Hydrocarbon exploration and extraction can be facilitated using machine-learning models. For example, a system described herein can receive seismic data indicating locations of geological bodies in a target area of a subterranean formation. The system can provide the seismic data as input to a trained machine-learning model for determining whether the target area of the subterranean formation includes one or more types of geological bodies. The system can receive an output from the trained machine-learning model indicating whether or not the target area of the subterranean formation includes the one or more types of geological bodies. The system can then execute one or more processing operations for facilitating hydrocarbon exploration or extraction based on the seismic data and the output from the trained machine-learning model.

Limitations in accuracy and computing power requirements impeding conventional Kirchhoff migration and reverse time migration are overcome by using the wave-equation Kirchhoff, WEK, technique with Kirchhoff migration. WEK technique includes forward-propagating a low-frequency wavefield from a shot location among pre-defined source locations, calculating an arrival traveltime of a maximum amplitude of the low-frequency wavefield, and applying Kirchhoff migration using the arrival traveltime and the maximum amplitude.

Methods and apparatus are described for reducing noise in measurements made by one or more pressure sensors disposed in a cable having a generally longitudinal axis. Estimated axial vibration noise at a location along the cable is determined based at least in part on measurements from one or more motion sensors disposed along the cable. The estimated axial vibration noise is subtracted from pressure sensor measurements corresponding to the location. The result is noise-attenuated pressure sensor measurements corresponding to the location.

The present disclosure describes methods and systems, including computer-implemented methods, computer program products, and computer systems, for smoothing seismic data. One computer-implemented method includes obtaining, by a hardware data processing apparatus, a plurality of seismic data samples; forming, by the hardware data processing apparatus, guiding vectors using the plurality of seismic data samples and a plurality of guiding structure attributes; generating, by the hardware data processing apparatus, a structure guided directional weighted vector filter using the guiding vectors and a plurality of weighting factors; filtering, by the hardware data processing apparatus, the seismic data samples using the structure guided directional weighted vector filter to generate smoothed seismic data; and initiating output of the smoothed seismic data.

The invention is a method for processing seismic data to eliminate the coherent noise arising from multiple seismic reflections. A decomposition procedure is applied to decompose along N directions of decomposition seismic data into a set of N components. At least one model of multiple reflections is decomposed according to the same decomposition procedure and along the same N directions. For each direction of decomposition, a relative concentration between the component of the model of multiples and the component of the data in the direction concerned is calculated. Next a procedure of adaptive filtering of the multiple reflections is applied to each of the seismic components. The filtered seismic components are recombined with the recombination being weighted by a weighting dependent on the relative concentrations calculated for each direction of decomposition.

A method for deghosting marine seismic streamer data includes receiving the marine seismic data recorded from a single streamer. A representation of an upgoing wavefield in the marine seismic data is defined. The representation of the upgoing wavefield includes a first wave component in a plane and a second wave component that is perpendicular to the plane. A linear system is built that models a wavefield using the representation. The wavefield includes the upgoing wavefield and a downgoing ghost wavefield. The upgoing wavefield is estimated, within the representation, by inverting the linear system.

The present disclosure describes computer-implemented methods, computer-program products, and computer systems, for providing parameters for successful automated fault patch extraction. Attributes are selected for annotating images generated from, and for interpretation of, seismic amplitude volume. Images are generated from layers of a seismic cube, each generated using a different attribute of the plural attributes. The plural images are blended using customized palettes and initial parameters to create a blended image illuminating discontinuities in the layers. Optimal parameters are iteratively determined for automatic derivation of fault discontinuities on an interpreter-selected edge-enhanced sub-volume. The iterations are controlled and terminated based on interpreter inputs. The optimal extraction parameters are applied to an entire edge-enhanced volume. Important extracted fault discontinuities are isolated using commercial filtering tools. Extracted fault patches are refined based on received manual interpretation. Patch results are converted to traditional fault objects for further interpretation and refinement.

Computing device, computer instructions and method for correcting an image, of a surveyed surface, due to a free-surface reflection. The method includes calculating a free surface reflection operator for a seismic source displaced in water based on a position of the source, and an air-water interface datum; receiving recorded seismic data d recorded with seismic sensors (r), wherein the recorded seismic data is associated with a pressure and/or a particle motion produced by a seismic wave in earth; correcting the recorded seismic data d based on the free surface reflection operator to obtain transformed seismic data; and generating an image of the surveyed subsurface, based on the transformed seismic data, wherein the image is indicative of various layers of the earth. The free surface reflection operator varies while a source signal is being emitted by the source.

Seismic image filtering machines, systems, program products, and computer implemented methods are provided to generate a filtered seismic image responsive to filtered seismic image data generated by attenuating coherent seismic noise from surface waves of an unfiltered wavefield constructed from unfiltered seismic image data through a single downward extrapolation of the unfiltered wavefield using a plurality of nonstationary convolution operators to perform localized filtering at each of a plurality of spatial locations of the unfiltered wavefield. Various embodiments, for example, can beneficially handle strong lateral velocity variations thus making various embodiments effective tools to remove complicated coherent seismic noise which is typically in the form of exponentially decaying evanescent waves. Embodiments of the present invention, for example, can use, as a part of the filtering mechanism, specially designed nonstationary convolution operators that are implemented in the space-frequency domain as nonstationary filters.

A method is described that includes obtaining a multicomponent seismic data set for a subterranean region of interest and determining, using a computer processor, a PP stacked time-domain seismic image and a PS stacked time-domain seismic image from the multicomponent seismic data set. The method further includes transforming a recording-time axis of at least one of the PP stacked time-domain seismic image and the PS stacked time-domain seismic image to produce a pair of coarsely-registered PP and PS seismic images and filtering at least one of the pair to produce a pair of spectrally-matched PP and PS seismic images. Further, the method includes dynamically warping at least one of the pair of spectrally-matched PP and PS seismic images to produce a pair of fully-registered PP and PS seismic images.