Exxonmobil Upstream Research Company

Exxonmobil Technology and Engineering Company

Anatoly I Baumstein

Partha Sarathi Routh

Yaxun Tang

Spyridon K Lazaratos

Sunwoong Lee

Tetyana Vdovina

John E Anderson

Jerome R Krebs

Ramesh Neelamani

Reeshidev Bansal

Di Yang

Ivan Chikichev

Ke Wang

Thomas A Dickens

Warren S Ross

Carey M Marcinkovich

Valeriy V Brytik

Junzhe Sun

Rongrong Lu

Mehdi Aharchaou

Gboyega Ayeni

Sirui Tan

Volkan Akcelik

David L Hinkley

Erik R Neumann

Jaewoo Park

Jia Yan

Peter Traynin

Mohamed T Hadidi

Gopalkrishna Palacharla

A methodology for extending bandwidth of geophysical data is disclosed. Geophysical data, obtained via a towed streamer, may have significant noise in a certain band (such as less than 4 Hz), rendering the data in the certain band unreliable. To remedy this, geophysical data, from a band that is reliable, may be extended to the certain band, resulting in bandwidth extension. One manner of bandwidth extension comprises using machine learning to generate a machine learning model. Specifically, because bandwidth may be viewed as a sequence, machine learning configured to identify sequences, such as recurrent neural networks, may be used to generate the machine learning model. In particular, machine learning may use a training dataset acquired via ocean bottom nodes in order to generate the machine learning model. After which, the machine learning model may be used to extend the bandwidth of a test dataset acquired via a towed streamer.

Bandwith extension of geophysical data

A method for enhancing properties of geophysical data with deep learning networks. Geophysical data may be acquired by positioning a source of sound waves at a chosen shot location, and measuring back-scattered energy generated by the source using receivers placed at selected locations. For example, seismic data may be collected using towed streamer acquisition in order to derive subsurface properties or to form images of the subsurface. However, towed streamer data may be deficient in one or more properties (e.g., at low frequencies). To compensate for the deficiencies, another survey (such as an Ocean Bottom Nodes (OBN) survey) may be sparsely acquired in order to train a neural network. The trained neural network may then be used to compensate for the towed streamer deficient properties, such as by using the trained neural network to extend the towed streamer data to the low frequencies.

Methodology for enhancing properties of geophysical data with deep learning networks

Iterative methods for inversion of seismic data to update a physical property model are disclosed. Such methods may comprise iteratively updating the model until a first predetermined resolution is achieved, using full wavefield inversion of the seismic data up to a first frequency threshold and assuming the seismic data is free of attenuation effects; extracting geobodies from the updated model; obtaining a Q model using the geobodies; and updating the physical property model using an inversion process, wherein the Q model is incorporated into the inversion process. These steps may be repeated until a second predetermined resolution of the physical property model is achieved, wherein the first frequency threshold is progressively increased in each repetition. The Q model may be updated with seismic data at all available frequencies to obtain a full-band Q model; and the physical property model may be updated using full-band migration and the full-band Q model.

Accurate velocity model estimation and imaging in the presence of localized attenuation (Q) anomalies

A computer-implemented method for updating subsurface models including: using an offset continuation approach to update the model, and at each stage defining a new objective function where a maximum offset for each stage is set, wherein the approach includes, performing a first stage iterative full wavefield inversion with near offset data, as the maximum offset, to obtain velocity and density or impedance models, performing subsequent stages of iterative full wavefield inversion, each generating updated models, relative to a previous stage, wherein the subsequent stages include incrementally expanding the maximum offset until ending at a full offset, wherein a last of the stages yields finally updated models, the subsequent stages use the updated models as starting models, and the full wavefield inversions include constraining scales of the velocity model updates at each stage of inversion as a function of velocity resolution; and using the finally updated models to prospect for hydrocarbons.

Full wavefield inversion with reflected seismic data starting from a poor velocity model

A computer-implemented method for updating a physical properties model of a subsurface region in an iterative inversion of seismic data using a gradient of a cost function that compares the seismic data to model-simulated data, said method comprising: obtaining a contrast model of a subsurface physical parameter that is sensitive to data dynamics and a kinematic model of a subsurface physical parameter; determining a gradient of a cost function using the contrast model and the kinematic model, wherein the cost function compares seismic data to model-simulated data; updating the kinematic model using a search direction derived from the gradient; adapting the contrast model according to an update to the kinematic model performed in the updating step; iteratively repeating the determining, updating, and adapting steps until a predetermined stopping criteria is reached, and generating a subsurface image from a finally updated kinematic model; and using the subsurface image to prospect for hydrocarbons.

Tomographically enhanced full wavefield inversion

A multi-stage FWI workflow uses multiple-contaminated FWI models to predict surface-related multiples. A method embodying the present technological advancement, can include: using data with free surface multiples as input into FWI; generating a subsurface model by performing FWI with the free-surface boundary condition imposed on top of the subsurface model; using inverted model from FWI to predict multiples; removing predicted multiples from the measured data; using the multiple-free data as input into FWI with absorbing boundary conditions imposed on top of the subsurface model; and preparing a multiple free data set for use in conventional seismic data processing.

Multistage full wavefield inversion process that generates a multiple free data set

A method, including: obtaining a seismic dataset that is separated into subsets according to predetermined subsurface reflection angle ranges; performing, with a computer, an acoustic full wavefield inversion process on each of the subsets, respectively, to invert for density and generate respective density models; generating acoustic impedances for each of the subsets, as a function of reflection angle, using the respective density models; and transforming, using a computer, the acoustic impedances for each of the subsets into reflectivity sections, wherein the transforming includes normalizing the reflectivity sections by their respective bandwidth.

FWI model domain angle stacks with amplitude preservation

An extended subspace method for inverting geophysical data to infer models for two or more subsurface physical properties, using gradients of an objective function as basis vectors for forming model updates. The extended set of basis vectors provides explicit mixing between gradient components corresponding to different medium parameters, for example P-wave velocity and an anisotropy parameter. In a preferred embodiment, off-diagonal elements of the mixing matrix may be scaled to adjust the degree of mixing between gradient components. Coefficients of the basis vector expansion are determined in a way that explicitly accounts for leakage or crosstalk between different physical parameters. The same extended subspace approach may be used to make further improvement to the model updates by incorporating well constraints, where well log data are available.

Extended subspace method for cross-talk mitigation in multi-parameter inversion

A method for iteratively inverting seismic data to jointly infer a model for at least P-wave velocity and attenuation parameters of the subsurface, the method including: jointly inverting the P-wave velocity and attenuation parameters with an iterative visco-acoustic full wavefield inversion process, wherein the iterative visco-acoustic full wavefield inversion process includes computing a gradient of an objective function, the objective function measuring a misfit between all or part of the seismic data and corresponding model-simulated seismic data; for each of the P-wave velocity and attenuation parameters, computing a search direction in model space from the gradient; determining line search step sizes &#x3b1; and &#x3b2; for the search directions for the P-wave velocity and attenuation parameters, respectively, wherein a ratio of the step sizes is a function of the P-wave velocity parameter; and using the step sizes &#x3b1; and &#x3b2; and the search directions for each of the P-wave velocity and attenuation parameters, computing a new search direction in model space, then performing a line search along the new search direction to arrive at a new step size, and using the new step size and the new search direction to generate an updated model for a current iteration of the iterative visco-acoustic full wavefield inversion process.

Joint full wavefield inversion of P-wave velocity and attenuation using an efficient first order optimization

A method, including performing, with a computer, up/down separation of geophysical data, which produces an approximate up-going wavefield and an approximate down-going wavefield; creating an areal source based at least in part on the down-going wavefield; and performing, with a computer, a full wavefield inversion process with the areal source, and an objective function measuring a misfit between modeled up-going wavefields and recorded up-going wavefields, wherein the full wavefield inversion process generates a final subsurface physical property model.

FWI with areal and point sources

Method for correcting seismic simulations, RTM, and FWI for temporal dispersion due to temporal finite difference methods in which time derivatives are approximated to a specified order of approximation. Computer-simulated seismic data () are transformed from time domain to frequency domain (), and then resampled using a mapping relationship that maps, in the frequency domain, to a frequency at which the time derivative exhibits no temporal dispersion (), or to a frequency at which the time derivative exhibits a specified different order of temporal dispersion. Alternatively, measured seismic data from a field survey () may have temporal dispersion of a given order introduced, by a similar technique, to match the order of approximation used to generate simulated data which are to be compared to the measured data.

Method for temporal dispersion correction for seismic simulation, RTM and FWI

Method for using the full wavefield (primaries, internal multiples and free-surface multiples) in inversion of marine seismic data, including both pressure and vertical velocity data (), to infer a subsurface model of acoustic velocity or other physical property. The marine seismic data are separated () into up-going () and down-going () wavefields, and both wavefields are inverted in a joint manner, in which the final model is impacted by both wavefields. This may be achieved by inverting both wavefields simultaneously (), or one after the other, i.e. in a cascaded approach (, or), for the subsurface properties ().

Full-wavefield inversion of primaries and multiples in marine environment

The invention includes a method for reducing noise in migration of seismic data, particularly advantageous for imaging by simultaneous encoded source reverse-time migration (SS-RTM). One example embodiment includes the steps of obtaining a plurality of initial subsurface images; decomposing each of the initial subsurface images into components; identifying a set of components comprising one of (i) components having at least one substantially similar characteristic across the plurality of initial subsurface images, and (ii) components having substantially dissimilar characteristics across the plurality of initial subsurface images; and generating an enhanced subsurface image using the identified set of components. For SS-RTM, each of the initial subsurface images is generated by migrating several sources simultaneously using a unique random set of encoding functions. Another embodiment of the invention uses SS-RTM for velocity model building.

Seismic data processing

Method for estimating the Hessian of the objective function, times a vector, in order to compute an update in an iterative optimization solution to a partial differential equation such as the wave equation, used for example in full wave field inversion of seismic data. The Hessian times vector operation is approximated as one forward wave propagation () and one gradient computation () in a modified subsurface model (). The modified subsurface model may be a linear combination of the current subsurface model () and the vector () to be multiplied by the Hessian matrix. The forward-modeled data from the modified model are treated as a field measurement in the data residual of the objective function for the gradient computation in the modified model. In model parameter estimation by iterative inversion of geophysical data, the vector in the first iteration may be the gradient of the objective function.

Methods for approximating hessian times vector operation in full wavefield inversion

Method for stabilizing the updated model () in iterative seismic data inversion so that the model-simulated data for the next iteration does not 'blow up.' A Projection Onto Convex Sets ('POCS') operator is defined that converts the matrix corresponding to the model to a positive semi-definite matrix. The stability projection operator may be looped with physical constraint projection operators until the model converges (). The resulting stable and constrained model is then used to simulate seismic data in the next cycle of the outer iteration loop ().

Growing community of inventors

Houston, TX, United States of America

Anatoly I Baumstein