In the realm of energy exploration, understanding the subsurface structures is crucial for locating resources and ensuring safe extraction. A team of researchers from the University of Southern California, Kamal Aghazade, Toktam Zand, and Ali Gholami, has been working on improving a technique called Full-Waveform Inversion (FWI) to enhance our ability to image these subsurface structures with greater accuracy. Their recent work, published in the journal Geophysics, focuses on automating a key aspect of FWI, making it more robust and efficient.
Full-Waveform Inversion is a powerful seismic imaging technique used to estimate the physical properties of subsurface structures by minimizing the differences between observed and modeled seismic data. However, FWI is a complex and ill-posed problem, meaning there can be multiple solutions that fit the data equally well. To address this, researchers use extended-source approaches like the augmented Lagrangian (AL) method to improve the stability and robustness of the solution. A critical component of this method is the penalty parameter, which balances the trade-off between fitting the data and satisfying the wave-equation constraint. The challenge lies in selecting the appropriate penalty parameter, as traditional methods either require accurate noise estimates or involve computationally expensive trial-and-error processes.
To overcome these limitations, the researchers integrated two data-driven strategies—the Residual Whiteness Principle (RWP) and a stable variant of Generalized Cross-Validation (RGCV)—within a multiplier-oriented AL framework. They adopted a dual-space AL formulation, which keeps the background wave-equation operator fixed and requires only a single LU factorization per frequency. This design allows for dynamic adjustment of the penalty parameter during iterations at a negligible computational cost, making the algorithm scalable for large-scale applications.
Numerical experiments conducted by the team demonstrated that, when combined with the dual-space formulation, RWP provides strong noise robustness. This means the method can reliably handle both white and colored noise, making it a practical solution for large-scale seismic inversion tasks. The automated nature of the penalty parameter selection also reduces the need for manual tuning, saving time and computational resources.
For the energy industry, this research offers a more efficient and reliable way to perform seismic imaging. By automating the penalty parameter selection process, energy companies can achieve higher-resolution subsurface images with less effort and cost. This can lead to more accurate resource estimation and better-informed decision-making in exploration and extraction activities. The practical applications of this research extend to various sectors within the energy industry, including oil and gas exploration, geothermal energy assessment, and even carbon sequestration projects, where understanding subsurface structures is paramount.
In summary, the work of Aghazade, Zand, and Gholami represents a significant advancement in the field of seismic imaging. By integrating RWP and RGCV into the FWI process, they have developed a more robust and efficient method for subsurface imaging, which can greatly benefit the energy sector. The research was published in the journal Geophysics, a leading publication in the field of geophysical research.
This article is based on research available at arXiv.