A method for training a backpropagation-enabled segmentation process is used for identifying an occurrence of a sub-surface feature. A multi-dimensional seismic data set with an input dimension of at least two is inputted into a backpropagation-enabled process. A prediction of the occurrence of the subsurface feature has a prediction dimension of at least 1 and is at least 1 dimension less than the input dimension.

A method for training a backpropagation-enabled regression process is used for predicting values of an attribute of subsurface data. A multi-dimensional seismic data set with an input dimension of at least two is inputted into a backpropagation-enabled process. A predicted value of the attribute has a prediction dimension of at least 1 and is at least 1 dimension less than the input dimension.

A method of generating diffraction images based on wave equations includes generating a source wavefield and a receiver wavefield. Based on the source wavefield, a first source wavefield propagating in a first direction and a second source wavefield propagating in a second direction are generated. Based on the receiver wavefield, a first receiver wavefield propagating in the first direction and a second receiver wavefield propagating in the second direction are generated. A first seismic image is generated based on the first source wavefield and the first receiver wavefield. A second seismic image is generated based on the second source wavefield and the second receiver wavefield. A final seismic image is generated based on the first seismic image and the second seismic image.

A system includes a processor and a memory. The memory includes instructions that are executable by the processor to access a plurality of seismic images of a subterranean formation in a first geological area. The instructions are also executable to generate a plurality of fault estimates from each of the plurality of seismic images. Further, the instructions are executable to generate a processed seismic image of the first geological area by normalizing and merging the plurality of seismic images and the plurality of fault estimates. Additionally, the instructions are executable to generate a statistical fault uncertainty volume of the first geological area using the processed seismic image. Furthermore, the instructions are executable to control a drilling operation in the first geological area using the statistical fault uncertainty volume of the first geological area.

A method includes receiving a training selection of a first set of faults located in a first subset of a seismic dataset for a subsurface geologic formation, detecting a second set of faults in the seismic dataset based on fault interpretation operations using a first set of interpretation parameters, and determining a difference between the first set of faults and the second set of faults. The method also includes generating a second set of interpretation parameters for the fault interpretation operations based on the difference between the first set of faults and the second set of faults, and determining a feature of the subsurface geologic formation based on fault interpretation operations using the second set of interpretation parameters.

Methods for dolomite mapping using multiscale fracture characterization include using a computer system to receive seismic data for a geographical area. The computer system identifies one or more macroscale fractures located within the geographical area based on a three-dimensional (3D) visualization of the seismic data. The computer system identifies one or more mesoscale fractures located within the geographical area based on a curvature map generated from the seismic data. The computer system identifies one or more microscale fractures located within the geographical area based on an amount of chaotic seismic reflections indicated by the seismic data. The computer system identifies a dolomite distribution of the geographical area based on the one or more macroscale fractures, the one or more mesoscale fractures, and the one or more microscale fractures. A display device of the computer system generates a graphical representation of the dolomite distribution.

The method for producing a geological vector model (GVM) based on seismic data includes the step of forming a Model-Grid, which includes creating a network of small units, called patches, to which a relative geological age is assigned, a set of patches with the same relative geological age corresponding to a geological layer, called the geological horizon. The method includes the step of sampling the Model-Grid in two directions perpendicular to each other, enabling the Model-Grid to be sampled in a plurality of vertical planes and the step of forming two-dimensional geological vector models (2DGVM). The step of forming includes forming two-dimensional horizons (Hb) with distinct relative geological ages using the patches belonging to each sampled plane, each two-dimensional geological vector model (2DGVM) corresponding to a vertical plane originating from the sampling of the Model-Grid.

A method includes receiving information that defines a three-dimensional subterranean structure; splitting the three-dimensional subterranean structure into portions; generating convex hulls for the portions; and generating a discrete fracture network based at least in part on the convex hulls.

A fault indicator calculator, a method for determining a fault indicator, and a fault indicator calculating system are disclosed herein. One embodiment of a fault indicator calculator includes: 1) an interface configured to receive seismic data, and 2) a processor configured to scan a manifold-shaped operator through said seismic data at a range of dips and azimuths and calculate fault throws at various orientations of said dips and azimuths independent of determining other fault indicators.

A method to automatically interpret a subsurface feature within geophysical data, the method including: storing, in a computer memory, geophysical data obtained from a survey of a subsurface region; and extracting, with a computer, a feature probability volume by processing the geophysical data with one or more fully convolutional neural networks, which are trained to relate the geophysical data to at least one subsurface feature, wherein the extracting includes fusing together outputs of the one or more fully convolutional neural networks.