity model which is often not available
Solution Approach 1:
The patent performs preliminary processing of the seismic data through predictive deconvolution to generate synthetic gathers before velocity analysis. This preliminary action prepares the data in a form that directly enables P-velocity estimation without requiring an initial velocity model, eliminating the need for the complex iterative process of Full Waveform Inversion and its associated initial model requirements.
4Quantity of substance
If near-offset traces are present in acquisition geometry, then complete data coverage is achieved, but acquisition constraints such as obstacles may cause missing near-offset traces
Solution Approach 1:
The patent uses the available seismic data itself to generate the synthetic gathers needed for velocity analysis, without requiring complete near-offset coverage. The predictive deconvolution process works with the recorded data to create the necessary synthetic information, allowing the method to be self-sufficient and not dependent on ideal acquisition geometry that may be constrained by obstacles.
Data Source
AI summary
A method for mapping near-surface velocities to layers of a subsurface includes receiving seismic data D associated with the subsurface, wherein the seismic data D includes at least one of P-wave energy, S-wave energy, or a mixture of P- and S-wave energy, applying a predictive deconvolution method to the seismic data D to calculate a synthetic gather F, of the subsurface, and generating a velocity model of the subsurface based on the synthetic gather F, where the velocity model maps near-surface velocities to the layers of the subsurface. A prediction deconvolution operator of the predictive deconvolution method, which corresponds to the synthetic gather F with changed sign, is a Green's function of the subsurface without any free surface.