Back to results
US 11,448,784 B2Granted
Full waveform inversion using time delayed seismic data
AssigneeSaudi Arabian Oil Company
InventorsJewoo Yoo, Roald van Borselen
Life of this patent
Estimating the term…
Abstract
Systems, methods, and apparatuses directed to determining a long wavelength velocity model using acquired seismic data lacking low-frequency data is disclosed. Determination of the long wavelength velocity model may include generating a time-delayed signal from the acquired seismic data to produce low-frequency information and reducing a residual energy between the time-delayed signal from the acquired seismic data and a time-delayed signal of modeled data using an objective function to produce an optimized initial velocity model using full waveform inversion. Moreover, a full waveform inversion on the optimized initial velocity model using acquired data can be used to produce a velocity model more accurately representing subterranean formations.
Claims13
A computer program product encoded on a non-transitory medium, the product comprising computer readable instructions for causing one or more processors to perform operations comprising:
- receiving a set of observed seismic data;
- receiving an initial velocity model and an initial source wavelet, the initial velocity model being a user-selected velocity model and the initial source wavelet being a user-selected wavelet;
- performing a wavefield simulation using the initial source wavelet and the initial velocity model to generate modeled seismic data;
- converting the received set of observed seismic data into a first time-delayed signal;
- converting the modeled seismic data into a second time-delayed signal;
- optimizing the initial velocity model containing long wavelength subterranean formations using a full waveform inversion utilizing the first time-delayed signal and the second time-delayed signal;
- and
- performing a full waveform inversion utilizing observed data and modeled data on the optimized initial velocity model to produce a detailed velocity model that more accurately represents subterranean formations;
- wherein optimizing the initial velocity model containing long wavelength subterranean formations using a full waveform inversion utilizing the first time-delayed signal and the second time-delayed signal comprises reducing a residual energy between the first time-delayed signal and the second time-delayed signal;
- and
- wherein reducing a residual energy between the first time-delayed signal and the second time-delayed signal comprises:
- applying an objective function to the first time-delayed signal and the second time-delayed signal to determine the residual energy;
- determining whether the residual energy satisfies a predetermined condition;
- and
- if the residual energy does not meet the predetermined condition, repeating an iterative loop until an occurrence of the predetermined condition or until a selected number of iterative loops is completed, the iterative loop comprising:
- taking a partial derivative of the objective function with respect to velocity to determine a directional change of the velocity;
- updating the initial velocity model according to the determined directional change of velocity to form a new velocity model;
- updating the initial source wavelet via a source estimation algorithm by minimizing a difference between a modeled green-function and the received set of observed seismic data forming a new source wavelet;
- performing a wavefield simulation with the new source wavelet and the new velocity model to generate new modeled seismic data;
- converting the received set of observed seismic data into a new first time-delayed signal;
- converting the new modeled seismic data into a new second time-delayed signal;
- applying the objective function to the new first time-delayed signal and the new second time-delayed signal to determine a new residual energy;
- and
- determining whether the new residual energy satisfies the predetermined condition.
21 more elements
A computer implemented method performed by one or more processors for automatically generating a long wavelength velocity model that more accurately represents subterranean formations, the method comprising:
- receiving a set of observed seismic data;
- receiving an initial velocity model and an initial source wavelet, the initial velocity model being a user-selected velocity model and the initial source being a user-selected wavelet;
- performing a wavefield simulation using the initial source wavelet and the initial velocity model to generate modeled seismic data;
- converting the received set of observed seismic data into a first time-delayed signal;
- converting the modeled seismic data into a second time-delayed signal;
- optimizing the initial velocity model containing long wavelength subterranean formations using full waveform inversion utilizing the first time-delayed signal and the second time-delayed signal;
- and
- performing a full waveform inversion utilizing observed data and modeled data on the optimized initial velocity model to produce a detailed velocity model that more accurately represents subterranean formations;
- wherein optimizing the initial velocity model utilizing the first time-delayed signal and the second time-delayed signal comprises reducing a residual energy between the first time-delayed signal and the second time-delayed signal;
- and
- wherein reducing a residual energy between the first time-delayed signal and the second time-delayed signal comprises:
- applying an objective function to the first time-delayed signal and the second time-delayed signal to determine the residual energy;
- determining whether the residual energy satisfies a predetermined condition;
- and
- if the residual energy does not meet the predetermined condition, repeating an iterative loop until an occurrence of the predetermined condition or until a selected number of iterative loops is completed, the iterative loop comprising:
- taking a partial derivative of the objective function with respect to velocity to determine a directional change of the velocity;
- updating the initial velocity model according to the determined directional change of velocity to form a new velocity model;
- updating the initial source wavelet via a source estimation algorithm by minimizing a difference between a modeled green-function and the received set of observed seismic data forming a new source wavelet;
- performing a wavefield simulation with the new source wavelet and the new velocity model to generate new modeled seismic data;
- converting the received set of observed seismic data into a new first time-delayed signal;
- converting the new modeled seismic data into a new second time-delayed signal;
- applying the objective function to the new first time-delayed signal and the new second time-delayed signal to determine a new residual energy;
- and
- determining whether the new residual energy satisfies the predetermined condition.
21 more elements
An apparatus for generating a long wavelength velocity model that more accurately represents subterranean formations, the apparatus comprising:
- one or more processors;
- and
- a non-transitory computer-readable storage medium coupled to the one or more processors and storing programming instructions for execution by the one or more processors, the programming instructions instruct the one or more processors to:
- receive a set of observed seismic data;
- receive an initial velocity model and an initial source wavelet, the initial velocity model being a user-selected velocity model and the initial source wavelet being a user-selected wavelet;
- perform a wavefield simulation using the initial source wavelet and the initial velocity model to generate modeled seismic data;
- convert the received set of observed seismic data into a first time-delayed signal;
- convert modeled seismic data into a second time-delayed signal;
- optimize the initial velocity model containing long wavelength subterranean formations using full waveform inversion utilizing the first time-delayed signal and the second time-delayed signal;
- and
- performing a full waveform inversion utilizing observed data and modeled data on the optimized initial velocity model to produce a detailed velocity model that more accurately represents subterranean formations;
- reducing a residual energy between the first time-delayed signal and the second time-delayed signal;
- and
- wherein reducing a residual energy between the first time-delayed signal and the second time-delayed signal comprises:
- applying an objective function to the first time-delayed signal and the second time-delayed signal to determine the residual energy;
- determining whether the residual energy satisfies a predetermined condition;
- and
- if the residual energy does not meet the predetermined condition, repeat an iterative loop until an occurrence of the predetermined condition or until a selected number of iterative loops is completed, the iterative loop comprising:
- taking a partial derivative of the objective function with respect to velocity to determine a directional change of the velocity;
- updating the initial velocity model according to the determined directional change of velocity to form a new velocity model;
- updating the initial source wavelet via a source estimation algorithm by minimizing a difference between a modeled green-function and the received set of observed seismic data forming a new source wavelet;
- performing a wavefield simulation with the new source wavelet and the new velocity model to generate new modeled seismic data;
- converting the received set of observed seismic data into a new first time-delayed signal;
- converting the new modeled seismic data into a new second time-delayed signal;
- applying the objective function to the new first time-delayed signal and the new second time-delayed signal to determine a new residual energy;
- and
- determining whether the new residual energy satisfies the predetermined condition.
24 more elements
Dependencies are inferred from the claim text.