Abstract: This specification describes workflows for, but is not limited to, performing full waveform inversion (FWI) to build high resolution velocity models to improve the accuracy of seismic imaging of a subterranean formation. This specification describes processes to automatically edit and enhance S/N quality of seismic data (such as land seismic data) to prepare the datasets for FWI. The methods for automatic corrections and pre-processing include: automatic iterative surface-consistent residual statics calculation, automatic rejection of anomalous traces (such as dead traces), and the automatic correction of surface-consistent amplitude anomalies (such as by scalar or deconvolution approaches). The operations include automatic “muting” of noise before first arrivals.

Abstract: A system and methods are disclosed for determining a model of a subterranean region. The method includes obtaining an observed dataset and a current model for the subterranean region, simulating a dataset from the current model, and determining a data penalty function based on a difference between the observed and simulated datasets. The method further includes training a machine learning (ML) network to predict a model from the observed dataset and determining the predicted model using the trained ML network. The method further includes determining a first model penalty function based on the current model, a second model penalty function based on a difference between the current and the predicted models, and a composite penalty function based on a weighted sum of the data penalty function, the first and the second model penalty functions. Finally, the method includes determining the model based on an extremum of a composite penalty function.

Abstract: A system and methods for determining a refined seismic model of a subterranean region are disclosed. The method includes obtaining an observed seismic dataset and a current seismic model for the subterranean region and training a machine learning (ML) network using seismic training models and corresponding seismic training datasets and predicting, using the trained ML network, a predicted seismic model from the observed seismic dataset. The method further includes determining a simulated seismic dataset from the current seismic model and a seismic wavelet, a data penalty function based on a difference between the observed and the simulated seismic datasets and a model penalty function from the difference between the current the predicted seismic models. The method still further includes determining the refined seismic model based on an extremum of a composite penalty function based on a weighted sum of the data penalty function and the model penalty function.

Abstract: A method may include obtaining various seismic traces for a geological region of interest. The method may further include determining an offset attribute and an azimuthal attribute. The method may further include determining, using the offset attribute and the azimuthal attribute, a virtual trace bin for the geological region of interest. The method may further include generating a virtual trace using a subset of the seismic traces and corresponding to the virtual trace bin. The method may further include generating a velocity model for the geological region of interest using a virtual shot gather including the virtual trace and various virtual traces. A respective virtual trace among the virtual traces may correspond to a respective virtual trace bin among various virtual trace bins. The method may further include generating a seismic image of the geological region of interest using the velocity model.

Abstract: Disclosed are methods, systems, and computer-readable medium to perform operations including: receiving for a plurality of common midpoint-offset bins each comprising a respective plurality of seismic traces, respective candidate pilot traces representing the plurality of common midpoint-offset bins; generating, based on the respective candidate pilot traces, a respective plurality of corrected seismic traces for each of the plurality of common midpoint-offset bins; grouping the respective pluralities of corrected seismic traces into a plurality of enhanced virtual shot gathers (eVSGs); generating, based on the plurality of common midpoint-offset bins, a common-midpoint (CMP) velocity model; calibrating the CMP velocity model using uphole velocity data to generate a pseudo-3 dimensional (3D) velocity model; performing, based on the plurality of enhanced virtual shot gathers and the pseudo-3D velocity model, a 1.

Abstract: In seismic imaging, accurate velocity functions (velocity model) defining seismic velocity as a function of depth in the earth are required. The velocity model is obtained as a result of seismic surveying. Delayed travel times in near surface refraction seismic surveys, an effect known as shingling, can result from an anomalous condition, seismic velocity decreasing with depth. Inclusion of such delayed travel times in a tomographic process for seismic imaging would otherwise cause large errors in determination of a seismic velocity model for seismic imaging of subsurface features. At locations (source-receiver offset) in the survey where the shingling occurs, the velocity inversions are identified. The undesirable effects the delayed travel times caused by the velocity inversions are removed from the survey dataset.

Abstract: An example method is for producing a seismic wave velocity model of a subsurface area. The method includes receiving, by a processor of a computing system, from a seismic receiver, seismic data input comprising a recorded seismic wave field. The method includes receiving, by the processor, an initial 1D velocity model of the subsurface area. The method includes determining, by the processor, a Laplace-Fourier transform of the recorded seismic wave field. The method includes regenerating, by the processor, the current 1D velocity model to generate inverted data representing the subsurface area. The method may include performing, by the processor, an upscaling of a plurality of 1D velocity models to produce a 3D velocity model.

Abstract: This specification describes workflows for, but is not limited to, performing full waveform inversion (FWI) to build high resolution velocity models to improve the accuracy of seismic imaging of a subterranean formation. This specification describes processes to automatically edit and enhance S/N quality of seismic data (such as land seismic data) to prepare the datasets for FWI. The methods for automatic corrections and pre-processing include: automatic iterative surface-consistent residual statics calculation, automatic rejection of anomalous traces (such as dead traces), and the automatic correction of surface-consistent amplitude anomalies (such as by scalar or deconvolution approaches). The operations include automatic “muting” of noise before first arrivals.

Abstract: Systems and methods are provided to correct seismic data for the undesired effects caused by near surface velocity variations. In one embodiment, a method includes receiving travel time data for a near surface region and estimating an initial velocity model for the near surface region using the travel time data. The method can include updating the velocity model by performing an inversion iteration of including inversion of travel times to estimate slowness. The process can also include calculating at least one long-wavelength static for the near surface region. The long-wavelength statics may be used to correct for undesired effects caused by near surface velocity variations.

Abstract: Methods for full waveform inversion (FWI) in the midpoint-offset domain include using a computer system to sort seismic traces into common midpoint-offset bins (XYO bins). For each XYO bin, a linear moveout correction is applied to a collection of seismic traces within the XYO bin. The collection of seismic traces is stacked to form a pilot trace. The computer system determines a surface-consistent residual static correction for each seismic trace. The computer system determines that the surface-consistent residual static correction for each seismic trace is less than a threshold. Responsive to the determining that the surface-consistent residual static correction is less than the threshold, the computer system stacks the collection of seismic traces to provide the pilot trace. The computer system groups the pilot traces for the XYO bins into a set of virtual shot gathers. The computer system performs one-dimensional FWI based on each virtual shot gather.

Abstract: Systems and methods are provided to correct seismic data for the undesired effects caused by near surface velocity variations. In one embodiment, a method includes receiving travel time data for a near surface region and estimating an initial velocity model for the near surface region using the travel time data. The method can include updating the velocity model by performing an inversion iteration of including inversion of travel times to estimate slowness. The process can also include calculating at least one long-wavelength static for the near surface region. The long-wavelength statics may be used to correct for undesired effects caused by near surface velocity variations.

Abstract: Seismic data from seismic exploration surveys are mapped into a hypercube of bins or voxels in a four-dimensional space (X, Y, Offset, and Azimuth) according to Common Mid-Point (or CMP) between source and receivers. The mapped data from individual voxels or bins is then analyzed by multimodal statistics. Robust estimates of first break picks are obtained from the analysis. The first break picks are then used to as seed inputs for autopicking iteration, which proceeds to convergence. Estimates of confidence levels in the data are provided for re-picking to reduce computer processing time in successive autopicking iterations. Analysis is provided of different seismic attributes such as azimuthal velocity variations indicative of anisotropy, positioning errors of sources/receivers, geometry errors, and three dimensional distribution of inversion residuals. Analysis is also performed of standard deviation of the travel time data useful for estimating data errors in the inversion covariance matrix.

Abstract: Seismic data from seismic exploration surveys are mapped into a hypercube of bins or voxels in a four-dimensional space (X, Y, Offset, and Azimuth) according to Common Mid-Point (or CMP) between source and receivers. The mapped data from individual voxels or bins is then analyzed by multimodal statistics. Robust estimates of first break picks are obtained from the analysis. The first break picks are then used to as seed inputs for autopicking iteration, which proceeds to convergence. Estimates of confidence levels in the data are provided for re-picking to reduce computer processing time in successive autopicking iterations. Analysis is provided of different seismic attributes such as azimuthal velocity variations indicative of anisotropy, positioning errors of sources/receivers, geometry errors, and three dimensional distribution of inversion residuals. Analysis is also performed of standard deviation of the travel time data useful for estimating data errors in the inversion covariance matrix.

Abstract: A deep learning framework includes a first model for predicting one or more attributes of a system; a second model for predicting one or more attributes of the system; at least one coupling operator combining the first and second models; and at least one inversion module for receiving the combined first and second models from the coupling operator. The inversion module simultaneously optimizes the first model and the second model, thereby resulting in a composite objective function representative of a prediction that is outputted to at least one user.

Abstract: Methods for full waveform inversion (FWI) in the midpoint-offset domain include using a computer system to sort seismic traces into common midpoint-offset bins (XYO bins). For each XYO bin, a linear moveout correction is applied to a collection of seismic traces within the XYO bin. The collection of seismic traces is stacked to form a pilot trace. The computer system determines a surface-consistent residual static correction for each seismic trace. The computer system determines that the surface-consistent residual static correction for each seismic trace is less than a threshold. Responsive to the determining that the surface-consistent residual static correction is less than the threshold, the computer system stacks the collection of seismic traces to provide the pilot trace. The computer system groups the pilot traces for the XYO bins into a set of virtual shot gathers. The computer system performs one-dimensional FWI based on each virtual shot gather.

Abstract: In some implementations, airborne electromagnetic (AEM) data and seismic data for a geographic region including sand dunes are received, and the AEM data identifies apparent resistivity as a function of depth within the sand dunes. An inversion with cross-domain regularization is calculated of the AEM data and the seismic data to generate a velocity-depth model, and the velocity depth model identifies velocity variations within the sand dunes. A seismic image using the velocity-depth model is generated.

Abstract: A method for refraction-based surface-consistent amplitude compensation and deconvolution includes receiving seismic traces, the seismic traces generated using at least one source and at least one receiver; calculating an amplitude residual for each seismic trace; determining surface-consistent amplitude residuals for the at least one source and the at least one receiver based on the amplitude residual for each seismic trace; and performing surface-consistent amplitude correction to each seismic trace by applying the determined surface-consistent amplitude residuals for the at least one source and the at least one receiver.

Abstract: In some implementations, airborne electromagnetic (AEM) data and seismic data for a geographic region including sand dunes are received, and the AEM data identifies apparent resistivity as a function of depth within the sand dunes. An inversion with cross-domain regularization is calculated of the AEM data and the seismic data to generate a velocity-depth model, and the velocity depth model identifies velocity variations within the sand dunes. A seismic image using the velocity-depth model is generated.

Abstract: Seismic data from seismic exploration surveys are mapped into a hypercube of bins or voxels in a four-dimensional space (X, Y, Offset, and Azimuth) according to Common Mid-Point (or CMP) between source and receivers. The mapped data from individual voxels or bins is then analyzed by multimodal statistics. Robust estimates of first break picks are obtained from the analysis. The first break picks are then used to as seed inputs for autopicking iteration, which proceeds to convergence. Estimates of confidence levels in the data are provided for re-picking to reduce computer processing time in successive autopicking iterations. Analysis is provided of different seismic attributes such as azimuthal velocity variations indicative of anisotropy, positioning errors of sources/receivers, geometry errors, and three dimensional distribution of inversion residuals. Analysis is also performed of standard deviation of the travel time data useful for estimating data errors in the inversion covariance matrix.

Abstract: Seismic data from seismic exploration surveys are mapped into a hypercube of bins or voxels in a four-dimensional space (X, Y, Offset, and Azimuth) according to Common Mid-Point (or CMP) between source and receivers. The mapped data from individual voxels or bins is then analyzed by multimodal statistics. Robust estimates of first break picks are obtained from the analysis. The first break picks are then used to as seed inputs for autopicking iteration, which proceeds to convergence. Estimates of confidence levels in the data are provided for re-picking to reduce computer processing time in successive autopicking iterations. Analysis is provided of different seismic attributes such as azimuthal velocity variations indicative of anisotropy, positioning errors of sources/receivers, geometry errors, and three dimensional distribution of inversion residuals. Analysis is also performed of standard deviation of the travel time data useful for estimating data errors in the inversion covariance matrix.