A computer system manages model information for defining a U-Net configured to execute, on the input time-series data, an encoding operation for extracting a feature map relating to the target wave by using downsampling blocks and a decoding operation for outputting data for predicting the first motion time of the target wave by using upsampling blocks, executes the encoding operation and the decoding operation on the input time-series data by using the model information. The downsampling blocks and the upsampling blocks each includes a residual block. The residual block includes a time attention block calculates a time attention for emphasizing a specific time domain in the feature map. The time attention block includes an arithmetic operation for calculating attentions different in time width, and calculates a feature map to which the time attention is added by using the attentions.

A method and apparatus for identifying features of a subsurface region, including: obtaining an initial physical property model and survey data for the subsurface region; identifying a current model to be the initial physical property model; and executing one or more iterations of: generating synthetic data and forward wavefields with the current model and the survey data by forward modeling with forward wave equations representing isotropic wave-mode-independent attenuation; generating adjoint wavefields with the synthetic data and the survey data by adjoint modeling with adjoint wave equations representing isotropic wave-mode-independent attenuation; computing an objective function gradient with the forward wavefields and the adjoint wavefields by solving gradient equations with the corresponding wave equations representing isotropic wave-mode-independent attenuation; computing a search direction of the objective function; searching for a possible improved model along the search direction; and updating the current model to be the possible improved model.

A method may include obtaining seismic data regarding a geological region of interest. The method may further include obtaining a property model regarding the geological region of interest. The method may further include determining an adjoint migration operator based on the property model. The method may further include updating the property model using the seismic data and a conjugate gradient solver in a least-squares reverse time migration to produce a first updated property model. The conjugate gradient solver is based on the adjoint migration operator. The method may further include updating the first updated property model using a threshold shrinkage function to produce a second updated property model. The threshold shrinkage function comprises a sign function and a maximum function that are applied to the first updated property model. The method may further include generating a seismic image of the geological region of interest using the second updated property model.

A system for seismic imaging of a subterranean geological formation uses a two-way imaging condition. A seismic signal is emitted into a subterranean formation and recorded at receiver(s). Source and receiver wavefields are decomposed into respective right-down/left-up and left-down/right-up propagating waves. The right-down/left-up and left-down/right-up direction can be defined along the direction emitted from the source or receiver to corresponding direction in two dimensional (2D) case. An imaging condition for generating both a positive-dip structure image and a negative-dip structure image is the inner product of the wavefields. Applying the sample-by-sample multiplication imaging condition to the opposite dip images, the diffraction energy is retained while the reflection energy is significantly attenuated. The diffraction image can be used to detect faults and fractures in subsurface regions.

The present disclosure includes a method for combining controlled and uncontrolled seismic data. The method includes accessing one or more controlled signals, each controlled signal associated with a respective receiver of a plurality of receivers. The method also includes accessing one or more uncontrolled signals, each uncontrolled signal associated with a respective receiver of the plurality of receivers. The method also includes generating one or more reconstructed signals based on the one or more uncontrolled signals. The method also includes generating a composite image based at least on the one or more controlled signals and the one or more reconstructed signals. The present disclosure may also include associated systems and apparatuses.

Techniques to avoid a cycle skip in conjunction with a full waveform inversion are disclosed herein. A method includes selecting a first objective function of a full waveform inversion (FWI) from a set of objective functions, selecting a second objective function of the FWI from the set of objective functions, calculating a first misfit based upon the first objective function using modeled data with respect to observed data, calculating a first search direction based upon the first misfit between the modeled data and the observed data, calculating a second misfit based upon the second objective function using the modeled data with respect to the observed data, calculating a second search direction based upon the second misfit between the modeled data and the observed data, combining the first search direction with the second direction and computing an update to the modeled data based upon the first search direction and the second search direction combination.

A method for seismic imaging of a subsurface using a full waveform inversion, FWI, includes the steps of obtaining an initial P-velocity model that describes how a seismic wave propagates through the subsurface, generating an S-velocity model based on the initial P-velocity model, generating an elastic synthetic dataset p based on the initial P-velocity model and the S-velocity model, receiving a recorded seismic dataset d of the subsurface, updating the P-velocity model based on a comparison between the synthetic dataset p and the recorded seismic dataset d, updating a reflectivity at various locations in the subsurface based on the updated P-velocity model, and generating an image of the subsurface based on the reflectivity.

A method for seismic exploration uses visco-acoustic FWI to model velocity and quality factor Q for an explored subsurface formation. The method employs frequency-dependent velocity to reduce cross-talk between Q and velocity and may be used for both isotropic and anisotropic media.

A method for generating a seismic image representing a subsurface includes receiving seismic data for the subsurface formation, including receiver wavelet data and source wavelet data. Source wavefield data are generated based on a forward modeling of the source wavelet data. Receiver wavefield data are generated that compensate for distortions in the seismic data by: applying a dispersion-only model to the receiver wavelet data to generate a first reconstructed back-propagated receiver wavefield portion, applying a dissipation-only model to the receiver wavelet data to generate a second reconstructed back-propagated receiver wavefield portion, and combining the first back-propagated receiver wavefield portion and the second back-propagated receiver wavefield portion into the receiver wavefield data. The method includes applying an imaging condition to the receiver wavefield data and the source wavefield data and generating, based on applying the imaging condition, visco-acoustic reverse time migration (VARTM) result data.

A method may include obtaining an input gather regarding a geological region of interest. The method may further include obtaining parameterization data regarding a seismic processing operation. The parameterization data may correspond to a first set of process parameter values that are different from a second set of process parameter values that are used to generate the input gather. The method may further include generating a predicted output gather using a machine-learning model, the input gather, and the parameterization data. The machine-learning model may include an encoder model and a decoder model. The method may further include generating a seismic image of the geological region of interest using the predicted output gather.