[论文解读] Seeing through the CO2 plume: joint inversion-segmentation of the Sleipner 4D Seismic Dataset

Juan Romero, Nick Luiken+1 more|arXiv (Cornell University)|Mar 21, 2023

Seismic Imaging and Inversion TechniquesEarth and Planetary Sciences被引用 3

一句话总结

本文提出了一种用于Sleipner CO₂储存场地4D地震反演的联合反演-分割（JIS）框架，结合全变差和分割先验，以提高分辨率并抑制非重复性噪声。该方法生成了高分辨率的声阻抗模型及时间-剖面变化的体积分类，从而更清晰地检测CO₂引起的地下动态变化，并为储层模拟提供更优的输入。

ABSTRACT

4D seismic inversion is the leading method to quantitatively monitor fluid flow dynamics in the subsurface, with applications ranging from enhanced oil recovery to subsurface CO2 storage. The process of inverting seismic data for reservoir properties is, however, a notoriously ill-posed inverse problem due to the band-limited and noisy nature of seismic data. This comes with additional challenges for 4D applications, given inaccuracies in the repeatability of the time-lapse acquisition surveys. Consequently, adding prior information to the inversion process in the form of properly crafted regularization terms is essential to obtain geologically meaningful subsurface models. Motivated by recent advances in the field of convex optimization, we propose a joint inversion-segmentation algorithm for 4D seismic inversion, which integrates Total-Variation and segmentation priors as a way to counteract the missing frequencies and noise present in 4D seismic data. The proposed inversion framework is applied to a pair of surveys from the open Sleipner 4D Seismic Dataset. Our method presents three main advantages over state-of-the-art least-squares inversion methods: 1. it produces high-resolution baseline and monitor acoustic models, 2. by leveraging similarities between multiple data, it mitigates the non-repeatable noise and better highlights the real time-lapse changes, and 3. it provides a volumetric classification of the acoustic impedance 4D difference model (time-lapse changes) based on user-defined classes. Such advantages may enable more robust stratigraphic and quantitative 4D seismic interpretation and provide more accurate inputs for dynamic reservoir simulations. Alongside our novel inversion method, in this work, we introduce a streamlined data pre-processing sequence for the 4D Sleipner post-stack seismic dataset, which includes time-shift estimation and well-to-seismic tie.

研究动机与目标

• 为解决由于带限、含噪数据以及非重复性采集噪声导致的4D地震反演病态性问题。
• 通过叠加4D地震数据提高基线和监测阶段声阻抗模型的分辨率和地质保真度。
• 通过生成阻抗变化的分段体积分类模型，实现对时移变化的定量解释。
• 展示JIS算法在GPU加速优化下对大规模3D地震数据集的可扩展性和高效性。

提出的方法

• 该方法采用联合反演方法，针对基线和监测阶段的地震数据，使用包含子波w₁和w₂的联合建模算子G̃。
• 应用全变差正则化以促进分段平滑解，减少反演模型中噪声的放大。
• 通过分割步骤，根据时移行为将4D差分模型中的体素分类为用户定义的类别（例如，阻抗变化范围为-50%至+50%）。
• 利用邻近算法结合交替最小化求解反演问题，借助凸优化确保稳定性与收敛性。
• 在反演前通过高斯-牛顿非线性反演方法集成时移估计，以对齐基线与监测调查数据。
• 该算法在开源、GPU加速的框架中实现，以高效处理大规模3D地震体数据。

Figure 1: Sleipner seismic dataset used in this work. a) Baseline seismic survey (1994), b) monitor seismic survey (2001), and c) the difference between them. The well shown in a) does not intersect the CO 2 plume.

实验结果

研究问题

• RQ1 与标准Tikhonov正则化相比，结合分割先验的联合反演是否能提升4D地震反演的分辨率和抗噪能力？
• RQ2 分割步骤在Sleipner储层中对CO₂注入引起的地下变化的分类与可视化效果如何？
• RQ3 JIS框架在保留真实地质变化的同时，能在多大程度上减少时移差分模型中的非重复性噪声？
• RQ4 尽管存在振幅调谐和低对比度效应，分段4D模型能否以可接受的精度估算注入的CO₂质量？
• RQ5 当在GPU架构上实现时，JIS算法在大规模3D地震数据集上的可扩展性如何？

主要发现

• 与Tikhonov正则化反演相比，JIS方法生成了更高分辨率的声阻抗模型，地质边界更加清晰。
• JIS生成的4D差分模型显示出更清晰的CO₂注入引起的时移变化，且非重复性噪声显著低于标准最小二乘反演。
• 分段模型识别出阻抗降低区域（例如-50%）和增加区域，表明砂岩和泥岩层因孔隙压力变化导致的压实体和膨胀。
• 基于分段地质体估算的CO₂质量为380万吨，相对于2001年实际注入的420万吨低估了9.5%，可能由于CO₂在含水层中溶解所致。
• GPU加速将反演时间从CPU的约29小时缩短至GPU的约22分钟，证明了该算法在大规模4D监测中的高可扩展性和实际可行性。

Figure 2: Well-analysis and well-to-seismic tie. From left to right, sonic and density logs (edited to eliminate spikes), TD-TWT relationship obtained by correlating regional seismic surfaces in time with well-picks in depth, impedance log (black) and its smoothed version used to create the backgrou