WO2025199448A1 - A foundation model methodology for subsurface analysis and hydrocarbon identification - Google Patents

Abstract

A method for detecting hydrocarbon-related seismic anomalies includes receiving first input data. The method also includes training a plurality of subnet models based upon the first input data to produce a plurality of trained subnet models. The method also includes building a multi-modal foundation model using the trained subnet models. The method also includes receiving second input data. The method also includes generating a multi-modal feature space using the multi-modal foundation model based upon the second input data.

Description

A FOUNDATION MODEL METHODOLOGY FOR SUBSURFACE ANALYSIS AND HYDROCARBON IDENTIFICATION

Cross-Reference to Related Applications

[0001] This application claims priority to U.S. Provisional Patent Application No. 63/568,233, filed on March 21, 2024, which is incorporated by reference in its entirety.

Background

[0002] The challenge of identifying seismic anomalies for exploration purposes is caused by the diversity of geological settings for hydrocarbon accumulation and the limitations and ambiguity of the imaging capabilities of seismic data. For this reason, geoscientists go through extensive process of tests and validations to mature the identified seismic anomaly to a drillable prospect. In addition, the ability to learn from previous success or failures is limited by how many examples were available to the geoscientist. Moreover, applying the exact same standards from a case to another is a very interpretive task that is highly dependent on the geoscientist experience and skills. [0003] On the other hand, the ability of ML models to generalize over large set of instances of success and failure and apply the same standards when assessing each case makes it possible to detect the potential exploration success or failures in a more consistent manner. Therefore, what is needed is a system and method to leverage the artificial intelligence (Al) technologies to automate this process and increase the certainty of the results.

Summary

[0004] A method for detecting hydrocarbon-related seismic anomalies is disclosed. The method includes receiving first input data. The method also includes training a plurality of subnet models based upon the first input data to produce a plurality of trained subnet models. The method also includes building a multi-modal foundation model using the trained subnet models. The method also includes receiving second input data. The method also includes generating a multi-modal feature space using the multi-modal foundation model based upon the second input data.

[0005] A computing system is also disclosed. The computing system includes one or more processors and a memory system. The memory system includes one or more non-transitory computer-readable media storing instructions that, when executed by at least one of the one or more processors, cause the computing system to perform operations. The operations include receiving first input data. The first input data includes first subsurface data. The first subsurface data includes first formation tops data, first well log data from a plurality of first well logs, first post-stack seismic trace data, a first pre-stack seismic gather tile, and/or a first post-stack seismic image tile. The operations also include training a plurality of subnet models based upon the first input data to produce a plurality of trained subnet models. A first of the subnet models is trained based upon the first well log data from plurality of first well logs. A second of the subnet models is trained based upon the first pre-stack seismic gather tile. A third of the subnet models is trained based upon the first post-stack image tile. The operations also include building or updating a multi-modal foundation model using the trained subnet models. The multi-modal foundation model is built or updated based upon features extracted from the trained subnet models. The trained subnet models operate in parallel within the multi-modal foundation model. The operations also include receiving second input data. The second input data includes second subsurface data. The second subsurface data includes second formation tops data, second well log data from a plurality of second well logs, second post-stack trace data, a second pre-stack gather tile, and/or a second post-stack image tile. The operations also include generating a multi-modal feature space using the multi-modal foundation model based upon the second input data.

[0006] A non-transitory computer-readable medium is also disclosed. The medium stores instructions that, when executed by one or more processors of a computing system, cause the computing system to perform operations. The operations include receiving first input data. The first input data includes first subsurface data. The first subsurface data includes first formation tops data, first well log data from a plurality of first well logs, first post-stack seismic trace data, a first pre-stack seismic gather tile, and a first post-stack seismic image tile. Curves in the first well logs correspond to a position in the first post-stack trace data. A zero-offset trace of the first prestack seismic gather tile corresponds to the position. A vertical centerline of the first post-stack image tile corresponds to the position. The operations also include training a plurality of subnet models based upon the first input data to produce a plurality of trained subnet models. A first of the subnet models is trained based upon the first well log data from plurality of first well logs. A second of the subnet models is trained based upon the first pre-stack seismic gather tile. A third of the subnet models is trained based upon the first post-stack image tile. Each of the subnet models is trained to pre-process the first input data to produce pre-processed data, pool the pre- processed data to produce pooled data, and extract features from the pooled data after the pooled data is additionally processed. The extracted features are obtained by the subnet models that are trained based on the pooled data using self-supervised and/or multi-task approaches. The selfsupervised and/or multi-task approaches are used to train the subnet models in a way for the trained subnet models to extract the features, which contain sufficient information to solve for multiple downstream applications as a foundation model for a specific data type. The operations also include building or updating a multi-modal foundation model using the trained subnet models. The multi-modal foundation model is built or updated based upon the extracted features from the trained subnet models. The trained subnet models operate in parallel within the multi-modal foundation model. Weights of neurons of the trained subnet models are shared between the trained subnet models. The operations also include receiving second input data. The second input data includes second subsurface data. The second subsurface data includes second formation tops data, second well log data from a plurality of second well logs, second post-stack trace data, a second pre-stack gather tile, and a second post-stack image tile. The operations also include generating a multi-modal feature space using the multi-modal foundation model based upon the second input data. Generating the multi-modal feature space includes extracting first features from the second well log data using the first subnet model. The first features include first filter activations of intermediate layers. Generating the multi-modal feature space also includes extracting second features from the second pre-stack seismic gather tile using the second subnet model. The second features include second filter activations of the intermediate layers. Generating the multi-modal feature space also includes extracting third features from the second post-stack image tile using the third subnet model. The third features include third filter activations of the intermediate layers. Generating the multi-modal feature space also includes fusing the first, second, and third features using a fusion layer of the multi-modal foundation model. The fusion layer is configured to match resolutions of the first, second, and third features; align and/or register the first, second, and third features; and merge the first, second, and third features into the multi-modal feature space. The operations also include performing a downstream task based upon the multi-modal feature space. The downstream task is performed by a multi-task head network that is used to train the multimodal foundation model. The downstream task includes detecting direct hydrocarbon indicators (DHIs) in the second input data, detecting an anomaly in the second input data, determining a reservoir type based upon the second input data, and/or reconstructing an image based upon the second input data.

[0007] It will be appreciated that this summary is intended merely to introduce some aspects of the present methods, systems, and media, which are more fully described and/or claimed below. Accordingly, this summary is not intended to be limiting.

Brief Description of the Drawings

[0008] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the present teachings and together with the description, serve to explain the principles of the present teachings. In the figures:

[0009] Figure 1 illustrates an example of a system that includes various management components to manage various aspects of a geologic environment, according to an embodiment.

[0010] Figure 2 illustrates a model application framework for Al-assisted hydrocarbon anomaly identification, according to an embodiment.

[0011] Figure 3 illustrates an interdisciplinary Al model-building framework, according to an embodiment.

[0012] Figure 4 illustrates input data (e.g., subsurface data), according to an embodiment.

[0013] Figure 5 illustrates a well log subnet, according to an embodiment.

[0014] Figure 6 illustrates a unified multi-modal foundation model (MMFM) network, according to an embodiment.

[0015] Figure 7 illustrates a flowchart of a method for detecting hydrocarbon-related seismic anomalies, according to an embodiment.

[0016] Figure 8 illustrates a schematic view of a computing system for performing at least a portion of the method(s) described herein, according to an embodiment.

Detailed Description

[0017] 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 invention. However, it will be apparent to one of ordinary skill in the art that the invention 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.

[0018] 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 only 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 present disclosure. 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.

[0019] 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 this description 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 combinations 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. Further, 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.

[0020] Attention is now directed to processing procedures, methods, techniques, and workflows that are in accordance with some embodiments. Some operations in the processing procedures, methods, techniques, and workflows disclosed herein may be combined and/or the order of some operations may be changed.

System Overview

[0021] Figure 1 illustrates an example of a system 100 that includes various management components 110 to manage various aspects of a geologic environment 150 (e.g., an environment that includes a sedimentary basin, a reservoir 151, one or more faults 153-1, one or more geobodies 153-2, etc.). For example, the management components 110 may allow for direct or indirect management of sensing, drilling, injecting, extracting, etc., with respect to the geologic environment 150. Tn turn, further information about the geologic environment 150 may become available as feedback 160 (e.g., optionally as input to one or more of the management components 110).

[0022] In the example of Figure 1, the management components 110 include a seismic data component 112, an additional information component 114 (e.g., well/logging data), a processing component 116, a simulation component 120, an attribute component 130, an analysis/visualization component 142 and a workflow component 144. In operation, seismic data and other information provided per the components 112 and 114 may be input to the simulation component 120.

[0023] In an example embodiment, the simulation component 120 may rely on entities 122. Entities 122 may include earth entities or geological objects such as wells, surfaces, bodies, reservoirs, etc. In the system 100, the entities 122 can include virtual representations of actual physical entities that are reconstructed for purposes of simulation. The entities 122 may include entities based on data acquired via sensing, observation, etc. (e.g., the seismic data 112 and other information 114). An entity may be characterized by one or more properties (e.g., a geometrical pillar grid entity of an earth model may be characterized by a porosity property). Such properties may represent one or more measurements (e.g., acquired data), calculations, etc.

[0024] In an example embodiment, the simulation component 120 may operate in conjunction with a software framework such as an object-based framework. In such a framework, entities may include entities based on pre-defined classes to facilitate modeling and simulation. A commercially available example of an object-based framework is the MICROSOFT® .NET® framework (Redmond, Washington), which provides a set of extensible object classes. In the .NET® framework, an object class encapsulates a module of reusable code and associated data structures. Object classes can be used to instantiate object instances for use in by a program, script, etc. For example, borehole classes may define objects for representing boreholes based on well data.

[0025] In the example of Figure 1, the simulation component 120 may process information to conform to one or more attributes specified by the attribute component 130, which may include a library of attributes. Such processing may occur prior to input to the simulation component 120 (e.g., consider the processing component 116). As an example, the simulation component 120 may perform operations on input information based on one or more attributes specified by the attribute component 130. In an example embodiment, the simulation component 120 may construct one or more models of the geologic environment 150, which may be relied on to simulate behavior of the geologic environment 150 (e.g., responsive to one or more acts, whether natural or artificial). In the example of Figure 1, the analysis/visualization component 142 may allow for interaction with a model or model-based results (e.g., simulation results, etc.). As an example, output from the simulation component 120 may be input to one or more other workflows, as indicated by a workflow component 144.

[0026] As an example, the simulation component 120 may include one or more features of a simulator such as the ECLIPSE™ reservoir simulator (SLB, Houston Texas), the INTERSECT™ reservoir simulator (SLB, Houston Texas), etc. As an example, a simulation component, a simulator, etc. may include features to implement one or more meshless techniques (e.g., to solve one or more equations, etc ). As an example, a reservoir or reservoirs may be simulated with respect to one or more enhanced recovery techniques (e.g., consider a thermal process such as SAGD, etc ).

[0027] As an example, the simulation component 120 may include one or more features of a simulator such as SYMMETRY software (SLB, Houston, Texas). More particularly, SYMMETRY may process workflows in a single integrated environment with accurate thermodynamic fluid representation and consistent modeling across multiple disciplines including process, production, and HSE. The simulator integrates steady-state and transient (e.g., dynamic) analyses that can be tailored for each domain. This approach enables users to optimize processes in upstream, midstream, and downstream sectors while maximizing profits and minimizing capital expenditures. It may also help reduce emissions, energy consumption, and waste.

[0028] As an example, the simulation component 120 may include one or more features of a simulator such as PIPESIM (SLB, Houston, Texas). More particularly, PIPESIM is steady-state multiphase flow simulator that incorporates the three areas of flow modeling: multiphase flow, heat transfer and fluid behavior.

[0029] As an example, the simulation component 120 may include one or more features of a simulator such as OLGA™ (SLB, Houston, Texas). More particularly, OLGA™ is a dynamic multiphase flow simulator that models transient flow (e.g., time-dependent behaviors) to maximize production potential. Transient modeling is a component for feasibility studies and field development design. Dynamic simulation is useful in deep water and is used in both offshore and onshore developments to investigate transient behavior in pipelines and wellbores. Transient simulation with the OLGA™ simulator provides an added dimension to steady-state analysis by predicting system dynamics, such as time-varying changes in flow rates, fluid compositions, temperature, solids deposition, and operational changes.

[0030] As an example, the simulation component 120 may include one or more features of a simulator such as SYMMETRY software (SLB, Houston, Texas). More particularly, SYMMETRY may process workflows in a single integrated environment with accurate thermodynamic fluid representation and consistent modeling across multiple disciplines including process, production, and HSE. The simulator integrates steady-state and transient (e.g., dynamic) analyses that can be tailored for each domain. This approach enables users to optimize processes in upstream, midstream, and downstream sectors while maximizing profits and minimizing capital expenditures. It may also help reduce emissions, energy consumption, and waste.

[0031] As an example, the simulation component 120 may include one or more features of a simulator such as PIPESIM (SLB, Houston, Texas). More particularly, PIPESIM® is steady-state multiphase flow simulator that incorporates the three areas of flow modeling: multiphase flow, heat transfer and fluid behavior.

[0032] As an example, the simulation component 120 may include one or more features of a simulator such as OLGA™ (SLB, Houston, Texas). More particularly, OLGA™ is a dynamic multiphase flow simulator that models transient flow (e.g., time-dependent behaviors) to maximize production potential. Transient modeling is a component for feasibility studies and field development design. Dynamic simulation is useful in deep water and is used in both offshore and onshore developments to investigate transient behavior in pipelines and wellbores. Transient simulation with the OLGA™ simulator provides an added dimension to steady-state analysis by predicting system dynamics, such as time-varying changes in flow rates, fluid compositions, temperature, solids deposition, and operational changes.

[0033] In an example embodiment, the management components 110 may include features of a commercially available framework such as the PETREL® seismic to simulation software framework (SLB, Houston, Texas). The PETREL*' framework provides components that allow for optimization of exploration and development operations. The PETREL® framework includes seismic to simulation software components that can output information for use in increasing reservoir performance, for example, by improving asset team productivity. Through use of such a framework, various professionals (e.g., geophysicists, geologists, and reservoir engineers) can develop collaborative workflows and integrate operations to streamline processes. Such a framework may be considered an application and may be considered a data-driven application (e.g., where data is input for purposes of modeling, simulating, etc.).

[0034] In an example embodiment, various aspects of the management components 110 may include add-ons or plug-ins that operate according to specifications of a framework environment. For example, a commercially available framework environment marketed as the OCEAN® framework environment (SLB, Houston, Texas) allows for integration of add-ons (or plug-ins) into a PETREL" framework workflow. The OCEAN " framework environment leverages .NET® tools (Microsoft Corporation, Redmond, Washington) and offers stable, user-friendly interfaces for efficient development. In an example embodiment, various components may be implemented as add-ons (or plug-ins) that conform to and operate according to specifications of a framework environment (e.g., according to application programming interface (API) specifications, etc.).

[0035] Figure 1 also shows an example of a framework 170 that includes a model simulation layer 180 along with a framework services layer 190, a framework core layer 195 and a modules layer 175. The framework 170 may include the commercially available OCEAN® framework where the model simulation layer 180 is the commercially available PETREL® model-centric software package that hosts OCEAN® framework applications. In an example embodiment, the PETREL® software may be considered a data-driven application. The PETREL® software can include a framework for model building and visualization.

[0036] As an example, a framework may include features for implementing one or more mesh generation techniques. For example, a framework may include an input component for receipt of information from interpretation of seismic data, one or more attributes based at least in part on seismic data, log data, image data, etc. Such a framework may include a mesh generation component that processes input information, optionally in conjunction with other information, to generate a mesh.

[0037] In the example of Figure 1, the model simulation layer 180 may provide domain objects 182, act as a data source 184, provide for rendering 186 and provide for various user interfaces 188. Rendering 186 may provide a graphical environment in which applications can display their data while the user interfaces 188 may provide a common look and feel for application user interface components. [0038] As an example, the domain objects 182 can include entity objects, property objects and optionally other objects. Entity objects may be used to geometrically represent wells, surfaces, bodies, reservoirs, etc., while property objects may be used to provide property values as well as data versions and display parameters. For example, an entity object may represent a well where a property object provides log information as well as version information and display information (e.g., to display the well as part of a model).

[0039] In the example of Figure 1, data may be stored in one or more data sources (or data stores, generally physical data storage devices), which may be at the same or different physical sites and accessible via one or more networks. The model simulation layer 180 may be configured to model projects. As such, a particular project may be stored where stored project information may include inputs, models, results and cases. Thus, upon completion of a modeling session, a user may store a project. At a later time, the project can be accessed and restored using the model simulation layer 180, which can recreate instances of the relevant domain objects.

[0040] In the example of Figure 1, the geologic environment 150 may include layers (e.g., stratification) that include a reservoir 151 and one or more other features such as the fault 153-1, the geobody 153-2, etc. As an example, the geologic environment 150 may be outfitted with any of a variety of sensors, detectors, actuators, etc. For example, equipment 152 may include communication circuitry to receive and to transmit information with respect to one or more networks 155. Such information may include information associated with downhole equipment 154, which may be equipment to acquire information, to assist with resource recovery, etc. Other equipment 156 may be located remote from a well site and include sensing, detecting, emitting or other circuitry. Such equipment may include storage and communication circuitry to store and to communicate data, instructions, etc. As an example, one or more satellites may be provided for purposes of communications, data acquisition, etc. For example, Figure 1 shows a satellite in communication with the network 155 that may be configured for communications, noting that the satellite may additionally or instead include circuitry for imagery (e.g., spatial, spectral, temporal, radiometric, etc.).

[0041] Figure 1 also shows the geologic environment 150 as optionally including equipment 157 and 158 associated with a well that includes a substantially horizontal portion that may intersect with one or more fractures 159. For example, consider a well in a shale formation that may include natural fractures, artificial fractures (e.g., hydraulic fractures) or a combination of natural and artificial fractures. As an example, a well may be drilled for a reservoir that is laterally extensive. In such an example, lateral variations in properties, stresses, etc. may exist where an assessment of such variations may assist with planning, operations, etc. to develop a laterally extensive reservoir (e.g., via fracturing, injecting, extracting, etc.). As an example, the equipment 157 and/or 158 may include components, a system, systems, etc. for fracturing, seismic sensing, analysis of seismic data, assessment of one or more fractures, etc.

[0042] As mentioned, the system 100 may be used to perform one or more workflows. A workflow may be a process that includes a number of worksteps. A workstep may operate on data, for example, to create new data, to update existing data, etc. As an example, a may operate on one or more inputs and create one or more results, for example, based on one or more algorithms. As an example, a system may include a workflow editor for creation, editing, executing, etc. of a workflow. In such an example, the workflow editor may provide for selection of one or more predefined worksteps, one or more customized worksteps, etc. As an example, a workflow may be a workflow implementable in the PETREL® software, for example, that operates on seismic data, seismic attribute(s), etc. As an example, a workflow may be a process implementable in the OCEAN® framework. As an example, a workflow may include one or more worksteps that access a module such as a plug-in (e.g., external executable code, etc.).

AI-Assisted Hydrocarbon Anomaly Identification Using Seismic and Log Foundation Model [0043] Figure 2 illustrates a model application framework for Al-assisted hydrocarbon anomaly identification, according to an embodiment. The method described herein utilizes generalized artificial intelligence (Al) for hydrocarbon-related seismic anomaly detection. The scope of this development may be applied to a defined global play type such as Class II deep water channel sands. The output may be an optimized (e.g., deep learning) and/or generalized (e.g., foundation model) pipeline that automatically identifies and rates seismic anomalies to accelerate and improve exploration processes and workstreams (Figure 2).

[0044] The method creates an “optimized” deep learning model that is trained for the specific task of anomaly detection and/or classification. The model may focus on identifying predetermined (e.g., best) geophysical features using appropriate training datasets. The method may create a “foundation model” that may broadly accept a variety of geophysical inputs simultaneously or individually to fundamentally learn the underlying nature and dependencies of the input data; and/or learn multiple interpretation tasks. Various machine-learning formulations and training paradigms may be employed to build such a foundation model (e.g. self-learning, multi-task, multi-objective, etc.). A trained foundation model may further be re-purposed to accomplish one or more interpretation tasks. To re-purpose such a foundation model, fine-tuning and few-shot learning may be employed. Such a foundation model may directly, or via retraining and re-purposing, identify anomalies, and also be the basis of other geophysical tasks like fault detection, horizon picking, salt body segmentation, noise removal, data interpolation, and property inversion. To accomplish specific tasks, the user may provide the model with labels for re-training or examples of the intended task. Subsets of input data types may be used to build a multitude of specialized foundation models.

Interdisciplinary Al Model-Building Framework

[0045] Figure 3 illustrates an interdisciplinary Al model-building framework, according to an embodiment. The process of building and training seismic anomaly identification models is interdisciplinary and involves close collaboration with interpreters, geophysicists, data scientists, and Al-platform architects to build and validate the model output in the context of exploration processes (Figure 3). The method may analyze and/or output interpretation training labels, geophysical attribute and anomaly inference volumes, and their respective ML-pipelines. A user may elect to develop a graphical user interface (GUI) or plugin for additional scope.

Input Data

[0046] Figure 4 illustrates input data (e.g., subsurface data), according to an embodiment. The input data may be or include formation tops 410, well logs 420, 430, 440 from one or more wellbores, a post-stack seismic trace 450, a pre-stack seismic gather 460, a post-stack seismic tile 470, or a combination thereof. The formation tops 410 indicate geological markers. Curves from the well log tracks 420, 430, 440 may correspond to a position of the single post-stack seismic trace 450 (e.g., the dashed range in the post-stack trace track). The zero-offset trace of the prestack gather tile 460 (e.g., shown by the arrow) corresponds to the position of the well logs 420, 430, 440 and the post-stack seismic trace 450. The center line of the post-stack seismic image tile 470 corresponds to the well logs 420, 430, 440 and the position of the zero-offset pre-stack trace 460. Well Log Subnet

[0047] Figure 5 illustrates a well log subnet 500, according to an embodiment. The subnet 500 may process one or more modes of the input data (e.g., from Figure 4). The subnet 500 may include one or more preprocessing layers 510, pooling layers 520, intermediate layers 530, and feature extraction layers 540. Multi-task heads 550 may be used in the training of the subnet 500 to elicit feature discovery. The subnet 500 may be or include a pre-trained foundation model of sufficient complexity to support self-learning and/or multi-tasking, which can be fine-tuned on new data. Feature extraction layers from the subnet 500 may be passed to fusion layers of a multimodal foundation model, as described below.

Multi-Modal Foundation Model

[0048] Figure 6 illustrates a unified multi-modal foundation model (MMFM) network 600, according to an embodiment. Multi-modal data including, for example, pre-stack seismic gathers 460, post-stack image tiles 470, and well logs 420-440 may be preprocessed through respective subnets 500A-500C. The subnets 500A-500C may be pre-trained foundation models. Weights between the subnets 500A-500C may be partially shared during training (as shown by the curved arrows). Feature extraction layers from the subnets 500A-500C may be passed to fusion layers 620 of the multi-modal foundation model 600. The fusion layers 620 may: 1) match resolution of the extracted features 610A-610C, 2) align and/or register the extracted features 610A-610C, and/or 3) merge the extracted features 610A-610C into a common subspace of the input streams from the different subnets. The MMFM 600 may be trained in a self-supervised and multi-tasking scheme including seismic anomaly detection.

Exemplary Method

[0049] Figure 7 illustrates a flowchart of a method 700 for detecting hydrocarbon-related seismic anomalies, according to an embodiment. An illustrative order of the method 700 is provided below; however, one or more portions of the method 700 may be performed in a different order, simultaneously, repeated, or omitted. At least a portion of the method 700 may be performed by a computing system (described below). [0050] In an embodiment, the following may be performed in a standalone fashion: (a) build a seismic foundation model; (b) build a log foundation model; and/or (c) build a multimodal model. Each of these can be used for various interpretation purposes, and also for anomaly detection. For instance, while examples of seismic foundation model downstream interpretation applications (e g., faults, horizons, noise, etc.) are provided, this may not suggest a specific focus on seismic foundation models. Some examples of log foundation model downstream interpretation tasks can be log correction, zonation, formation evaluation, marker detection, log prediction, outlier detection, or a combination thereof.

[0051] The method 700 may include receiving first input data, as at 705. The first input data may be or include first subsurface data. The first subsurface data may be or include first formation tops data 410, first well log data 420-440 from a plurality of first well logs, first post-stack seismic trace data 450, a first pre-stack seismic gather tile 460, a first post-stack seismic image tile 470, or a combination thereof. As mentioned above, curves in the first well logs 420-440 may correspond to a position in the first post-stack trace data 450. A zero-offset trace of the first pre-stack seismic gather tile 460 may correspond to the position, and a vertical centerline of the first post-stack image tile 470 may correspond to the position.

[0052] The method 700 may also include training one or more subnet models 500A-500C based upon the first input data to produce a plurality of trained subnet models, as at 710. In an example, a first of the subnet models may be trained based upon the first well log data 420-440 from plurality of first well logs, a second of the subnet models may be trained based upon the first pre-stack seismic gather tile 460, and a third of the subnet models may be trained based upon the first poststack image tile 470.

[0053] Each of the subnet models 500A-500C may be trained to pre-process the first input data to produce pre-processed data. Each of the subnet models 500A-500C may also be trained to pool the pre-processed data to produce pooled data. Each of the subnet models 500A-500C may also be trained to extract features from the pooled data (e.g., after the pooled data is additionally processed). The extracted features may be obtained by the subnet models 500A-500C that are trained based on the pooled data using self-supervised and/or multi-task approaches. The selfsupervised and/or multi-task approaches may be used to train the subnet models 500A-500C in a way for the trained subnet models 500A-500C to extract the features, which contain sufficient information to solve for one or more downstream applications as a foundation model for a specific data type.

[0054] The method 700 may also include building, updating, and/or training a multi-modal foundation model 600 using the trained subnet models 500A-500C, as at 715. The multi-modal foundation model 600 may be built or updated based upon the extracted features from the trained subnet models 500A-500C. The trained subnet models 500A-500C may operate in parallel within the multi-modal foundation model 600. Weights of neurons of the trained subnet models 500A- 500C may be shared between the trained subnet models 500A-500C.

[0055] The method 700 may also include receiving second input data, as at 720. The second input data may be or include second subsurface data. More particularly, the second subsurface data may be or include second formation tops data, second well log data from a plurality of second well logs, second post-stack trace data, a second pre-stack gather tile, a second post-stack image tile, or a combination thereof.

[0056] The method 700 may also include generating a multi-modal feature space using the multimodal foundation model based upon the second input data, as at 725. Generating the multi-modal feature space may include (1) extracting first features 610A from the second well log data using the first subnet model 500 A, (2) extracting second features 610B from the second pre-stack seismic gather tile using the second subnet model 500B, and extracting third features 610C from the second post-stack image tile using the third subnet model 500C. The first, second, and third features 610A-610C may be different from one another and include filter activations of the intermediate layers.

[0057] Generating the multi-modal feature space may also include fusing the first, second, and third features 610A-610C using a fusion layer 620 of the multi-modal foundation model 600. The fusion layer 620 may be configured to (1) match resolutions of the first, second, and third features 610A-610C, align and/or register the first, second, and third features 610A-610C, and merge the first, second, and third features 610A-610C into a common multi-modal feature space 630.

[0058] The method 700 may also include performing a downstream task 645A-645D based upon the multi-modal feature space, as at 730. The downstream task 645A-645D may be performed by a multi-task head network 640 that is used to train the multi-modal foundation model 600. In an example, the downstream task 645A-645D may be or include detecting direct hydrocarbon indicators (DHIs) in the second input data, detecting an anomaly in the second input data, determining a reservoir type based upon the second input data, and/or reconstructing an image based upon the second input data.

[0059] The method 700 may also include updating the multi-modal foundation model 600 based upon the downstream task 645A-645D, as at 735. The multi-modal foundation model may be updated by modifying weights of the trained subnet models 500A-500C, the extracted features 610A-610C, the fusion layer 620, and/or the multi-modal feature layer/space 630 of the multimodal foundation model 600. In another embodiment, the weights of the multi-modal foundation model 600 may be locked (i.e., not modified). Instead, weights of the multi-task head network 640 may be modified.

[0060] The method 700 may also include displaying the multi-modal feature layer/space 630 and/or an output of the downstream task, as at 740.

[0061] The method 700 may also include performing a wellsite action in response to the multimodal feature layer/space 630 and/or the downstream task 645A-645D, as at 745. The wellsite action may be or include generating and/or transmitting a signal (e.g., using a computing system) that recommends, instructs, or causes a physical action to occur at a wellsite. The action may also or instead include performing the physical action at the wellsite. The physical action may include selecting where to drill a wellbore, drilling the wellbore, varying a weight and/or torque on a drill bit that is drilling the wellbore, varying a drilling trajectory of the wellbore, varying a concentration and/or flow rate of a fluid pumped into the wellbore, or the like.

[0062] In one embodiment, if the network is optimized using self-learning, the residual from the trained model (or re-trained model) may be subject to additional analysis to detect and classify anomalies. The residual is the difference between the network input and output. Additional transformations may be applied to the residual to assist in the detection and classification of hydrocarbons, noise removal, and other interpretation tasks. Additional processing of the residual by the same network may be employed, possibly in combination with the input and output.

[0063] For both log and seismic foundation models, as well as multi-modal foundation models, user-provided examples of tasks may be employed to explicitly influence the network’s inner representational spaces, as well as the network output. These tasks include various above-noted interpretation tasks.

[0064] Additional outputs may be used when training the foundational model to help with the model generalization including, but not limited to, pseudo properties using varying parameters when appropriate (e.g., acoustic impedance, reflection coefficient, POR, SW, VSHALE, VOIL, GAS detector, SALT detector, caliper (logs)).

[0065] The foundational model may be or include trained deep learning methods that extract information from a variety of large datasets. Those models may be based on neural networks, and may include multiple types of layers such as convolutional layers, activation layers, residual layers, attention layers, dropout layers, and/or pooling and up sampling layers, and other layers. The deep learning methods may be or include convolution-based or transformer-based architectures.

[0066] The number of trainable layers may impact the memory usage (e.g., model storage costs), computational cost, and data for multiple downstream applications and the risk of overfitting in case of small, labelled datasets. As a result, models may be used that can achieve high performance with small trainable parameters. Additional techniques may be used for helping the pre-trained method with the generalization regarding the regional differences or learning new tasks without full fine-tuning the entire model. One or more of those techniques may be influenced and adapted from the following NLP, such as parameter-efficient fine-tuning (PEFT), low rank adaptation, weight decomposed low-rank adaptation, etc., as well as possibly including additional domain specific methods and additional information.

[0067] The models may be built for identifying “common anomalies” as associated with washouts, misreading’s in top/bottom, coal/kerogen/bitumen, gas, highly radioactive layers, or a combination thereof. The models may use metadata (e.g., well locations, field name, etc.) to select a set of wells/logs in the pre-built database for fine tuning the pre-trained model or a previously fine-tuned model for a first estimate of basic desirable properties if not available in the dataset. If a pre-built model or database is not available, the method may obtain a first estimate, possibly from similar wells in different regions or using a general global model. If those curves are available in the dataset, the method may instead calculate the error (e.g., MAE/MSE, Accuracy) and indicate that the first model provides this amount of error, while organizing the data and retraining the model with the available curves to consider the information from the current database as a refined answer.

[0068] The distribution of the ML reconstructed logs versus the logged data may be used to indicate that normalization may be applied to the data. The foundational model may be used to construct acoustic impedance and reflection coefficients in areas where density or p waves are not available - for helping with the seismic to well tie. User input may be used to define the self- supervised tasks to be included in the conditioned model training (e.g., focus on anomaly detection, filling missing parts, creating reflection coefficient or other additional curves, propagating some property to the reservoir, etc.), and/or proposing workflows around those prompts.

Multi-Modal Large, Foundat ion Model Methodology

[0069] Large models (LM)/foundation models (FM) have amply demonstrated excellent generalization properties and offer benefits for rapid development of downstream ML applications using fewer labels in a computationally efficient manner. The size and depth of such LM/FM also opens an opportunity to simultaneously process multiple input data types (i.e., multimodal LM/FM). Multi-modal LM/FMs may permit coherent fusion of different sources of complementary information and such an approach may permit better detection and classification of anomalies.

[0070] The method 700 employs large multi-modal foundation modeling (multi-modal LM/FM) with additional downstream processing as a solution methodology for detecting HC anomaly identifications. More particularly, a deep network, with sufficient sequential processing depth, may provide a means to disentangle the underlying complexity, and provide a larger effective receptive field for wider contextual understanding. Deep networks may learn the regional characteristics of a survey and provide a semantically structured latent space. The ‘foundation’ of FM implies that the model has sufficient capacity, which, along with the deep processing, and large volumes of data will endow a network with generalization capabilities.

[0071] Network generalization may play a role in the detection of unseen HC anomalies anomaly identification. For example, in the high dimensional latent space, an unseen anomaly segment may be projected close to and in-between known anomalies (i.e., interpolation) or may be projected peripherally to known anomalies (i.e., extrapolation). Mathematically, if HC anomalies are viewed as a multi-attributed object, and the unseen anomaly shares a few attributes with known anomalies, the interpolation/extrapolation argument holds. Hence, the model may be able to generalize. With such broad generalization, transfer learning to other related tasks should also be possible.

[0072] To incorporate different sources of complementary information, the method 700 adopts a multi-modal ML formulation to allow the network to fuse the various data sources (e.g., well logs) simultaneously and coherently. Such complementary information may provide for better characterization of HC anomalies. While derived data (e.g., RGT, faults, stratigraphy, frequency domain calculations, etc.) may not be strictly complementary to the base data, the method 700 may incorporate these as a means to constrain and induce the network towards specific intermediate or final objectives. Moreover, from a practical standpoint, inclusion of such derived data as input may permit the method 700 to reduce the size of the model and provide faster experimentation cycles.

[0073] The data processed through the LM/FM may be repurposed for direct anomaly detection, or intermediate targets which together can be used for anomaly detection. As noted above, due to better generalization properties of such models, the prospects are promising in that the model may work in more diverse geological settings.

[0074] A methodology is in place to build seismic foundation models, and based on the details of the data provided, convolution-based or transformer-based architectures may be adopted. Preliminary runs may be conducted to establish whether the depth domain (e.g., post-stack) or time domain (e.g., AVO, pre-stack, etc.). Other options include a pre-stack gather or post-stack image domain. To endow the multi-modal foundation model with functionalities noted earlier, due consideration may be given to the training paradigm. For example, the method 700 may include training the network using self-supervised learning and/or multi-tasking.

[0075] On completion of training, a pretrained LM/FM model may exist for further processing and use. For instance, to detect common HC anomalies, the model may be fine-tuned using a few anomaly examples. One focus may be the discovery of unseen HC anomalies. The pretrained FM model may provide a well -structured latent space capturing high-level semantics of the core data. Due to the deep sequence of nonlinear transformations, data projection onto such LM/FM latent space may provide semantically meaningful clusters which are also well separated. HC anomaly occurrences may appear as outliers in such latent space. Various means to analyze this highdimensional latent space for HC anomalies may be explored (e.g., clustering, etc.).

[0076] The exploration may not be purely focused on the LM/FM latent space analysis. Given that the unseen HC anomalies occur infrequently, it is also likely that a network may not learn such instances and may effectively fdter out such HC anomaly occurrences or their partial characteristics. To address this issue, the method 700 may analyze the residuals obtained from the differences between the model input and output. Any unseen HC anomaly features that are fdtered out by the model may appear in the residual. Such residuals may contain signal leakage, random noise, processing artifacts, etc. Additional means may be devised to analyze such residuals for HC anomalies, as mentioned above.

Exemplary Computing System

[0077] In some embodiments, the methods of the present disclosure may be executed by a computing system. Figure 8 illustrates an example of such a computing system 800, in accordance with some embodiments. The computing system 800 may include a computer or computer system 801A, which may be an individual computer system 801A or an arrangement of distributed computer systems. The computer system 801 A includes one or more analysis modules 802 that are configured to perform various tasks according to some embodiments, such as one or more methods disclosed herein. To perform these various tasks, the analysis module 802 executes independently, or in coordination with, one or more processors 804, which is (or are) connected to one or more storage media 806. The processor(s) 804 is (or are) also connected to a network interface 807 to allow the computer system 801 A to communicate over a data network 809 with one or more additional computer systems and/or computing systems, such as 80 IB, 801C, and/or 80 ID (note that computer systems 80 IB, 801C and/or 80 ID may or may not share the same architecture as computer system 801 A, and may be located in different physical locations, e.g., computer systems 801 A and 801B may be located in a processing facility, while in communication with one or more computer systems such as 801 C and/or 80 ID that are located in one or more data centers, and/or located in varying countries on different continents).

[0078] A processor may include a microprocessor, microcontroller, processor module or subsystem, programmable integrated circuit, programmable gate array, or another control or computing device.

[0079] The storage media 806 may be implemented as one or more computer-readable or machine-readable storage media. Note that while in the example embodiment of Figure 8 storage media 806 is depicted as within computer system 801A, in some embodiments, storage media 806 may be distributed within and/or across multiple internal and/or external enclosures of computing system 801 A and/or additional computing systems. Storage media 806 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), BLURAY® disks, or other types of optical storage, or other types of storage devices. Note that the instructions discussed above may be provided on one computer-readable or machine-readable storage medium, or may be provided on multiple computer-readable or machine-readable storage media distributed in a large system having possibly plural nodes. Such computer-readable or machine-readable storage medium or media is (are) considered to be part of an article (or article of manufacture). An article or article of manufacture may refer to any manufactured single component or multiple components. The storage medium or media may be located either in the machine running the machine-readable instructions, or located at a remote site from which machine-readable instructions may be downloaded over a network for execution.

[0080] In some embodiments, computing system 800 contains one or more hydrocarbon anomaly identification module(s) 808. In the example of computing system 800, computer system 801 A includes the hydrocarbon anomaly identification module 808. In some embodiments, a single hydrocarbon anomaly identification module may be used to perform some aspects of one or more embodiments of the methods disclosed herein. In other embodiments, a plurality of hydrocarbon anomaly identification modules may be used to perform some aspects of methods herein.

[0081] It should be appreciated that computing system 800 is merely one example of a computing system, and that computing system 800 may have more or fewer components than shown, may combine additional components not depicted in the example embodiment of Figure 8, and/or computing system 800 may have a different configuration or arrangement of the components depicted in Figure 8. The various components shown in Figure 8 may 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.

[0082] Further, the steps in the processing methods described herein may be implemented by running one or more functional modules in information processing apparatus such as general purpose processors or application specific chips, such as ASICs, FPGAs, PLDs, or other appropriate devices. These modules, combinations of these modules, and/or their combination with general hardware are included within the scope of the present disclosure. [0083] Computational interpretations, models, and/or other interpretation aids may be refined in an iterative fashion; this concept is applicable to the methods discussed herein. This may include use of feedback loops executed on an algorithmic basis, such as at a computing device (e.g., computing system 800, Figure 8), and/or through manual control by a user who may make determinations regarding whether a given step, action, template, model, or set of curves has become sufficiently accurate for the evaluation of the subsurface three-dimensional geologic formation under consideration.

[0084] 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 limiting to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings. Moreover, the order in which the elements of the methods described herein are illustrated and described may be re-arranged, and/or two or more elements may occur simultaneously. The embodiments were chosen and described in order to best explain the principles of the disclosure and its practical applications, to thereby enable others skilled in the art to best utilize the disclosed embodiments and various embodiments with various modifications as are suited to the particular use contemplated.

Claims

CLAIMS What is claimed is:

1. A method for detecting hydrocarbon-related seismic anomalies, the method comprising: receiving first input data; training a plurality of subnet models based upon the first input data to produce a plurality of trained subnet models; building a multi-modal foundation model using the trained subnet models; receiving second input data; and generating a multi-modal feature space using the multi-modal foundation model based upon the second input data.

2. The method of Claim 1 , wherein the first input data comprises subsurface data, and wherein the subsurface data comprises formation tops data, well log data from a plurality of well logs, poststack seismic trace data, a pre-stack seismic gather tile, and/or a post-stack seismic image tile.

3. The method of Claim 2, wherein curves in the well logs correspond to a position in the post-stack trace data, wherein a zero-offset trace of the pre-stack seismic gather tile corresponds to the position, and wherein a vertical centerline of the post-stack image tile corresponds to the position.

4. The method of Claim 2, wherein a first of the subnet models is trained based upon the first well log data from plurality of well logs, wherein a second of the subnet models is trained based upon the pre-stack seismic gather tile, wherein a third of the subnet models is trained based upon the post-stack image tile.

5. The method of Claim 1, wherein training the subnet models comprises extracting features from the first input data, and wherein the multi-modal foundation model is trained based upon the extracted features.

6. The method of Claim 1 , wherein the trained subnet models operate in parallel within the multi-modal foundation model.

7. The method of Claim 1 , wherein weights of neurons of the trained subnet models are shared between the trained subnet models.

8. The method of Claim 1, further comprising performing a downstream task based upon the multi-modal feature space.

9. The method of Claim 1, further comprising displaying the multi-modal feature space.

10. The method of Claim 1, further comprising performing a wellsite action in response to the multi-modal feature space.

11. A computing system, comprising: one or more processors; and a memory system comprising one or more non-transitory computer-readable media storing instructions that, when executed by at least one of the one or more processors, cause the computing system to perform operations, the operations comprising: receiving first input data, wherein the first input data comprises first subsurface data, wherein the first subsurface data comprises first formation tops data, first well log data from a plurality of first well logs, first post-stack seismic trace data, a first pre-stack seismic gather tile, and/or a first post-stack seismic image tile; training a plurality of subnet models based upon the first input data to produce a plurality of trained subnet models, wherein a first of the subnet models is trained based upon the first well log data from plurality of first well logs, wherein a second of the subnet models is trained based upon the first pre-stack seismic gather tile, wherein a third of the subnet models is trained based upon the first post-stack image tile; building or updating a multi-modal foundation model using the trained subnet models, wherein the multi-modal foundation model is built or updated based upon features extracted from the trained subnet models, and wherein the trained subnet models operate in parallel within the multi-modal foundation model; receiving second input data, wherein the second input data comprises second subsurface data, wherein the second subsurface data comprises second formation tops data, second well log data from a plurality of second well logs, second post-stack trace data, a second pre-stack gather tile, and/or a second post-stack image tile; and generating a multi-modal feature space using the multi-modal foundation model based upon the second input data.

12. The computing system of Claim 11, wherein each of the subnet models is trained to: pre-process the first input data to produce pre-processed data; pool the pre-processed data to produce pooled data; and extract features from the pooled data, wherein the extracted features are obtained by the subnet models that are trained based on the pooled data using self-supervised and/or multi-task approaches, and wherein the self-supervised and/or multi-task approaches are used to train the subnet models in a way for the trained subnet models to extract the features.

13. The computing system of Claim 11, wherein generating the multi-modal feature space comprises: extracting first features from the second well log data using the first subnet model; extracting second features from the second pre-stack seismic gather tile using the second subnet model; extracting third features from the second post-stack image tile using the third subnet model, wherein the first, second, and third features are different and comprise third filter activations of intermediate layers; and fusing the first, second, and third features using a fusion layer of the multi-modal foundation model.

14. The computing system of Claim 13, wherein the fusion layer is configured to: match resolutions of the first, second, and third features; align and/or register the first, second, and third features; and merge the first, second, and third features into the multi-modal feature space.

15. The computing system of Claim 11, wherein the operations further comprise performing a downstream task based upon the multi-modal feature space, wherein the downstream task is performed by a multi-task head network that is used to train the multi-modal foundation model, wherein the downstream task comprises detecting direct hydrocarbon indicators (DHIs) in the second input data, detecting an anomaly in the second input data, determining a reservoir type based upon the second input data, and/or reconstructing an image based upon the second input data.

16. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors of a computing system, cause the computing system to perform operations, the operations comprising: receiving first input data, wherein the first input data comprises first subsurface data, wherein the first subsurface data comprises first formation tops data, first well log data from a plurality of first well logs, first post-stack seismic trace data, a first pre-stack seismic gather tile, and a first post-stack seismic image tile, wherein curves in the first well logs correspond to a position in the first post-stack trace data, wherein a zero-offset trace of the first pre-stack seismic gather tile corresponds to the position, and wherein a vertical centerline of the first post-stack image tile corresponds to the position; training a plurality of subnet models based upon the first input data to produce a plurality of trained subnet models, wherein a first of the subnet models is trained based upon the first well log data from plurality of first well logs, wherein a second of the subnet models is trained based upon the first pre-stack seismic gather tile, wherein a third of the subnet models is trained based upon the first post-stack image tile, and wherein each of the subnet models is trained to: pre-process the first input data to produce pre-processed data; pool the pre-processed data to produce pooled data; and extract features from the pooled data after the pooled data is additionally processed, wherein the extracted features are obtained by the subnet models that are trained based on the pooled data using self-supervised and/or multi-task approaches, wherein the selfsupervised and/or multi-task approaches are used to train the subnet models in a way for the trained subnet models to extract the features, which contain sufficient information to solve for multiple downstream applications as a foundation model for a specific data type; building or updating a multi-modal foundation model using the trained subnet models, wherein the multi-modal foundation model is built or updated based upon the extracted features from the trained subnet models, wherein the trained subnet models operate in parallel within the multi-modal foundation model, and wherein weights of neurons of the trained subnet models are shared between the trained subnet models; receiving second input data, wherein the second input data comprises second subsurface data, wherein the second subsurface data comprises second formation tops data, second well log data from a plurality of second well logs, second post-stack trace data, a second pre-stack gather tile, and a second post-stack image tile; generating a multi-modal feature space using the multi-modal foundation model based upon the second input data, wherein generating the multi-modal feature space comprises: extracting first features from the second well log data using the first subnet model, wherein the first features comprise first filter activations of intermediate layers; extracting second features from the second pre-stack seismic gather tile using the second subnet model, wherein the second features comprise second filter activations of the intermediate layers; extracting third features from the second post-stack image tile using the third subnet model, wherein the third features comprise third filter activations of the intermediate layers; and fusing the first, second, and third features using a fusion layer of the multi-modal foundation model, wherein the fusion layer is configured to: match resolutions of the first, second, and third features; align and/or register the first, second, and third features; and merge the first, second, and third features into the multi-modal feature space; and performing a downstream task based upon the multi-modal feature space, wherein the downstream task is performed by a multi-task head network that is used to train the multi-modal foundation model, wherein the downstream task comprises detecting direct hydrocarbon indicators (DHIs) in the second input data, detecting an anomaly in the second input data, determining a reservoir type based upon the second input data, and/or reconstructing an image based upon the second input data.

17. The non-transitory computer-readable medium of Claim 16, wherein the operations further comprise updating the multi-modal foundation model based upon the downstream task, wherein the multi-modal foundation model is updated by modifying weights of the trained subnet models and/or the fusion layer of the multi-modal foundation model.

18. The non-transitory computer-readable medium of Claim 16, wherein the operations further comprise updating the multi-task head network based upon the downstream task, wherein the multi-task head network is updated by modifying weights of the multi-task head network, and wherein weights of the multi-modal foundation model are not modified.

19. The non-transitory computer-readable medium of Claim 16, wherein the operations further comprise displaying an output of the downstream task,

20. The non-transitory computer-readable medium of Claim 16, wherein the operations further comprise performing a wellsite action in response to the downstream task.