CGG发布了全新油藏描述软件，具有基于云平台的创新机器学习能力。

编译 | 惊蛰

CGG公司宣布，其地质科学部门旗下的Geosoftware，推出了新一代基于云平台的油藏描述解决方案，包括Jason 10.0，HampsonRussell 10.4与PowerLog 10.0。这些软件具备先进的机器学习能力和强大的跨产品集成能力，可提高勘探开发项目的效益，提供更易理解的油藏描述。

这三个新版本软件都可以在Microsoft Azure的云环境中无缝运行，并且很快可用于其他的云平台。利用这些新软件，地球物理学家可以实现计算密集型的作业流程，并可运行大型项目，以前所未有的速度处理数千口井数据。

机器学习的优势在于，能够持续提高应用能力，应对复杂的地质挑战。目前，HampsonRussell Emerge以深度前馈神经网络的形式提供深度学习，可更好地预测储层物性。PowerLog中的开放式python生态系统，支持在作业流程中使用机器学习与深度学习，以提高自动化，并实现更准确的相预测。

新版本的特点在于集成能力的进步，例如“一次加载，处处可用”的能力，简化了跨软件应用的工作流程，并能够满足用户提出的其他需求。HampsonRussell 10.4增加了先进的地震数据处理功能，提高了地震数据质量，可获得更准确的反演结果。AVO建模现提供了更多样的工具，来研究叠前数据的地震响应。Jason 10.0可以很容易地利用岩石物性测井资料，设计或审查相分类，并能立即获得弹性反演域的效果。它还改进了时深转换与深度反演的速度标定。PowerLog 10.0的改进使用户能够有效地解释井组，用户还可基于PowerLog生态系统，应用机器学习来解决岩石物理的挑战。

GeoSoftware高级副总裁Kamal al-Yahya表示：“我们一如既往地关注客户需求，确保我们的软件及时升级，真正可以带来有助于提高效率的创新。随着地球物理学、地质学以及岩石物理学之间的联系日益紧密，产品集成成为跨学科工作流程的关键。这些新升级软件也是第一批获益于我们数字化发展的产品，集成了机器学习与云端应用程序，可为客户提供高效的工作体验。”

JASON

Jason Workbench拥有先进的油藏描述模块与工作流程，可帮助用户及时了解缓解风险的方法。利用Jason技术，地质科学家与工程师可做出优化油井产能、油田开发以及储层管理相关的重要决策。Jason 10提供了油藏描述的新功能，包括更好的可用性与性能、与其他GeoSoftware软件的无缝连接、新的相驱动技术，能够满足关键工作流程中更多地质约束的需求。Jason Workbench 10.0致力于提供先进、直观的工作流程，提供卓越的用户体验。此版本扩展并增强了Jason的用户体验，提供了现代油藏描述所需的关键工作流程。

1. 核心的反演工具拥有全新的用户界面与更友好的用户体验，使用户可更快的工作；
2. 新的相驱动反演技术满足了关键工作流程中更多地质限制的需求；
3. 可用性的改进，简化了Jason Workbench中的加载与查看数据的流程；
4. 可在少至三个方位扇区上完成各向异性的反演，以提供基本的层向各向异性特性。

HampsonRussell

HampsonRussell是一套完整的地球物理解释工具套件，主要用于油藏描述。HampsonRussell以其易用性而闻名，它可为各种水平的地球物理学家提供先进的地球物理技术。目前最新的10.4.1版本中拥有油藏描述的新功能。HampsonRussell 10.4.1版的新功能可持续改进油藏描述与地震解释技术，旨在使每位地球科学家都能进行复杂的地球物理分析与工作流程。

Emerge中的机器学习能力可以更好地进行油藏描述。几十年来，机器学习一直是Emerge的组成部分。新版本中增强了Emerge深度学习的功能，可以更精确的进行油藏描述。先进的地震数据处理。这是一个新模块，可以提高地震数据质量，并可增加地震反演与AVO的可信度。

AVO快速建模功能拥有众多工具，可用于生成一维、二维、三维的合成模型，并可分析潜在的地震响应。它的RockSI组件可分析岩石物理学的相关信息。多节点处理(MNP)：分布在多台Linux机器上的多节点处理，可以从根本上缩短堆栈正反演、AVO以及数据处理的运行时间。它可在云端使用，从而实现了硬件资源的可扩展性。高级Python脚本：用于地震处理的高级Python脚本支持添加Python插件，使客户能够自由地将高级R＆D算法直接整合到HampsonRussell中。

POWERLOG

岩石力学、岩石物理学、相分析与统计矿物学的基准。多井测井资料的协同分析可帮助作业者轻松做出更好的决策。PowerLog是一款多用户、多井、多解释器的软件，适用于灵活的岩石物理工作流程。PowerLog可在任何地方使用，使您的观点能够快速体现在重要的钻井决策上。

PowerLog 10提供了新功能，包括：机器学习工作流程、新的截面对比模块、地层柱状图编辑器。PowerLog使用户能够简单高效地构建出自用的测井资料处理模块，并开发出专有的、复杂的、高级的解释模型。PowerLog 10版本中增加了易于使用的工具，并通过新的机器学习工作流程提高了工作效率。

1. 新的工作流程利用了机器学习与深度学习，用户可以开发出自动化编辑测井资料的工作流程，以及其他提高生产力与理解力的流程；
2. PowerLog已成功测试，可在Azure云平台上运行；
3. 新的截面对比模块，可实现井间的拖拽投影曲线、建立截面、创建层位、地质信息可视化等功能，能够帮助用户交互选择相关层位的顶部；
4. 新的地层柱状图编辑器有助于确定地质时间与建立沉积环境，使用该编辑器，用户可以将地层添加到测井图、示意图、截面图中，以方便交流与理解。

CGG announces that GeoSoftware, part of its Geoscience Division, has launched a new generation of cloud-ready reservoir characterization solutions. Jason 10.0, HampsonRussell 10.4 and PowerLog 10.0 also feature advanced machine learning capabilities and greater cross-product integration, improving E&P project performance and providing a better understanding of reservoir properties.

The three new releases already run seamlessly on Microsoft Azure’s Cloud Environment and will soon be available on other major Cloud platforms. Geoscientists can now implement compute-intensive workflows and run very large projects, to process thousgands of wells, faster than ever before.

The benefits of machine learning continue to drive new capabilities to address complex geological challenges. HampsonRussell Emerge now delivers deep learning in the form of Deep Feed Forward Neural Networks for better prediction of reservoir properties. An open Python ecosystem in PowerLog enables the routine use of machine and deep learning in workflows to increase automation and achieve more accurate facies predictions.

The new releases feature integration advancements, such as the ability to “load once, use everywhere” to streamline cross-product workflows, as well as other user-driven improvements. HampsonRussell 10.4 updates Advanced Seismic Conditioning to improve seismic data quality for better inversion outcomes and AVO Modeling now offers a wider range of tools for investigating the seismic response of pre-stack data. Jason 10.0 makes it easy to design or vet facies classifications from petrophysical logs and immediately see the effects in the elastic inversion domain. It also has improved velocity calibration for time-to-depth conversion and depth inversion. PowerLog 10.0 advancements enable users to efficiently interpret groups of wells and apply machine learning to solve petrophysical challenges using the PowerLog Ecosystem.

Kamal al-Yahya, senior V.P., GeoSoftware & Smart Data Solutions, said: “As always, we listen to our clients to ensure our new software releases bring innovations that really help them increase efficiency. Greater product integration is key for inter-disciplinary workflows as links between geophysics, geology and petrophysics grow ever stronger. This new set of releases is also the first to deliver real benefits from our digitalization roadmap, offering a high-quality experience with the integration of Machine Learning and Cloud-ready applications.”

JASON?Advanced Seismic Reservoir Characterization

The advanced seismic reservoir characterization software products and workflows within the Jason Workbench yield timely answers to mitigate risk. Jason technology supports critical decisions for geoscientists and engineers to optimize well productivity, field development, and reservoir management.

New capabilities for reservoir characterization now available in Jason 10 including better usability and performance, seamless connectivity with other GeoSoftware products and new facies-driven technology that addresses the need for more geological constraints in critical workflows.

Jason Workbench 10.0 demonstrates our commitment to delivering leading-edge, intuitive workflows offering a superior user experience. This version release extends and enhances the Jason User experience, offering the critical workflows required for modern reservoir characterization.

• New user interface and more user friendly experience for the core deterministic inversion product, RockTrace, lets users work faster
• New Facies-Driven Inversion technology addresses the need for more geological constraints in critical workflows
• Usability improvements streamlines loading and viewing data in Jason Workbench
• Anisotropic inversion can now be completed on as few as three azimuthal sectors to deliver basic, layer-oriented anisotropic properties
• Jason Workbench has been successfully tested and is now Azure Cloud-ready

HAMPSONRUSSELL?Geophysical Interpretation Tools

HampsonRussell offers a fully integrated suite of world-class geophysical interpretation tools for reservoir characterization. Known for its ease of use, HampsonRussell makes sophisticated geophysical techniques accessible to all levels of geophysicists. New features for reservoir characterization now available in 10.4.1.

New features in HampsonRussell version 10.4.1 delivers continual improvements in reservoir characterization and seismic interpretation technologies aimed at making sophisticated geophysical analysis and workflows accessible to every geoscientist.

Power of Machine Learning in Emerge for better reservoir characterization：Machine learning has been a component in Emerge for decades. A key enhancement is Emerge Deep Learning allowing for more accurate reservoir characterization.

Advanced Seismic Conditioning：This is a new module to improve seismic data quality and achieve more confidence in inversion and AVO.

AVO Quick Modeling：AVO Quick Modeling has an expanded range of tools to generate 1D/2D/3D synthetic models and analyze potential seismic response. Its RockSI component provides linkage to rock physics.

Multi Node Processing (MNP)：Multi Node Processing distributed across several Linux machines can radically reduce the run-time of pre and post-stack inversion, AVO and gather conditioning processes. It can be used in the cloud which enables scalability of hardware resources.

Advanced Python Scripting：Advanced Python Scripting for seismic processes enables adding Python plug-ins, giving you a freedom to incorporate your advanced R&D algorithms directly into HampsonRussell.

POWERLOG?Industry-Leading Petrophysical Interpretation

The benchmark for petrophysics, rock physics, facies analysis and statistical mineralogy. Collaborative multi-well log analysis made easy for better drilling decisions. PowerLog is multi-user, multi-well, multi-interpreter software tuned for flexible petrophysical workflows. PowerLog can be used anywhere to get your point across for quick collaboration on important drilling decisions.

New features including machine learning workflows, a new correlation cross-section module and stratigraphic column editor are now available in PowerLog 10.

PowerLog gives users the ability to build their own advanced log processing modules simply and efficiently as well as develop proprietary, sophisticated and advanced interpretation models. New features in the latest PowerLog 10 release adds easy to use tools and allows for increased productivity with new machine learning workflows.

• New workflows utilize machine learning and deep learning so users can develop workflows for automated log editing along and other activities that increase productivity and understanding
• PowerLog has been successfully tested and is ready to work in the Azure Cloud platform
• New Correlation Cross-Section Module aids users in interactively picking tops correlated by dragging ghosted curves from well to well, building cross-sections, creating zones, and visualizing the geology
• Newly built Stratigraphic Column Editor helps define geologic time and establish depositional environments. With this editor, users can add the stratigraphic columns to log plots, collage displays, and cross sections to improve communication and understanding