This disclosure describes a system and method for generating a subsurface model representing lithological characteristics and attributes of the subsurface of a celestial body or planet. By automatically ingesting data from many sources, a machine learning system can infer information about the characteristics of regions of the subsurface and build a model representing the subsurface rock properties. In some cases, this can provide information about a region using inferred data, where no direct measurements have been taken. Remote sensing data, such as aerial or satellite imagery, gravimetric data, magnetic field data, electromagnetic data, and other information can be readily collected or is already available at scale. Lithological attributes and characteristics present in available geoscience data can be correlated with related remote sensing data using a machine learning model, which can then infer lithological attributes and characteristics for regions where remote sensing data is available, but geoscience data is not.

A method, system and non-transitory computer-readable medium for forming a seismic image of a geological structure are provided. After obtaining seismic wave data including a plurality of seismic wave traces at a first region of the geological structure, a predicted time dispersion error of an actual time dispersion error that results from a use of a finite difference approximation in calculating predicted seismic wave data at a second region of the geological structure as if a seismic wave propagates from the first region to the second region of the geological structure, is calculated. A corrected predicted seismic wave data at the second region of the geological structure is calculated by applying the finite difference approximation to the seismic wave data at the first region of the geological structure compensated with the predicted time dispersion error. A seismic image of the second region of the geological structure is generated using the corrected predicted seismic wave data, such that the actual time dispersion error is negated by the predicted time dispersion error.

A method, and a system for implementing the method, are disclosed wherein coordinates of survey region are used to locate small pieces of a seismic wave model, usually defined by their nodes (or vertices) and contain information about physical properties, such as liquid or solid, density, velocity that seismic waves propagates in it; and connects them to the appropriate source and receiver sensor. In particular, the method and system disclosed, generates a multi-layer mapping of the survey region by decomposing the survey region into cubes containing small pieces of seismic wave models (the elements), as well as source and receiver location. Those cubes are then indexed depending upon their location and the elements, sources and receivers are assigned to a particular cube thereby creating a multi-layer relationship between the survey region map, the cube map, the elements map, as well as the source and receiver locations.

A method can include acquiring imagery of an exposed surface of the Earth; generating a multi-dimensional model based at least in part on the imagery; generating synthetic seismic data utilizing the multi-dimensional model; acquiring seismic data of a subsurface region of the Earth; performing a search that matches a portion of the acquired seismic data and a portion of the synthetic seismic data; and characterizing the subsurface region of the Earth based at least in part on the portion of the synthetic seismic data.

This disclosure describes a system and method for generating a subsurface model representing lithological characteristics and attributes of the subsurface of a celestial body or planet. By automatically ingesting data from many sources, a machine learning system can infer information about the characteristics of regions of the subsurface and build a model representing the subsurface rock properties. In some cases, this can provide information about a region using inferred data, where no direct measurements have been taken. Remote sensing data, such as aerial or satellite imagery, gravimetric data, magnetic field data, electromagnetic data, and other information can be readily collected or is already available at scale. Lithological attributes and characteristics present in available geoscience data can be correlated with related remote sensing data using a machine learning model, which can then infer lithological attributes and characteristics for regions where remote sensing data is available, but geoscience data is not.

A method including selecting a forward model based on a modeled well structure and including a single modeled acoustic source located in a modeled wellbore and a plurality of modeled acoustic sensors located in a modeled source area, simulating an acoustic signal generated by the single modeled acoustic source and received by each modeled acoustic sensor, calculating phases of the simulated acoustic signals received at each modeled acoustic sensor, obtaining with a principle of reciprocity a plurality of modeled acoustic sources in the modeled source area and a single modeled acoustic sensor in the modeled wellbore, calculating phase delays of the simulated acoustic signals between each modeled acoustic source and the single modeled acoustic sensor, detecting acoustic signals generated by a flow of fluid using acoustic sensors in a wellbore, and processing the acoustic signals using the phase delays to generate a flow likelihood map.

A method of seismic modeling using an elastic model, the elastic model including a grid having a grid spacing sized such that, when synthetic seismic data is generated using the elastic model, synthetic shear wave data exhibits numerical dispersion, the method including: generating generated synthetic seismic data using the elastic model, wherein the generated synthetic seismic data includes synthetic compression wave data and synthetic shear wave data, and modifying the generated synthetic seismic data to produce modified synthetic seismic data by attenuating at least some of the synthetic shear wave data in order to attenuate at least some of the numerically dispersive data.

Systems and methods of identifying a drilling target are disclosed. The method includes obtaining a training set of base subsurface models and generating, using a first artificial intelligence neural network, a plurality of subsurface model realizations based on the base subsurface models. The method further includes simulating, for each subsurface model realization, a synthetic seismic dataset and training a second artificial intelligence neural network, using the plurality of subsurface model realizations and the corresponding synthetic seismic dataset for each subsurface model realization, to predict an inferred subsurface model from a seismic dataset. The method still further includes obtaining an observed seismic dataset for a subterranean region of interest, predicting, using the trained second artificial intelligence neural network, an inferred subsurface model from the observed seismic dataset, and identifying the drilling target based on the inferred subsurface model.

The present invention discloses a model parameter design method and device for simulating propagation of seismic waves at any discontinuous interface, including: reading source wavelet and model parameters; selecting a space step and a time step according to the model parameters, and setting a space order of the finite difference numerical simulation; discretizing the discontinuous interface in the medium, and according to the threshold judgment criteria, obtaining many staircase discretization results for the discontinuous interface; conducting a forward modeling on the obtained multiple results of staircase discretization for the discontinuous interface by using the finite difference method, and setting parameters of the forward modeling; and obtaining the final simulation results by the superposition of the forward modeling results of different staircase-discretization of the discontinuous interface. The present invention can greatly improve the calculation efficiency.

A method for partial differential equation inversion, the method including receiving measured data do; selecting an objective function having first and second measures N.sub.1 and N.sub.2, wherein the objective function depends on three independent variables V, u, and f, V being a perturbation of a wave equation operator L from a background operator L.sub.0, u being a wavefield that satisfies the wave equation operator L, and f being a source function that describes the source of the waves; optimizing with a processor the objective function by finding a minimum or a maximum using the inversion; calculating with the processor solutions V*, u*, and f* of the three independent variables V, u, and f; and generating with the processor an image of an object based on the solutions V*, u*, and f*.