A method and apparatus for seismic analysis include obtaining an initial geophysical model and seismic data for a subsurface region; producing a subsurface image of the subsurface region with the seismic data and the geophysical model; generating a map of one or more geologic features of the subsurface region by automatically interpreting the subsurface image; and iteratively updating the geophysical model, subsurface image, and map of geologic features by: building an updated geophysical model based on the geophysical model of a prior iteration constrained by one or more geologic features from the prior iteration; imaging the seismic data with the updated geophysical model to produce an updated subsurface image; and automatically interpreting the updated subsurface image to generate an updated map of geologic features. The method and apparatus may also include post-stack migration, pre-stack time migration, pre-stack depth migration, reverse-time migration, gradient-based tomography, and/or gradient-based inversion methods.

Methods and systems for determining an interbed multiple attenuated pre-stack seismic dataset are disclosed. The methods include forming a post-stack seismic image composed of post-stack traces from the pre-stack seismic dataset and identifying a first, second, and third post-stack horizon on each of the post-stack traces. The methods further include for each pre-stack trace, generating a first, second, and third multiple-generator trace based on the first, second and third post-stack horizon and determining a correlation trace based, at least in part, on a correlation between the first multiple-generator trace and the second multiple-generator trace. The methods still further include predicting an interbed multiple trace by convolving the correlation trace and the third multiple-generator trace, determining an interbed multiple attenuated trace by subtracting the interbed multiple trace from a corresponding pre-stack seismic trace, and determining the interbed multiple attenuated pre-stack seismic dataset by combining the interbed multiple attenuated traces.

Optimizing seismic to depth conversion to enhance subsurface operations including measuring seismic data in a subsurface formation, dividing the subsurface formation into a training area and a study area, dividing the seismic data into training seismic data and study seismic data, wherein the training seismic data corresponds to the training area, and wherein the study seismic data corresponds to the study area, calculating target depth data corresponding to the training area, training a machine learning model using training inputs and training targets, wherein the training inputs comprise the training seismic data, and wherein the training targets comprise the target depth data, computing, by the machine learning model, output depth data corresponding to the study area based at least in part on the study seismic data; and modifying one or more subsurface operations corresponding to the study area based at least in part on the output depth data.

A system and method can be used for to calibrating time-lapse seismic volumes by cross-migration rescaling and reorientation for use in determining optimal wellbore placement or production in a subsurface environment. Certain aspects include methods for cross-migration of data sets processed using different migration techniques. Pre-processing of the data sets, optimization of rescaling and reorientation, and identification of adjustment parameters associated with minimum global error can be used to achieve a time-dependent formation data set that addresses error in all input data sets.

A method includes receiving seismic data for a geologic region of the Earth; building a velocity model of the geologic region of the Earth; selecting at least one mode of multiple and corresponding travel time data from a data storage where the travel time data correspond to at least one complex ray signature in the geologic region of the Earth and are based at least in part on the velocity model; performing migration on the seismic data using at least the selected travel time data to generate processed seismic data; and rendering an image of the geologic region of the Earth to a display where the image includes at least a multiple image.

A computer system receives a post-stack time-domain image having a first spectrum and representing one or more subsurface structures. The computer system reconstructs an increased-frequency version of the post-stack time-domain image using L0-constrained inversion and a least-squares mismatch ratio. The increased-frequency version of the post-stack time-domain image includes structural artifacts. The computer system removes the structural artifacts from the increased-frequency version of the post-stack time-domain image using singular value decomposition. The computer system combines the increased-frequency version of the post-stack time-domain image with the post-stack time-domain image using a weighting function. The computer system generates a combined version of the increased-frequency version of the post-stack time-domain image and the post-stack time-domain image. The combined version represents the one or more subsurface structures and has a second spectrum broader than the first spectrum.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining a velocity model for a geological region. In one aspect, a method comprises: obtaining a current velocity model for the geological region; obtaining pre-stack and post-stack seismic data characterizing the geological region; and for each of a plurality of iterations: identifying a plurality of reflection events from the post-stack seismic data and the current velocity model; determining a respective observed travel time for each of the plurality of reflection events, comprising, for each reflection event, determining the respective observed travel time for the reflection event based at least in part on kinematic features derived from a respective seismic trace included in the pre-stack seismic data; and updating the current velocity model based at least in part on the observed travel times of the plurality of reflection events.

The present disclosure describes methods and systems, including computer-implemented methods, computer program products, and computer systems, for suppressing noises in seismic data. One computer-implemented method includes receiving, at a data processing apparatus, a set of seismic data associated with a subsurface region; flattening, by the data processing apparatus, the set of seismic data according to an identified seismic event; dividing, by the data processing apparatus, the set of seismic data into a plurality of spatial windows; randomizing, by the data processing apparatus, the set of seismic data according to a random sequential order; filtering, by the data processing apparatus, the randomized seismic data; and reorganizing, by the data processing apparatus, the filtered seismic data according to a pre-randomization order.

A method involving automated salt body boundary interpretation employs multiple sequential supervised machine learning models which have been trained using training data. The training data may consist of pairs of seismic data and labels as determined by human interpretation. The machine learning models are deep learning models, and each of the deep learning models is aimed to address a specific challenge in the salt body boundary detection. The proposed approach consists of application of an ensemble of deep learning models applied sequentially, wherein each model is trained to address a specific challenge. In one example an initial salt boundary inference as generated by a first trained first deep learning model is subject to a trained refinement deep learning model for false positives removal.

A zero-offset wavefield synthesis workflow to calculate a synthesized zero-offset wavefield output without the commitment to an rms velocity field output to circumvent velocity uncertainty. Said zero-offset wavefield synthesis workflow comprises calculating a migration cube output. Rendering a demigration cube output from said migration cube output with a demigration cube calculation. Rendering said synthesized zero-offset wavefield output from said demigration cube output with a zero-offset wavefield synthesis procedure.