Automation of tomographic inversion picking with ml methods

Abstract

Disclosed are methods, systems, and computer programs that computationally select event data from captured seismic data. The methods comprise generating a learning model associated with a machine learning (ML) engine. The methods also comprise receiving seismic data that comprise a propagated wavefield transmitted within the subsurface of the resource site. The methods also comprise generating a data matrix using the received seismic data and resolving event data comprised in the received seismic data to generate a segmented wavefield. A geological model of the subsurface of the resource site may be generated using the segmented wavefield such that the geological model of the subsurface of the resource site indicates at least a multidimensional image of the subsurface of the resource site.

Description

AUTOMATION OF TOMOGRAPHIC INVERSION PICKING WITH ML METHODSINTRODUCTION

[0001] This disclosure is directed to analyzing seismic data using tomographic inversion operations to generate subsurface image data.BACKGROUND

[0002] The selection or picking of event data from a received seismic wavefield can be used in tomographic inversion operations to generate analysis data and / or analysis models, and / or geological models of a subsurface. The nature of the quantitative and / or qualitative properties of the analysis data and / or analysis models, and / or geological models associated with tomographic inversion is dependent on the process used to pick or select the event data comprised in a captured seismic wavefield.

[0003] While some deterministic picking methods (e.g., CIPPick, which is not based on ML data processing but on deterministic signal processing techniques such as cross-correlation) have been used for selecting or picking event data from seismic data as part of tomographic inversion operations, these techniques often operate independently on separate 2-dimensional (x-y) data / gathers by picking the same type of event data within a neighborhood of each other to generate deterministic event data. Moreover, some deep-learning pickers of event data within seismic data also have a limited “field of view” as such pickers are restricted to x-y dimensions such that only the neighborly situated event data are picked at a time. In particular, the pick variation indicating a variation of picked wavefield amplitude and phase relative to those of neighboring picks / event data for such scenarios, are not associated with subsurface velocity variation data of the propagated wavefield which could lead to inconsistencies in velocity updates associated with captured event data using such techniques.

[0004] As such, there is a need to develop efficient and accurate workflows and processes that select optimal events or event data from captured seismic data in order to optimize tomographic inversion processes as well as a method with inherent spatial consistency processes that are independent of any picking tool or methodology.SUMMARY

[0005] Disclosed are methods, systems, and computer programs for computationally selecting event data comprised in seismic data captured by a set of sensors deployed at a resource site.

[0006] According to an embodiment, a method for selecting event data from seismic data comprises: generating a learning model associated with a machine learning (ML) engine, wherein the learning model is parameterized using: first spatial data associated with one or more geo-targets in a subsurface of the resource site, and selected offset data associated with a set of sensors deployed at the resource site; receiving the seismic data from the set of sensors, the seismic data comprising a propagated wavefield that is transmitted within the subsurface of a resource site by a first sensor of the set of sensors, and received by a second sensor of the set of sensors; formatting a data matrix using the received seismic data, the data matrix comprising one or more multidimensional parameters that represent: second spatial data indicating the one or more geo-targets with which the propagated wavefield interacts, and determined offset data indicating displacement data of the first sensor relative to the second sensor; resolving event data comprised in the seismic data to generate a segmented wavefield by applying the learning model to the data matrix, the segmented wavefield indicating a computational selection of the event data comprised in the received propagated wavefield; and generating a geological model of the subsurface of the resource site using the segmented wavefield, the geological model of the subsurface of the resource site indicating at least a multidimensional image of the subsurface of the resource site.

[0007] In other embodiments, a system and a computer program can include or execute the method described above. These and other implementations may each optionally include one or more of the following features.

[0008] In one embodiment, resolving the event data comprises: identifying, using the learning model, one or more spatially consistent event data associated with, or comprised in the data matrix; selecting one or more geological boundaries associated with illuminated regions indicating the spatially consistent event data of the one or more geo-targets in the subsurface of the resource site, such that the illuminated regions are based on interactions of the propagated wavefield with the one or more geo-targets in the subsurface of the resource site; and generating, using the one or more geological boundaries, the segmented wavefield.

[0009] In one embodiment, the set of sensors comprise one or more of: hydrophonic sensors, geophonic sensors, broadband sensors, or a distributed acoustic (DAS) sensors. In some embodiments, the resource site comprises one or more of: an onshore resource site or an offshore resource site.

[0010] In one embodiment, the learning model may be parameterized using one or more of: training data including historical subsurface data associated with the first spatial data or the selected offset data; or simulation data indicating an initial boundary condition for the learning model based on the first spatial data or the selected offset data.

[0011] In one embodiment, the one or more multidimensional parameters of the data matrix include a parameter set associated with the second spatial data which in aggregate define one or more of area data or volume data, such that: the one or more geo-targets are comprised in, or are proximal to a location associated with the area data or volume data, and the propagated wavefield travels through the location within the subsurface of the resource site.

[0012] In one embodiment, the parameter set comprises geo-parameters indicating geolocation points associated with the one or more geo-targets in the subsurface of the resource site.

[0013] In one embodiment, the one or more multidimensional parameters of the data matrix include a displacement parameter associated with the determined offset data. The displacement parameter may indicate at least one of: an angular displacement of the propagated wavefield from the first sensor to the second sensor, an azimuthal displacement of the first sensor relative to the second sensor, or a distance between the first sensor and the second sensor.

[0014] In one embodiment, the propagated wavefield logs one or more of: geological boundary data associated with the one or more geo-targets; rock property data associated with the one or more geo-targets; fluid flow condition data associated with the one or more geotargets; air gap data associated with the one or more geo-targets; subsurface discontinuity data associated with the one or more geo-targets; subsurface layering data associated with the one or more geo-targets; hydrocarbon data associated with the one or more geo-targets; and mineral deposit data associated with the one or more geo-targets.

[0015] In one embodiment, the learning model comprises: a tomographic learning model with a trainable data layout structure; and the trainable data layout structure comprisesone of: a neural network, an intelligent data structure, an artificial intelligence engine, or a structure of an interconnected data learning pathway.

[0016] In one embodiment, formatting the data matrix comprises sorting or arranging at least event data comprised in the seismic data into one or more groups of data points associated with the one or more geo-targets.

[0017] In one embodiment, the learning model is a 4-dimensional model configured for selecting the event data comprised in the seismic data.

[0018] In one embodiment, a dimension of the learning model is based on the data matrix. The dimension of the learning model comprises one of a 4-dimensional model or a 5- dimensional model.

[0019] In one embodiment, the multidimensional image of the subsurface of the resource site comprises one of a 2-dimensional image or a 3-dimensional image.

[0020] In one embodiment, the multidimensional image of the subsurface of the resource site may be used to initiate one or more of: adjusting a drill bit spin rate at the resource site based on geological data comprised in the multidimensional image; regulating one or more flow control devices at the resource site based on fluid data comprised in the multidimensional image; optimizing gas storage operations in the subsurface of the resource site based on the multidimensional image; and optimizing energy development operations at the resource site based on the multidimensional image.

[0021] In one embodiment, the one or more multidimensional parameters of the data matrix include a parameter set associated with the second spatial data which in aggregate define one or more of area data or volume data; and the one or more multidimensional parameters of the data matrix include a displacement parameter associated with the determined offset data.

[0022] In one embodiment, a dimension of the learning model is based on at least three parameters comprised in the one or more multidimensional parameters of the data matrix.

[0023] In one embodiment, the seismic data comprises one of: subsurface volume data corresponding to the selected offset data that characterize one or more lumps of classified pixels within the multidimensional image with each lump of the one or more classified pixels corresponding to a geological segment that has an assigned code based on the learning model; or subsurface area data corresponding to the selected offset data that characterize the one or more lumps of classified pixels within the multidimensional image with each lump of the oneor more classified pixels corresponding to the geological segment that has an assigned code based on the learning model.BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The disclosure is illustrated by way of example, and not by way of limitation in the figures of the accompanying drawings in which like reference numerals are used to refer to similar elements. It is emphasized that various features may not be drawn to scale and the dimensions of various features may be arbitrarily increased or reduced for clarity of discussion.

[0025] FIG. 1A shows one or more stack volumes associated with a resource site used to generate a model of a subsurface of the resource site.

[0026] FIG. IB shows an exemplary propagation of sensor signals or wavefields into a subsurface of the resource site.

[0027] FIG. 2 shows a cross-sectional view of a resource site for which the process of FIG. 4 may be executed.

[0028] FIG. 3 shows a networked system illustrating a communicative coupling of devices or systems associated with the resource site of FIG. 2.

[0029] FIG. 4 provides an exemplary workflow for methods, systems, and computer programs that select event data from seismic data.

[0030] FIGS. 5A-5D provide exemplary segment data associated with geological models generated using the disclosed techniques.DETAILED DESCRIPTION

[0031] Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings and figures. In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the disclosed subject-matter. However, it will be apparent to one of ordinary skill in the art that the solutions disclosed may be practiced without these specific details. In other instances, well- known methods, procedures, components, circuits and networks have not been described in detail so as not to unnecessarily obscure aspects of the embodiments.

[0032] The disclosed systems and methods may be accomplished using interconnected devices and systems that obtain a plurality of data associated with various parameters of interest at a resource site. The workflows / flowcharts described in this disclosure, according to some embodiments, implicate a new processing approach (e.g., hardware, special purpose processors, and specially programmed general-purpose processors) because such analyses are too complex and cannot be done by a person in the time available or at all. Thus, the described systems and methods are directed to tangible implementations or solutions to specific technological problems in developing natural resources such as oil, gas, water well industries, and other mineral exploration operations. More specifically, the systems and methods presently disclosed may be applicable to operations associated with computationally selecting event data comprised in seismic data captured by a set of sensors deployed at a resource site.

[0033] Attention is now directed to methods, techniques, infrastructure, and workflows for operations that may be carried out at a resource site. Some operations in the processing procedures, methods, techniques, and workflows disclosed herein may be combined while the order of some operations may be changed. Some embodiments include an iterative refinement of one or more data associated with the resource site via feedback loops executed by one or more computing device processors and / or through other control devices / mechanisms that make determinations regarding whether a given action, template, or resource data, etc., is sufficiently accurate.

[0034] Disclosed are methods and systems for selecting event data comprised in captured seismic data for optimal tomographic inversion operations. According to one embodiment, the workflows disclosed are based on automated machine learning (ML) operations for selecting seismic events. In particular, the disclosed workflows produce spatially consistent event data (also referred to as picks) from captured seismic data relative to one or more of a surface offset, a subsurface angle associated with a propagated wavefield in the subsurface, and other relevant dimensions. The disclosed solutions may be formulated from a semantic segmentation perspective with the picks being predicted as the boundaries of segments associated with the captured seismic data or the picks being predicted as a segment themselves associated with the captured seismic data. In other embodiments, the disclosed solutions are formulated from an object detection perspective that allows for pick prediction along an offset dimension or an azimuth dimension, for example, using a backbone platform comprising oneor more of a learning model (e.g., convolutional neural network such as ResNet) associated with certain architectures (e.g., U-Net, DeepLab, flow field) and which has trainable data structures or networks (e.g., generative networks, recurrent networks, etc.). Such a learning model, for example, may be used together with a transformer to implement a workflow that selects optimal picks. According to one embodiment, a plurality of similar or dissimilar training datasets and / or model training strategies can be used, as part of initializing ML operations associated with the learning model, to account for data variations (e.g., wavelet variations) of dimensions. According to one embodiment, selected picks can be used independently from, or in combination with deterministic pick selection methodologies in order to further enhance the picks and / or facilitate quality control operations on picks based on workflows with reduced numbers of parameters.

[0035] According to one embodiment of this disclosure, a machine learning (ML) tool(e.g., a learning model) is used to automatically pick or select optimal event data (e.g., moveout event data) from captured seismic data. The event data, for example, may indicate a representative wavefield at different sensor locations as the wavefield travels through and illuminates target regions of the subsurface. The moveout event data may indicate a travel time difference of the wavefield between sensors based on subsurface variations associated with the traveling wavefield and different paths the wavefield takes through the subsurface.

[0036] It is appreciated that the wavefield can comprise an extended area, or an extended space, or an extended path, or an extended path volume or an extended path area taken up by a signal (e.g., a seismic wave or a seismic signal) transmitted and / or received via the subsurface by a set of sensors (e.g., seismic sensors). According to one embodiment, the moveout event data (also referred to as an event moveout) referenced above may be used for subsurface structure inference operations (e.g., subsurface model generation) associated with seismic tomographic inversion.Overview

[0037] In the proposed workflow, event data (e.g., derived from received seismic data) identified on a stack or a common surface offset and / or subsurface angle and / or some other offset parameter (denoted by h in FIG. 1A) may be tracked through an offset dimension relative to each x-y location or area associated with the subsurface or relative to each x-y-z location orvolume associated with the subsurface. This can be thought of as a surface tracking problem in a 4-dimensional space given by (x, y, z, h (offset dimension)) as further discussed in association with FIG. 1A.

[0038] FIG. 1A shows one or more stack volumes associated with a resource site used to generate a model of a subsurface of the resource site. As can be seen in FIG. 1A, one or more stack volumes 102 associated with a resource site may be used to generate a 4-dimensional model of the subsurface of the resource site. In particular, comprised in the one or more stack volumes 102 of FIG. 1A is a plurality of offsets hxysorted along a plurality of lines (e.g., geological lines indicated in the figure as solid lines, dashed arrow lines, and solid arrow lines). The one or more stack volumes 102 can also include a training offset hcassociated with, for example, one or more common image point (CIP) tomography panels which may be used to configure or otherwise train the 4-dimensional model. In addition, the one or more stack volumes may also include a plurality of inference offsets hc+iwhich may be used for inference operations associated with the 4-dimensional model such that i = 1, 2,and hc+Nrepresents the largest offset value for the inference operations. It is appreciated that the offsets depicted in FIG. 1A need not be visualized or displayed as shown in the figure. For example, the offset hccan represent a central or middle offset with i representing either positive or negative integers as the case may require.

[0039] According to one embodiment, the 4-dimensional model may comprise or be defined by x-y-z-hxydimensions as shown in the figure. However, it is contemplated that the model generated may be a 5 -dimensional model defined by x-y-z-hx-hydimensions or a 3- dimensional subset model of the 4-dimensional model. It is appreciated that the x, y, and z geoparameters refer to geolocation points associated with, for example, sensor data associated with the subsurface being observed. Furthermore, the subsurface can be represented as a volume defined by the geo-parameters x, y, and z. For example, assume two or more sensors (e.g., subsurface sensors, surface sensors, or other geological sensors) are used as part of tomographic inversion operations such that a first sensor comprised in the two or more sensors is separated by an offset hcfrom a source (e.g., a source sensor transmitting or propagating seismic signals in the subsurface). The distance between the source and any sensor comprised in the two or more sensors may be given by h = hc+ A / i, where A / i represents a distance between any given sensor comprised in the two or more sensors and the first sensor. For example, Ah = 0 if onlythe first sensor is being observed (e.g., the first sensor is receiving a propagated seismic wavefield). Assuming the first sensor is the source sensor in an alternate embodiment, then the first sensor may transmit (e.g., when a source is co-located with the first sensor (hc= 0)) and / or register a seismic signal (e.g., a wavefield) that travels through the subsurface such that the transmitted wavefield can be observed by a second sensor, which registers one or more targets hit or otherwise affected by the propagating seismic wavefield through the subsurface. The wavefield can move through a space or volume within the subsurface such that the space or volume may be characterized by the geo-parameters x-y-z (e.g., x x y x z). According to one embodiment, the value of hcis based on training data or simulation data, or testing data.

[0040] In some embodiments, the seismic wavefield can illuminate or register or log subsurface characteristics such as geological boundary data, and / or segment data, and / or rock property data, and / or fluid flow condition data, and / or air gap data, and / or subsurface discontinuity data, and / or subsurface layering data, and / or hydrocarbon data associated with the subsurface, and / or mineral deposit data associated with the subsurface, etc., within the volume characterized by the geo-parameters x-y-z or within an area in the subsurface characterized by x-y geo-parameters or y-z geo-parameters or x-z geo-parameters or a combination of areas defined by x-y, y-z, and x-z geo-parameters. According to some embodiments, the first and the second sensors discussed above, as well as other sensors may be located closer or farther from each other (for instance, the first sensor may be located with, for example, an offset distance value of hcfrom a source and from the second sensor with an offset distance value of hc+ A from the source, such that both hcand A may vary) thereby contracting or expanding the observation volume x-y-z, respectively. This would respectively decrease or increase the value of separation between the first and second sensors, as well as other sensors registering wavefield arrivals.

[0041] Also noteworthy is the effect of relative offset values in the modeling process as exemplified in FIG. IB. In particular, FIG. IB shows an exemplary propagation of sensor signals or wavefields having one or more offset values into a subsurface of the resource site. According to some embodiments, the offset h (e.g., h1?h2, and h3in FIG. IB) is not indicated by just the relative distance (e.g., surface offset) between a transmitting sensor and a receiving sensor. Rather, the offset h may indicate an angular displacement / deviation (e.g., referred to as subsurface angle elsewhere herein) of the first sensor relative to the second sensor. In someimplementations, the offset h indicates a reflection angle or an incident angle associated with a propagated wavefield from the first sensor such that the reflection angle or incident angle characterizes an angular deviation of the wavefield after hitting a target in the subsurface before traveling to the second sensor. In particular, and as shown in FIG. IB, the angular deviation can represent an angular directional change of a signal or wavefield once the signal or wavefield is reflected by one or more of the aforementioned subsurface characteristics or geo-targets before arriving at the second sensor.

[0042] According to one embodiment, the ML operations for selecting optimal picks may comprise determining, using picked surface data at a fixed offset ( / ic) volume data and / or other representative volume data, which may be a stack and / or coherency dip stack and / or equivalent in offset / angle-azimuth and / or other dimensions, for a propagated wavefield. In one embodiment, the volume data may comprise or be associated with an over-burden data class or an under-burden data class respectively indicating a first bounded segment above an interface comprised in the picked surface data and a second bounded segment below the interface comprised in the picked interface data. According to one embodiment, a plurality of classes comprising the over-burden class and / or the under-burden class and / or classes comprising whole geological sections of the resource site simultaneously for a plurality of boundaries with no over- or -under-burden references may be selected to indicate a plurality of bounded segments associated with the subsurface based on a reference point (e.g. the interface or a plurality of interfaces / geological boundaries comprised in the picked surface data).

[0043] In some implementations, a learning model (e.g., a tomographic picking learning model) with a trainable data layout structure (e.g., a neural network, an intelligent data structure, a structure of an interconnected data learning pathway, an artificial intelligence engine) associated with the ML operations may be parameterized or otherwise configured, or labeled to optimize the identification of the plurality of classes. In one embodiment, the trainable data layout structure may be labeled to indicate how the offset h associated with the volume data (e.g., volume data associated with the wavefield propagating through a volume of the subsurface) should be segmented as part of determining the volume data. In other embodiments, training data may be used for the parameterization of the learning model. The training data, for example, can comprise historical data or simulation data that indicate initial boundary' conditions for training the trainable data layout structure. In one embodiment, parameterizingthe trainable data layout structure using the training data enables the learning model to be applied to analyzing a plurality of offset values h during current and future picking operations.

[0044] While the subsurface volume defined by parameters x, y, and z may be the same for different offsets h, it is appreciated that the traveling wavefield within a given volume may experience a plurality of variation for the different offsets h as the wavefield travels through different areas or volumes of the subsurface while registering the plurality of different offsets h. After generating the volume data, one or more training examples may be formed using the volume data based on the trainable data layout structure of the learning model. In one embodiment, the training examples provide result data that confirm an optimality of one or more of the volume data and / or the learning model

[0045] According to one implementation, one or more analysis or inference operations may be executed using received seismic data based on the volume data and / or the learning model. The analysis operation may comprise executing one or more picking / selecting operations using the seismic data. For example, seismic data (e.g., historically received seismic data or newly received seismic data or real-time seismic data from a set of sensors deployed at a resource site) may be analyzed using the volume data together with the learning model by visually designating (e.g., textual coding or color coding) the volume data based on the seismic data to indicate one or more offset run inferences for offset h locations associated with the volume data. In implementations where the visual designation comprises color coding, colors such as red (e.g., a red star) or green (e.g., a green star) may each correspond to a fixed value of the offset / i.

[0046] It is appreciated that the captured seismic data being analyzed may comprise a segmented x-y-z volume (or a segmented area x-y) with offset h values that characterize (e.g., within an image of the subsurface) one or more lumps of classified pixels. Each lump of classified pixels, for example, may correspond to a geological segment with an assigned code associated with the learning model, such that the lumps of classified pixels may include textual or symbolic characters (e.g., stars) that indicate boundaries between subsurface segments which, for example, are assigned by the learning model during the analysis of the seismic data.

[0047] While hcmay be denoted by one or more identifiers (e.g., see star identifiers in FIG. 1A, dots, etc.), for example, and may represent a closest distance from a receiving sensor to a source sensor, hc(training offset) may also be determined for a plurality of sensorconfigurations relative to the source using a stack of volume data. As applied to a U-Net and DeepLab ML architectures, for example, the analysis of the seismic data based on the volume and / or learning model may facilitate semantic segmentation operations that segment different parts of transmitted wavefields interacting with subsurface structures.

[0048] In one embodiment, the volume data may be resolved into 2-dimensional area data parameterized by x-y geo-coordinates or x-z geo-coordinates or y-z geo-coordinates instead of the 3-dimensional volume data parameterized by x-y-z coordinates discussed above. In some embodiments, the 3 -dimensional volume data may be resolved into smaller volumes which may be further analyzed and or processed to improve the resolution of, for example, a geological model generated using the disclosed techniques. The 2-dimensional area data for example, may indicate 2-dimensional patch information associated with the subsurface such that the patch information may represent a 2-dimensional windowed view of an input wavefield (e.g., a transmitted seismic wavefield) within the subsurface. In the case of a reduced volume, for example, the 3 -dimensional volume (x-y-z) data may have dimensions 1000, 1000, 1000 respectively representing x, y, and z depth dimensions. This volume data may be split into overlapping regions of 128, 128, 128 dimensions which are processed at a more granular / micro level relative to processing the volume data at the macro level (e.g., processing the 1000, 1000, 1000 depth dimensions of the volume data).

[0049] This has the benefit of adapting the data size being analyzed based on the processing capacity of a signal processing engine and / or a screen size of a user device and has the added advantage of targeted analysis of the seismic wavefield at specific locations within the subsurface. In addition, this approach of resolving the macro volume into 2-dimensional patches or reduced 3 -dimensional volumes has the benefit of improving lateral and / or medial continuity or similarity or consistency between picks and thereby ensures visual alignment of said picks with one or more representative geological interfaces within the subsurface as seen in the received wavefield. It is appreciated that the 2-dimensional patches and / or reduced 3- dimensional volumes may comprise inline and / or crossline depth dimensions that facilitate the lateral continuity discussed above. In one embodiment, the forgoing benefits of resolving the macro volume (e.g., volume data) into 2-dimensional patches or reduced 3-dimensional volumes enables the use of trained structures (e.g., spatial extent of perception fields) comprised in the learning model such as convolutional neural networks (CNN) to analyze or otherwiseslide across the 2-dimensional patches and / or 3-dimensional reduced volumes and thereby analyze same.

[0050] It is appreciated that the CNN Kernel, for example, can enable analysis of the 2- dimensional patches and / or 3-dimensional reduced volumes at structural point(s) (e.g., discrete structural points / positions such as a neuron of the CNN Kernel) as part of the analysis of the wavefield. The structural point(s) of the learning model, for example, can be activated to provide output values (e.g., high output values) relative to a reference (e.g., zero output) output value if a similar pattern learnt by the Kernel is present in an input fragment (e.g., 2-dimensional patch or 3-dimensional reduced volume). The perceptive field of a neuron, for example, can correspond to an effective Kernel size through which the input fragment is fed such that the Kernel size corresponds to the input fragment size, according to some embodiments.

[0051] Given a dynamic wavelet distortion (e.g., frequency content variation comprised in a received seismic wavefield), it is necessary to factor such wave distortions, noise, or anomalies into the design or training of the learning model for optimal analysis of the seismic data. For example, dynamic wavelet distortion including signature artifacts associated with the source from which the seismic signal was transmitted may comprise complicated waveforms as well as a multipath wavefield (e.g., a wavefield traveling through different paths) due to different offsets h and subsurface variation characteristics, and may have different characteristics with respect to said offsets h such that the learning model (e.g., a CNN associated with the learning model) may fail for offsets that are out of range (e.g., far away offset values) relative to offsets hcused for training the learning model. Various training strategies including band-pass filtering, band-stop filtering, low-pass filtering, high-pass filtering, and other signal processing methodologies may be applied to received seismic data to clean the seismic data prior to processing by, for example, the learning model to match, for example, the frequency content of out of range offsets h and thereby enable the robustness of the learning model for said out of range offsets h. For instance, filtering of volume data associated with one or more offset values h and / or associated with the received seismic data may improve results generated for out of range offset values h.

[0052] According to some embodiments, the segmentation operations (e.g., semantic segmentation operations) may be conducted to allow for segmentation along a sequence of “frames” corresponding to offsets or angles or azimuths such as flow field generation networks,recurrent networks, and transformers. The transformers, for example, can allow for representative event identification based on a detected energy variation between “frames” associated with velocity variation within the subsurface, according to some implementations. In some embodiments, the segmentation operations are based on lumps of pixel data relative to a set of predefined categories. For instance, the segmentation operation enables the detection of localized event data indicating specific localized geological structures or characteristics as opposed to merely detecting boundaries between said localized geological structures. The problem may as well be formulated as detecting lumps of pixel data corresponding to boundaries or fragments around boundaries of interest; or / and as object detection, or / and a formulation allowing for pick prediction along a set of “frames" within captured seismic data.

[0053] The training process may include an augmentation strategy that boosts the amount of available labels, for instance elastic transformations. Other procedures that bring the training sample distribution closer to that of inference-based data for any volume data mentioned above may be applied. In one embodiment, new training data may be generated while performing inference along a set of "frames” corresponding to offsets or angles or azimuths associated with the subsurface through which the seismic data travels such that each subsequent frame may be used for additional training. A semi-supervised learning / training approach may also be used in combination with above-mentioned training or learning strategies to extend or otherwise expand the distribution of training data.

[0054] In one embodiment, set of sensors including marine sensors and / or land-based sensors may be used for transmitting and / or receiving the seismic wavefield via the subsurface. For example, such sensors may include hydrophonic sensors, geophonic sensors, broadband sensors, distributed acoustic sensors, and other sensors such as those discussed in the resource site of FIG. 2. In one embodiment, capturing or recording or receiving the propagated seismic wavefield can be done onshore, offshore, or in borehole settings. In one implementation, the received seismic wavefield may be subjected to signal processing operations and other preprocessing stages to remove noise including interference data and other random noise comprised in the received seismic wavefield. Furthermore, the received seismic wavefield may be formatted into a matrix data structure. The received seismic data may also be sorted into event data (also referred to as seismic gathers herein) by arranging recordings (e.g., the received seismic data or received propagated wavefield) from the set of sensors into groups illuminatingone or more subsurface locations associated with volume data (e.g., based on x, y, z coordinates) as well as the offset h. Thus, for each subsurface point, a multitude of records of various offset parameters h may be determined based on travel and location properties of the transmitted seismic wavefield.

[0055] According to one embodiment, segmented wavefields may be associated with: a set of offsets h; or a set of subsurface angles defined by, or associated with the offsets h; or a set of azimuths defined by or associated with the offsets h. In some cases, the segmented wavefields comprise geological boundaries associated with subsurface segments (e.g., segments with one or more geo-targets that indicate event data). In particular, when a wavefield is travelling through a particular part of the subsurface, geological structures within the subsurface affect or otherwise interact with the wavefield and thus encode or register information (e.g., event data) on the wavefield which can be extracted, picked, or selected using the disclosed techniques. Specifically, the learning model disclosed above can be used to extract or select or pick event data (e.g., indicating geological properties of the subsurface) comprised in the received seismic wavefield. The learning model may have a trainable data layout structure such as a neural network which can learn patterns present in the received wavefield for one or more offset values h with respect to geological boundaries of interest defined or characterized by the one or more coordinates (x, y, z) comprised in the volume data. Once a new offset h is presented to the trainable data layout structure of the learning model, the learning model applies its knowledge to find / identify similar and / or dissimilar geological boundary data comprised in the received seismic wavefield (also referred to as propagated wavefield elsewhere herein). Thus, the learning model can generate as output comprising one or more segmented wavefields corresponding to one or more offsets values, or one or more subsurface angular values, or one or more azimuthal values. In one embodiment, one or more geological boundaries can be indicated (e g., generated, drawn, patterned, or mapped) using geological boundaries derived from or between the segmented wavefields in a multidimensional image of the subsurface through which the seismic wavefield traveled. It is appreciated that the proposed approach enables selecting spatially consistent event data (e.g., seismic event data) associated with wavefield arrival time detection from geological boundaries of interest within the subsurface. In particular, the disclosed technique allows the learning model to analyze arrival times of the transmitted seismic wavefield on a point-by-point basis or based on a spatialview limited by a low number of data points associated with the subsurface, according to some embodiments.Resource Site

[0056] FIG. 2 shows a cross-sectional view of a resource site 200 for which the process of FIG. 1 may be executed. While the illustrated resource site 200 represents a subterranean formation, the resource site, according to some embodiments, may be below water bodies such as oceans, seas, lakes, ponds, wetlands, rivers, etc. According to one embodiment, various measurement tools capable of sensing one or more parameters such as seismic two-way travel time, density, resistivity, production rate, etc., of a subterranean formation and / or geological formations may be provided at the resource site. As an example, wireline tools may be used to obtain measurement information related to geological attributes (e.g., geological attributes of a wellbore and / or reservoir) including geophysical and / or chemical information. For example, the chemical information may include chemical information associated with the subsurface and / or chemical information associated with the surface / above ground areas of the resource site 200.

[0057] In some embodiments, various sensors may be located at various locations around the resource site 200 to monitor and collect data for executing the process of FIG. 4. In other embodiments, the techniques disclosed herein may be applied to surface seismic monitoring applications, surface gravity applications, surface electromagnetic applications, surface ground heave applications, and surface measurement of induced seismicity applications. According to some implementations, the disclosed techniques may be applied to remote sensing applications (e.g., satellite-based measurements), subsea applications associated with permanent sensors, temporary sensor applications, applications associated with remotely operated vehicles, and applications associated with aerial-based measurements (e.g., performed from planes, helicopters, and / or drones). Such measurements may include Synthetic Aperture Radar data, atmospheric concentration data associated with molecules such as CO2, CH4, and / or gas concentration data associated with gases within the seabed.

[0058] Part, or all, of the resource site 200 may be on land, on water, or below water.In addition, while a resource site 200 is depicted, the technology described herein may be used with any combination of one or more resource sites (e.g., multiple oil fields or multiple wellsites,one or more saline aquifers, one or more depleted oil / gas fields, etc.), one or more processing facilities, etc. As can be seen in FIG. 2, the resource site 200 may have data acquisition tools 202a, 202b, 202c, and 202d positioned at various locations within the resource site 200. The subterranean structure 204 may have a plurality of geological formations 206a-206d. As shown, this structure may have several formations or layers, including a shale layer 206a, a carbonate layer 206b, a shale layer 206c, and a sand layer 206d. A fault 207 may extend through the shale layer 206a and the carbonate layer 206b. The data acquisition tools, for example, may be adapted to take measurements and detect geophysical and / or chemical characteristics of the various formations shown.

[0059] While a specific subterranean formation with specific geological structures is depicted, it is appreciated that the oil field 200 may contain a variety of geological structures and / or formations, sometimes having extreme complexity. In some locations of a given geological structure, for example below a water line (e.g., aquifer) relative to the given geological structure, fluid may occupy pore spaces of the formations. Each of the measurement devices may be used to measure properties of the formations and / or other geological features. While each data acquisition tool is shown as being in specific locations in FIG. 2, it is appreciated that one or more types of measurement may be taken at one or more locations across one or more sources of the resource site 200 or other locations for comparison and / or analysis. The data collected from various sources at the resource site 200 may be processed and / or evaluated and / or used as training data, and or used to generate high resolution result sets for characterizing a resource at the resource site, and / or used for generating resource models, etc. In one embodiment, the data collected by a set of sensors at the resource site may include data associated with the number of wells of a first reservoir or second reservoir at the resource site, data associated with the number of grid cells of the first or second reservoir, data associated with the average permeability of the first or second reservoir, data associated with the production duration history (e.g., number of years of production) of the first reservoir or second, etc.

[0060] Data acquisition tool 202a is illustrated as a measurement truck, which may comprise devices or sensors that take measurements of the subsurface through sound vibrations such as, but not limited to, seismic measurements. Drilling tool 202b may include a downhole sensor adapted to perform logging while drilling (LWD) data collection. The wireline tool 202cmay include a downhole sensor deployed in a wellbore or borehole. Production tool 202d may be deployed from a production unit or Christmas tree into a completed wellbore. Examples of parameters that may be measured include weight on bit, torque on bit, subterranean pressures (e.g., underground fluid pressure), temperatures, flow rates, compositions, rotary speed, particle count, voltages, currents, and / or other parameters of operations as further discussed below.

[0061] Sensors may be positioned about the resource site to collect data relating to various resource site operations, such as sensors deployed by the data acquisition tools 202. The sensor may include any type of sensor such as a metrology sensor (e.g., temperature, humidity), an automation enabling sensor, an operational sensor (e.g., pressure sensor, H2S sensor, thermometer, depth, tension), evaluation sensors, that can be used for acquiring data regarding the formation, wellbore, formation fluid / gas, wellbore fluid, gas / oil / water comprised in the formation / wellbore fluid, or any other suitable sensor. For example, the sensors may include accelerometers, flow rate sensors, pressure transducers, electromagnetic sensors, acoustic sensors, temperature sensors, chemical agent detection sensors, nuclear sensor, and / or any additional suitable sensors. In one embodiment, the data captured by the one or sensors may be used to characterize, or otherwise generate one or more parameter values for a high resolution result set used to, for example, label or configure a machine learning (ML) engine, a resource model as the case may require. In other embodiments, test data or synthetic data may also be used in developing the ML engine or resource model via one or more parameterization / labeling operations such as those discussed in association with the workflows presented herein.

[0062] Evaluation sensors may be featured in downhole tools such as tools 202b-202d and may include for instance electromagnetic, acoustic, nuclear, and optic sensors. Examples of tools including evaluation sensors that can be used in the framework of the current method include electromagnetic tools including imaging sensors such as FMI™ or QuantaGeo™ (mark of Schlumberger, Houston, TX); induction sensors such as Rt Scanner™ (mark of Schlumberger, Houston, TX), multifrequency dielectric dispersion sensor such as Dielectric Scanner™ (mark of Schlumberger, Houston, TX); acoustic tools including sonic sensors, such as Sonic Scanner™ (mark of Schlumberger, Houston, TX) or ultrasonic sensors, such as pulseecho sensor as in UBI™ or PowerEcho™ (marks of Schlumberger, Houston, TX) or flexural sensors PowerFlex™ (mark of Schlumberger, Houston, TX); nuclear sensors such as LithoScanner™ (mark of Schlumberger, Houston, TX) or nuclear magnetic resonance sensors; fluid sampling tools including fluid analysis sensors such as InSitu Fluid Analyzer ™ (mark of Schlumberger, Houston, TX); distributed sensors including fiber optic. Such evaluation sensors may be used in particular for evaluating the formation in which the well is formed (z.c., determining petrophysical or geological properties of the formation), for verifying the integrity of the well (such as casing or cement properties) and / or analyzing the produced fluid (flow, type of fluid, etc.).

[0063] As shown, data acquisition tools 202a-202d may generate data plots or measurements 208a-208d, respectively. These data plots are depicted within the resource site 200 to demonstrate that data generated by some of the operations executed at the resource site 200.

[0064] Data plots 208a-208c are examples of static data plots that may be generated by data acquisition tools 202a-202c, respectively. However, it is herein contemplated that data plots 208a-208c may also be data plots that may be generated and updated in real time. These measurements may be analyzed to better define properties of the formation(s) and / or determine the accuracy of the measurements and / or check for and compensate for measurement errors. The plots of each of the respective measurements may be aligned and / or scaled for comparison and verification purposes. In some embodiments, base data associated with the plots may be incorporated into site planning, modeling a test at the resource site 200. The respective measurements that can be taken may be any of the above.

[0065] Other data may also be collected, such as historical data of the resource site 200 and / or sites similar to the resource site 200, user inputs, information (e.g., economic information) associated with the resource site 200 and / or sites similar to the resource site 200, and / or other measurement data and other parameters of interest. Similar measurements may also be used to measure changes in formation aspects over time.

[0066] Computer facilities such as those discussed in association with FIG. 3 may be positioned at various locations about the resource site 200 (e.g., a surface unit) and / or at remote locations. A surface unit (e.g., one or more terminals 320) may be used to communicate with the onsite tools and / or offsite operations, as well as with other surface or downhole sensors. The surface unit may be capable of sending commands to the oil field equipment / systems, and receiving data therefrom. The surface unit may also collect data generated during productionoperations and can produce output data, which may be stored or transmitted for further processing.

[0067] The data collected by sensors may be used alone or in combination with other data. The data may be collected in one or more databases and / or transmitted on or offsite. The data may be historical data, real time data, or combinations thereof. The real time data may be used in real time, or stored for later use. The data may also be combined with historical data or other inputs for further analysis or for modeling purposes to optimize production processes at the oil field 200. In one embodiment, the data is stored in separate databases, or combined into a single database.High-Level Networked System

[0068] FIG. 3 shows a high-level networked system diagram illustrating a communicative coupling of devices or systems associated with the resource site 200 as described in FIG. 2. The system shown in the figure may include a set of processors 302a, 302b, and 302c for executing one or more processes discussed herein. The set of processors 302 may be electrically coupled to one or more servers (e g., computing systems) including memory 306a, 306b, and 306c that may store for example, program data, databases, and other forms of data. Each server of the one or more servers may also include one or more communication devices 308a, 308b, and 308c. The set of servers may provide a cloudcomputing platform 310. In one embodiment, the set of servers includes different computing devices that are situated in different locations and may be scalable based on the needs and workflows associated with the oil field 200. The communication devices of each server may enable the servers to communicate with each other through a local or global network such as an Internet network. In some embodiments, the servers may be arranged as a town 312, which may provide a private or local cloud service for users. A town may be advantageous in remote locations with poor connectivity. Additionally, a town may be beneficial in scenarios with large networks where security may be of concern. A town in such large network embodiments can facilitate implementation of a private network within such large networks. The town may interface with other towns or a larger cloud network, which may also communicate over public communication links. Note that cloud-computing platform 310 may include a private networkand / or portions of public networks. In some cases, a cloud-computing platform 310 may include remote storage and / or other application processing capabilities.

[0069] The system of FIG. 3 may also include one or more user terminals 314a and314b each including at least a processor to execute programs, a memory (e.g., 316a and 316b) for storing data, a communication device and one or more user interfaces and devices that enable the user to receive, view, and transmit information. In one embodiment, the user terminals 314a and 314b is a computing system having interfaces and devices including keyboards, touchscreens, display screens, speakers, microphones, a mouse, styluses, etc. The user terminals 314 may be communicatively coupled to the one or more servers of the cloudcomputing platform 310. The user terminals 314 may be client terminals or expert terminals, enabling collaboration between clients and experts through the system of FIG. 3.

[0070] The system of FIG. 3 may also include at least one or more resource sites 200 having, for example, a set of terminals 320, each including at least a processor, a memory, and a communication device for communicating with other devices communicatively coupled to the cloud-computing platform 310. The resource site 200 may also have a set of sensors (e.g., one or more sensors described in association with FIG. 2) or sensor interfaces 322a and 322b communicatively coupled to the set of terminals 320 and / or directly coupled to the cloudcomputing platform 310. In some embodiments, data collected by the set of sensors / sensor interfaces 322a and 322b may be processed to generate a one or more resource models (e.g., reservoir models) or one or more resolved data sets used to generate the resource model which may be displayed on a user interface associated with the set of terminals 320, and / or displayed on user interfaces associated with the set of servers of the cloud computing platform 310, and / or displayed on user interfaces of the user terminals 314. Furthermore, various equipment / devices discussed in association with the resource site 200 may also be communicatively coupled to the set of terminals 320 and or communicatively coupled directly to the cloud-computing platform 310. The equipment and sensors may also include one or more communication device(s) that may communicate with the set of terminals 320 to receive orders / instructions locally and / or remotely from the resource site 200 and also send statuses / updates to other terminals such as the user terminals 314.

[0071] The system of FIG. 3 may also include one or more client servers 324 including a processor, memory and communication device. For communication purposes, the clientservers 324 may be communicatively coupled to the cloud-computing platform 310, and / or to the user terminals 314a and 314b, and / or to the set of terminals 320 at the resource site 200 and / or to sensors at the oil field, and / or to other equipment at the resource site 200.

[0072] A processor, as discussed with reference to the system of FIG. 3, may include a microprocessor, a graphical processing unit (GPU), a microcontroller, a processor module or subsystem, a programmable integrated circuit, a programmable gate array, or another control or computing device.

[0073] The memory / storage media discussed above in association with FIG. 3 can be implemented as one or more computer-readable or machine-readable storage media that are non-transitory. In some embodiments, storage media may be distributed within and / or across multiple internal and / or external enclosures of a computing system and / or additional computing systems. Storage media may include one or more different forms of memory including semiconductor memory devices such as dynamic or static random access memories (DRAMs or SRAMs), erasable and programmable read-only memories (EPROMs), electrically erasable and programmable read-only memories (EEPROMs) and flash memories; magnetic disks such as fixed, floppy and removable disks; other magnetic media including tape; optical media such as compact disks (CDs) or digital video disks (DVDs), BluRays or any other type of optical media; or other types of storage devices. “Non-transitory” computer readable medium refers to the medium itself (i.e., tangible, not a signal) and not data storage persistency (e.g., RAM vs. ROM).

[0074] Note that instructions can be provided on one computer-readable or machine- readable storage medium, or alternatively, can be provided on multiple computer-readable or machine-readable storage media distributed in a large system having possibly plural nodes and / or non-transitory storage means. Such computer-readable or machine-readable storage medium or media is (are) considered to be part of an article (or article of manufacture). The storage medium or media can be located either in a computer system running the machine- readable instructions, or located at a remote site from which machine-readable instructions can be downloaded over a network for execution.

[0075] It is appreciated that the described system of FIG. 3 is an example that may have more or fewer components than shown, may combine additional components, and / or may have a different configuration or arrangement of the components. The various components shownmay be implemented in hardware, software, or a combination of both, hardware and software, including one or more signal processing and / or application specific integrated circuits.

[0076] Further, the steps in the flowcharts described below may be implemented by running one or more functional modules in an information processing apparatus such as general-purpose processors or application specific chips, such as ASICs, FPGAs, PLDs, GPUs or other appropriate devices associated with the system of FIG. 3. For example, the flowchart of FIG. 1 as well as the flowcharts below may be executed using a signal processing engine / a data processing module (e.g., computing module) stored in memory 306a, 306b, or 306c such that the signal processing engine / data processing module includes instructions that are executed by the one or more processors such as processors 302a, 302b, or 302c as the case may be. The various modules of FIG. 3, combinations of these modules, and / or their combination with general hardware are included within the scope of protection of the disclosure. While one or more computing processors (e.g., processors 302a, 302b, or 302c) may be described as executing steps associated with one or more of the flowcharts described in this disclosure, the one or more computing device processors may be associated with the cloud-based computing platform 310 and may be located at one location or distributed across multiple locations. In one embodiment, the one or more computing device processors may also be associated with other systems of FIG. 3 other than the cloud-computing platform 310.

[0077] In some embodiments, a computing system is provided that includes at least one processor, at least one memory, and one or more programs stored in the at least one memory, such that the programs comprise instructions, which when executed by the at least one processor, are configured to perform any method disclosed herein.

[0078] In some embodiments, a computer readable storage medium is provided, which has stored therein one or more programs, the one or more programs including instructions, which when executed by a processor, cause the processor to perform any method disclosed herein. In some embodiments, a computing system is provided that includes at least one processor, at least one memory, and one or more programs stored in the at least one memory for performing any method disclosed herein. In some embodiments, an information processing apparatus for use in a computing system is provided for performing any method disclosed herein.Flowchart

[0079] FIG. 4 provides an exemplary workflow 400 for methods, systems, and computer programs that select event data from seismic data. It is appreciated that a data managing module stored in a memory device may cause a computer processor to execute the various processing stages of FIG. 4. For example, the disclosed techniques may be implemented as a signal processing engine within a geological software tool such that the signal processing engine enables the modeling of geological structures in the subsurface of a resource site based on the processes outlined herein.

[0080] At block 402, the data manager may generate a learning model associated with a machine learning (ML) engine. The learning model may be parameterized using: first spatial data associated with one or more geo-targets in a subsurface of the resource site, and selected offset data associated with a set of sensors deployed at the resource site, and / or selected from a combination of sensors deployed at the resources site and / or selected from a stack (e.g., a sum of offset values, synthetic or otherwise) associated with the resource site.

[0081] At block 404, the data manager receives the seismic data from the set of sensors such that the seismic data comprises a propagated wavefield transmitted within the subsurface of the resource site by a first sensor comprised in the set of sensors deployed at the resource site, and / or a propagated wavefield received by a second sensor comprised in the set of sensors at the resource site.

[0082] At block 406, the data manager may format or otherwise generate a data matrix using the received seismic data, the data matrix comprising one or more multidimensional parameters. In one embodiment, the one or more multidimensional parameters represent second spatial data indicating the one or more geo-targets with which the propagated wavefield interacts, and / or a determined offset data indicating displacement data of the first sensor relative to the second sensor.

[0083] At block 408, the data manager may resolve the event data comprised in the seismic data to generate a segmented wavefield by applying the learning model to the data matrix. In particular, the data manager may generate a segmented wavefield by applying the learning model to the data matrix. The segmented wavefield, for example, may indicate a computational selection of the event data comprised in the received propagated wavefield.

[0084] At block 410, the data manager may generate a geological model of the subsurface of the resource site using the segmented wavefield, the geological model of thesubsurface of the resource site indicating at least a multidimensional image of the subsurface of the resource site. In one embodiment, the multidimensional image may be visualized on a display device and / or used to automatically execute a plurality of operations associated with energy research, geological research and other energy development operations as further discussed below.

[0085] These and other implementations may each optionally include one or more of the following features. Resolving (e.g., discretizing or quantizing regions in the subsurface) the event data or generating the segmented wavefield may comprises: identifying, using the learning model, one or more spatially consistent event data comprising the resolved event data, or associated with the received seismic data; selecting one or more geological boundaries associated with illuminated regions indicating the spatially consistent event data of the one or more geo-targets in the subsurface of the resource site, such that the illuminated regions are based on interactions of the propagated wavefield with the one or more geo-targets in the subsurface of the resource site; and generating, using the one or more geological boundaries, the segmented wavefield.

[0086] Furthermore, the set of sensors comprise one or more of: hydrophonic sensors, geophonic sensors, broadband sensors, or a distributed acoustic (DAS) sensors; and the resource site comprises one or more of: an onshore resource site or an offshore resource site.

[0087] In addition, the learning model may be parameterized using one or more of: training data including historical subsurface data associated with the first spatial data or the selected offset data; or simulation data indicating an initial boundary condition for the learning model based on the first spatial data or the selected offset data.

[0088] In some embodiments, the one or more multidimensional parameters of the data matrix include a parameter set associated with the second spatial data which in aggregate define one or more of area data or volume data such that: the one or more geo-targets are comprised in, or are proximal to a location associated with the area data or volume data, and the propagated wavefield travels through the location within the subsurface of the resource site.

[0089] Furthermore, the parameter set may comprise geo-parameters indicating geolocation points associated with the one or more geo-targets in the subsurface of the resource site.

[0090] In addition, the one or more multidimensional parameters of the data matrix may include a displacement parameter associated with the determined offset data. The displacement parameter indicating at least one of: an angular displacement of the propagated wavefield from the first sensor to the second sensor; an azimuthal displacement of the first sensor relative to the second sensor; or a distance between the first sensor and the second sensor.

[0091] It is appreciated that the one or more multidimensional parameters of the data matrix may include a parameter set associated with the second spatial data which in aggregate define one or more of area data or volume data according to some embodiments. It is further appreciated that the one or more multidimensional parameters of the data matrix include a displacement parameter associated with the determined offset data discussed in association with FIG. 4

[0092] According to some implementations, the propagated wavefield logs, records, indicates, or captures one or more of: geological boundary data associated with the one or more geo-targets; rock property data associated with the one or more geo-targets; fluid flow condition data associated with the one or more geo-targets; air gap data associated with the one or more geo-targets; subsurface discontinuity data associated with the one or more geo-targets; subsurface layering data associated with the one or more geo-targets; hydrocarbon data associated with the one or more geo-targets; and mineral deposit data associated with the one or more geo-targets.

[0093] The learning model may comprise: a tomographic learning model with a trainable data layout structure; and the trainable data layout structure comprises one of: a neural network, an intelligent data structure, an artificial intelligence engine, or a structure of an interconnected data learning pathway.

[0094] In one embodiment, the learning model is a 4-dimensional model configured for selecting the event data comprised in the seismic data.

[0095] In other embodiments, a dimension of the learning model is based on the data matrix such that the dimension of the learning model is one of a 4-dimensional model or a 5- dimensional model.

[0096] In one embodiment, a dimension of the learning model is based on at least 3 parameters comprised in the one or more multidimensional parameters of the data matrix.

[0097] Moreover, generating or formatting the data matrix may comprises sorting or arranging at least event data comprised in the seismic data into one or more groups of data points associated with the one or more geo-targets.

[0098] In some cases, the multidimensional image of the subsurface of the resource site and referenced in association with FIG. 4 comprises one of a 2-dimensional image or a 3- dimensional image.

[0099] It is appreciated that the multidimensional image of the subsurface of the resource site is used to initiate one or more of: adjusting a drill bit spin rate at the resource site based on geological data comprised in the multidimensional image; regulating one or more flow control devices at the resource site based on fluid data comprised in the multidimensional image; optimizing gas storage operations in the subsurface of the resource site based on the multidimensional image; and optimizing energy development operations at the resource site based on the multidimensional image.

[0100] According to one embodiment, the seismic data comprises subsurface volume data corresponding to the selected offset data that characterize one or more lumps of classified pixels within the multidimensional image with each lump of the one or more classified pixels corresponding to a geological segment that has an assigned code based on the learning model. In other embodiments, the seismic data comprises subsurface area data corresponding to the selected offset data that characterize the one or more lumps of classified pixels within the multidimensional image with each lump of the one or more classified pixels corresponding to the geological segment that has an assigned code based on the learning model.Exemplary Segment Visualizations

[0101] FIGS. 5A-5D provide exemplary segment data associated with geological models generated using the disclosed techniques. FIG. 5A, for example provides an exemplary embodiment where the learning model is trained with training data comprising an offset value of hc= 350 meters. Such parameterization of the learning model can, for example, be used to generate a geological model of the resource site subsurface such that the geological model indicates the two segment data of FIG. 5B. Similarly, FIG. 5C shows an implementation where the learning model is trained with an offset value of hc= 3150 meters following which the learning model is used to generate the segment data shown in FIG. 5D.

[0102] While any discussion of or citation to related art in this disclosure may or may not include some prior art references, Applicant neither concedes nor acquiesces to the position that any given reference is prior art or analogous prior art.

[0103] The foregoing description, for purpose of explanation, has been described with reference to specific embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings. The embodiments were chosen and described in order to explain the principles of the invention and its practical applications, to thereby enable others skilled in the art to use the invention and various embodiments with various modifications as are suited to the particular use contemplated. It is appreciated that the term optimize / optimal and its variants (e.g., efficient or optimally) may simply indicate improving, rather than the ultimate form of 'perfection' or the like.

[0104] It will also be understood that, although the terms first, second, etc., may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used to distinguish one element from another. For example, a first object or step could be termed a second object or step, and, similarly, a second object or step could be termed a first object or step, without departing from the scope of the invention. The first object or step, and the second object or step, are both objects or steps, respectively, but they are not to be considered the same object or step.

[0105] The terminology used in the description herein is for the purpose of describing particular embodiments and is not intended to be limiting. As used in the description of the invention and the appended claims, the singular forms “a,” “an” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that the term “and / or” as used herein refers to and encompasses any possible combination of one or more of the associated listed items. It will be further understood that the terms “includes,” “including,” “comprises” and / or “comprising,” when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0106] As used herein, the term “if’ may be construed to mean “when” or “upon” or “in response to determining” or “in response to detecting,” depending on the context.

[0107] Those with skill in the art will appreciate that while some terms in this disclosure may refer to absolutes, e.g., all source receiver traces, each of a plurality of objects, etc., the methods and techniques disclosed herein may also be performed on fewer than all of a given thing, e.g., performed on one or more components and / or performed on one or more source receiver traces. Accordingly, in instances in the disclosure where an absolute is used, the disclosure may also be interpreted to be referring to a subset.

Claims

CLAIMSWhat is claimed is:

1. A method for selecting event data from seismic data, the method comprising: generating a learning model associated with a machine learning (ML) engine, wherein the learning model is parameterized using: first spatial data associated with one or more geo-targets in a subsurface of a resource site, and selected offset data associated with a set of sensors deployed at the resource site; receiving the seismic data from the set of sensors, the seismic data comprising a propagated wavefield that is transmitted within the subsurface of the resource site by a first sensor of the set of sensors and received by a second sensor of the set of sensors; formatting a data matrix using the received seismic data, the data matrix comprising one or more multidimensional parameters that represent: second spatial data indicating the one or more geo-targets with which the propagated wavefield interacts, and determined offset data indicating displacement data of the first sensor relative to the second sensor; resolving event data comprised in the seismic data to generate a segmented wavefield by applying the learning model to the data matrix, the segmented wavefield indicating a computational selection of the event data comprised in the received propagated wavefield; and generating a geological model of the subsurface of the resource site using the segmented wavefield, the geological model of the subsurface of the resource site indicating at least a multidimensional image of the subsurface of the resource site.

2. The method of claim 1, wherein resolving the event data comprises: identifying, using the learning model, one or more spatially consistent event data associated with, or comprised in the data matrix; selecting one or more geological boundaries associated with illuminated regions indicating the spatially consistent event data of the one or more geo-targets in the subsurface of the resource site, such that the illuminated regions are based on interactions of the propagated wavefield with the one or more geo-targets in the subsurface of the resource site; andgenerating, using the one or more geological boundaries, the segmented wavefield.

3. The method of claim 1, wherein: the set of sensors comprise one or more of: hydrophonic sensors, geophonic sensors, broadband sensors, or a distributed acoustic (DAS) sensors; and the resource site comprises one or more of: an onshore resource site or an offshore resource site.

4. The method of claim 1, wherein the learning model is parameterized using one or more of: training data including historical subsurface data associated with the first spatial data or the selected offset data; or simulation data indicating an initial boundary condition for the learning model based on the first spatial data or the selected offset data.

5. The method of claim 1, wherein the one or more multidimensional parameters of the data matrix include a parameter set associated with the second spatial data which in aggregate define one or more of area data or volume data such that: the one or more geo-targets are comprised in, or are proximal to a location associated with the area data or volume data, and the propagated wavefield travels through the location within the subsurface of the resource site.

6. The method of claim 5, wherein the parameter set comprises geo-parameters indicating geolocation points associated with the one or more geo-targets in the subsurface of the resource site.

7. The method of claim 1, wherein the one or more multidimensional parameters of the data matrix include a displacement parameter associated with the determined offset data, the displacement parameter indicating at least one of:an angular displacement of the propagated wavefield from the first sensor to the second sensor, an azimuthal displacement of the first sensor relative to the second sensor, or a distance between the first sensor and the second sensor.

8. The method of claim 1, wherein the propagated wavefield logs one or more of: geological boundary data associated with the one or more geo-targets; rock property data associated with the one or more geo-targets; fluid flow condition data associated with the one or more geo-targets; air gap data associated with the one or more geo-targets; subsurface discontinuity data associated with the one or more geo-targets; subsurface layering data associated with the one or more geo-targets; hydrocarbon data associated with the one or more geo-targets; and mineral deposit data associated with the one or more geo-targets.

9. The method of claim 1, wherein the learning model comprises: a tomographic learning model with a trainable data layout structure; and the trainable data layout structure comprises one of: a neural network, an intelligent data structure, an artificial intelligence engine, or a structure of an interconnected data learning pathway.

10. The method of claim 1, wherein formatting the data matrix comprises sorting or arranging at least event data comprised in the seismic data into one or more groups of data points associated with the one or more geo-targets.

11. The method of claim 1 , wherein the learning model is a 4-dimensional model configured for selecting the event data comprised in the seismic data.

12. The method of claim 1, wherein a dimension of the learning model is based on the data matrix, the dimension of the learning model being one of a 4-dimensional model or a 5- dimensional model.

13. The method of claim 1, wherein the multidimensional image of the subsurface of the resource site comprises one of a 2-dimensional image or a 3-dimensional image.

14. The method of claim 1, wherein the multidimensional image of the subsurface of the resource site is used to initiate one or more of: adjusting a drill bit spin rate at the resource site based on geological data comprised in the multidimensional image; regulating one or more flow control devices at the resource site based on fluid data comprised in the multidimensional image; optimizing gas storage operations in the subsurface of the resource site based on the multidimensional image; and optimizing energy development operations at the resource site based on the multidimensional image.

15. A system for selecting event data from seismic data, the system comprising: a computer processor, and memory storing a data processing engine that comprises instructions which are executable by the computer processor to: generate a learning model associated with a machine learning (ML) engine, wherein the learning model is parameterized using: first spatial data associated with one or more geo-targets in a subsurface of a resource site, and selected offset data associated with the a set of sensors deployed at the resource site; receive the seismic data from the set of sensors, the seismic data comprising a propagated wavefield that istransmitted within the subsurface of the resource site by a first sensor of the set of sensors, and received by a second sensor of the set of sensors; format a data matrix using the received seismic data, the data matrix comprising one or more multidimensional parameters that represent: second spatial data indicating the one or more geo-targets with which the propagated wavefield interacts, and determined offset data indicating displacement data of the first sensor relative to the second sensor; generate a segmented wavefield by applying the learning model to the data matrix, the segmented wavefield indicating computationally selected event data comprised in the received propagated wavefield; and generate a geological model of the subsurface of the resource site using the segmented wavefield, the geological model of the subsurface of the resource site indicating at least a multidimensional image of the subsurface of the resource site.

16. The system of claim 15, wherein to generate the segmented wavefield comprises: identifying, using the learning model, one or more spatially consistent event data associated with the received seismic data; selecting one or more geological boundaries associated with illuminated regions indicating the spatially consistent event data of the one or more geo-targets in the subsurface of the resource site, such that the illuminated regions are based on interactions of the propagated wavefield with the one or more geo-targets in the subsurface of the resource site; and generating, using the one or more geological boundaries, the segmented wavefield.

17. The system of claim 15, wherein: the one or more multidimensional parameters of the data matrix include a parameter set associated with the second spatial data which in aggregate define one or more of area data or volume data; and the one or more multidimensional parameters of the data matrix include a displacement parameter associated with the determined offset data.

18. A computer program for selecting event data from seismic data, the computer program comprising a non -transitory computer-readable medium comprising code configured to: generate a learning model associated with a machine learning (ML) engine, wherein the learning model is parameterized using: first spatial data associated with one or more geo-targets in a subsurface of a resource site, and selected offset data associated with a set of sensors deployed at the resource site; receive the seismic data from the set of sensors, the seismic data comprising a propagated wavefield that is: transmitted within the subsurface of the resource site by a first sensor of the set of sensors, and received by a second sensor of the set of sensors; generate a data matrix using the received seismic data, the data matrix comprising one or more multidimensional parameters that represent: second spatial data indicating the one or more geo-targets with which the propagated wavefield interacts, and determined offset data indicating displacement data of the first sensor relative to the second sensor; generate a segmented wavefield by applying the learning model to the data matrix, the segmented wavefield indicating computationally selected event data comprised in the received propagated wavefield; and generate a geological model of the subsurface of the resource site using the segmented wavefield, the geological model of the subsurface of the resource site indicating at least a multidimensional image of the subsurface of the resource site.

19. The computer program of claim 18, wherein a dimension of the learning model is based on at least three parameters comprised in the one or more multidimensional parameters of the data matrix.

20. The computer program of claim 18, wherein the seismic data comprises one of: subsurface volume data corresponding to the selected offset data that characterize one or more lumps of classified pixels within the multidimensional image with each lump of the one or more classified pixels corresponding to a geological segment that has an assigned code based on the learning model; or subsurface area data corresponding to the selected offset data that characterize the one or more lumps of classified pixels within the multidimensional image with each lump of the one or more classified pixels corresponding to the geological segment that has an assigned code based on the learning model.

Images