Target-oriented inversion (TOI) is an approach aimed at enhancing the ability of full-waveform inversion (FWI) to achieve a high-resolution delineation of a reservoir. FWI has demonstrated its potential to address the challenge of imaging complex structures on a considerable number of field data applications. Nevertheless, it is still impractical costwise to implement FWI with the full band of seismic data because, in this case, we need to discretize the whole subsurface model space with a fine grid to handle the high frequencies and satisfy the interpretation of, for example, reservoir-scale features. Redatuming techniques enable us to obtain a virtual data set at the target level from the original data acquisition that is most commonly deployed on the earth’s surface. The virtual data set can help us apply a high-resolution FWI to the target region, which often occupies a small area of the entire model space. To analyze such a redatuming process, we need to estimate an overburden model that can accurately describe the kinematics and dynamics of the wave propagation. Fortunately, our virtual data retrieval can rely on the overburden estimation with relatively low resolution, because the high-frequency multiple scattering has a limited effect on the deep part and on the corresponding virtual data. Therefore, we start with macro-overburden models that contain reasonably accurate kinematics, and we then apply FWI on the overburden with only low-frequency data. The resulting model is used to implement a least-squares waveform redatuming using the full band. The Marmousi model and Chevron 2014 benchmark data set are used to demonstrate the effectiveness of our strategy, which results in the high-resolution inversion of the target areas. Our TOI workflow leads to an obvious boost in efficiency and reduces the memory requirement because the finer grid needed for the high frequencies is only adopted for the redatuming and the TOI.

中文翻译：

---

面向目标的波形重新计算和高分辨率反演：覆盖层的作用

面向目标的反演（TOI）是一种旨在增强全波形反演（FWI）的能力，以实现油藏高分辨率描述的方法。FWI展示了其解决大量现场数据应用中复杂结构成像挑战的潜力。尽管如此，在整个地震数据带中实施FWI仍然不切实际，在这种情况下，因为在这种情况下，我们需要用精细的网格离散整个地下模型空间，以处理高频并满足例如以下解释：水库规模特征。借助重新引用技术，我们可以从最常部署在地球表面的原始数据获取中获得目标级别的虚拟数据集。虚拟数据集可以帮助我们将高分辨率FWI应用于目标区域，该区域通常占据整个模型空间的一小部分。要分析这种重新规划过程，我们需要估算一个可以准确描述波传播的运动学和动力学的覆盖模型。幸运的是，我们的虚拟数据检索可以依靠分辨率相对较低的覆盖层估计，因为高频多次散射对深部和相应的虚拟数据影响有限。因此，我们从包含相当准确的运动学的宏覆盖模型开始，然后我们仅使用低频数据对覆盖层应用FWI。所得的模型用于使用全频段实施最小二乘波形重新计算。Marmousi模型和Chevron 2014基准数据集用于证明我们策略的有效性，从而实现了目标区域的高分辨率反演。我们的TOI工作流程显着提高了效率并减少了内存需求，因为高频所需要的更精细的网格仅用于重新标定和TOI。