Least-squares reverse time migration (LSRTM) is currently one of the most advanced migration imaging techniques in the field of geophysics. It uses least-squares inversion to fit the observed data, resulting in high-resolution imaging results with more accurate amplitudes and better illumination compensation than conventional reverse time migration (RTM). However, noise in the observed data and the Born approximation forward operator can result in high-wavenumber artifacts in the final imaging results. Moreover, iteratively solving LSRTM leads to one or two orders of computational cost higher than conventional RTM, making it challenging to apply extensively in industrial applications. Simultaneous source acquisition technology can reduce the computational cost of LSRTM by reducing the number of wavefield simulations. However, this technique also can cause high-wavenumber crosstalk artifacts in the migration results. To effectively remove the high-wavenumber artifacts caused by these issues, simultaneous source and deep learning are combined to speed up LSRTM as well as suppress high-wavenumber artifacts. A deep residual neural network (DR-Unet) is trained with synthetic samples, which are generated by adding field noise to synthesized noise-free migration images. Then, the trained DR-Unet is applied on the gradient of LSRTM to remove high-wavenumber artifacts in each iteration. Compared to directly applying DR-Unet denoising to LSRTM results, embedding DR-Unet denoising into the inversion process can better preserve weak reflectors and improve denoising effects. Finally, the proposed LSRTM method is tested on two synthetic data sets and a land data set. The tests demonstrate that the proposed method can effectively remove high-wavenumber artifacts, improve imaging results, and accelerate convergence speed.
Finite-difference methods are the most widely used methods for seismic wavefield simulation. However, numerical dispersion is the main issue hindering accurate simulation. In the case where the finite-difference scheme is known, the time dispersion can be predicted mathematically and, thus, can be eliminated. However, when only pre-compiled software is available for wavefield simulation, which is common in practical applications, the software-used algorithm becomes a black box (unknown). Therefore, it is challenging to obtain the mathematical expression of the time dispersion, resulting in difficulty in eliminating the time dispersion. To solve this problem, we propose to use deep learning methods to eliminate time dispersion. We design a semi-supervised framework based on convolutional and recurrent neural networks for eliminating time dispersion caused by seismic wave modeling. The framework of our proposed neural network includes two main modules: Inverse Model and Forward Model, both of which have learnable parameters. The Inverse Model is used for eliminating time dispersion while the Forward Model is used for regularizing the training. Particularly, this framework includes two steps: Firstly, using the compiled modeling software to generate two data sets with large and small time steps. Secondly, we train these two modules for transformation between large time-step data (with time dispersion) and small time-step data (without time dispersion) by labeled and unlabeled data sets. In this work, the labeled data set is a paired data set with large time-step data and their corresponding small time-step data; the unlabeled data set is the large time-step data that need time-dispersion elimination. We use the unlabeled data set to guide the network. In this learning framework, re-training is required whenever the modeling algorithms, time interval, or frequency band is changed. Hence, we propose a transfer learning training method to extend from the trained model to another model, which reduces the computational cost caused by re-training. This minor drawback is offset overwhelmingly by the modeling efficiency gain with large time steps in large-scale production. Tests on two models confirm the effectiveness of the proposed method.
28 February 2026 in Journal of Geophysical Research: Machine Learning and Computation
Solving the wave equation is fundamental for many geophysical applications. However, numerical solutions of the Helmholtz equation face significant computational and memory challenges. Therefore, we introduce a physics‐informed convolutional neural operator (CNO) (PICNO) to solve the Helmholtz equation efficiently. PICNO takes both the background wavefield corresponding to a homogeneous medium and the velocity model as input function space, generating the scattered wavefield as the output function space. Our workflow integrates partial differential equation constraints directly into the training process, enabling the neural operator to not only fit the available data but also capture the underlying physics governing wave phenomena. PICNO allows for high‐resolution reasonably accurate predictions even with limited training samples, and it demonstrates significant improvements over a purely data‐driven CNO, particularly in predicting high‐frequency wavefields. These features and improvements are important for waveform inversion down the road.
Solving the wave equation is fundamental for geophysical applications. However, numerical solutions of the Helmholtz equation face significant computational and memory challenges. Therefore, we introduce a physics-informed convolutional neural operator (PICNO) to solve the Helmholtz equation efficiently. The PICNO takes both the background wavefield corresponding to a homogeneous medium and the velocity model as input function space, generating the scattered wavefield as the output function space. Our workflow integrates PDE constraints directly into the training process, enabling the neural operator to not only fit the available data but also capture the underlying physics governing wave phenomena. PICNO allows for high-resolution reasonably accurate predictions even with limited training samples, and it demonstrates significant improvements over a purely data-driven convolutional neural operator (CNO), particularly in predicting high-frequency wavefields. These features and improvements are important for waveform inversion down the road.
Extracting subsurface velocity information from seismic data is mainly an undetermined problem that requires injecting a priori information to constrain the inversion process. Machine learning has offered a platform to do so through the training process as we formulate our training dataset to inject as much a prior knowledge as possible in the trained ML model. Here, we use a neural-operator-based framework for high-resolution seismic velocity model building, which integrates Neural Operators with time-lag reverse time migration imaging. Unlike conventional full waveform inversion (FWI) methods that rely on iterative forward and adjoint-state computations, our approach learns a direct mapping from initial velocity models and extended seismic images to high-resolution velocity estimates through supervised learning. The network architecture enables mesh-independent generalization by learning mappings in infinite-dimensional function spaces, while synthetic velocity models and time-lag reverse time migration (RTM) images provide complementary high-frequency information critical for recovering mid- and high-wavenumber velocity components. Synthetic experiments demonstrate that the proposed method accurately reconstructs fine-scale structures and complex geologies, including out-of-distribution features such as salt bodies. Applications to real seismic field data acquired offshore Australia further validate the method’s robustness and resolution capability. The predicted models show enhanced structural details and improved consistency with well-log data, outperforming traditional multi-scale FWI in both accuracy and computational efficiency. The entire prediction process is completed within seconds, making the proposed approach highly suitable for rapid and scalable velocity model building in practical exploration scenarios.
Extracting subsurface velocity information from seismic data is mainly an undetermined problem that requires injecting a priori information to constrain the inversion process. Machine learning has offered a platform to do so through the training process, as we formulate our training dataset to inject as much prior knowledge as possible in the trained ML model. Here, we use a neural-operator-based framework for high-resolution seismic velocity model building, which integrates Neural Operators with time-lag reverse time migration imaging. Unlike conventional full waveform inversion (FWI) methods that rely on iterative forward and adjoint-state computations, our approach learns a direct mapping from initial velocity models and extended seismic images to high-resolution velocity estimates through supervised learning. The network architecture enables mesh-independent generalization by learning mappings in infinite-dimensional function spaces, while synthetic velocity models and time-lag reverse time migration (RTM) images provide complementary high-frequency information critical for recovering mid- and high-wavenumber velocity components. Synthetic experiments demonstrate that the proposed method accurately reconstructs fine-scale structures and complex geologies, including out-of-distribution features such as salt bodies. Applications to real seismic field data acquired offshore Australia further validate the method's robustness and resolution capability. The predicted models show enhanced structural details and improved consistency with well-log data, outperforming traditional multi-scale FWI in both accuracy and computational efficiency. The entire prediction process is completed within seconds, making the proposed approach highly suitable for rapid and scalable velocity model building in practical exploration scenarios.