Reverse time migration (RTM) for attenuating media should take amplitude compensation and phase correction into consideration. However, attenuation compensation during seismic propagation suffers from numerical instability because of the boosted high-frequency ambient noise. We have developed a novel adaptive stabilization method for $Q$-compensated RTM ($Q$-RTM), which exhibits superior properties of time variance and $Q$ dependence over conventional low-pass filtering-based method. We derive the stabilization operator by first analytically deriving $k$-space Green’s functions for a constant-$Q$ wave equation with decoupled fractional Laplacians and its compensated equation. The time propagator of Green’s function for the viscoacoustic wave equation decreases exponentially, whereas that of the compensated equation is exponentially divergent at a high wavenumber, and it is not stable after the wave is extrapolated for a long time. Therefore, the Green’s functions theoretically explain how the numerical instability existing in $Q$-RTM arises and shed light on how to overcome this problem pertinently. The stabilization factor required in the proposed method can be explicitly identified by the specified gain limit according to an empirical formula. The $Q$-RTM results for noise-free data using low-pass filtering and adaptive stabilization are compared over a simple five-layer model and the BP gas chimney model to verify the superiority of the proposed approach in terms of fidelity and stability. The $Q$-RTM result for noisy data from the BP gas chimney model further demonstrates that our method enjoys a better antinoise performance and helps significantly to enhance the resolution of seismic images.

We have designed a periodically varying code that can avoid the problem of the local coherency and make the interference distribute uniformly in a given range; hence, it was better at suppressing incoherent interference (blending noise) and preserving coherent useful signals compared with a random dithering code. We have also devised a new form of the iterative method to remove interference generated from the simultaneous source acquisition. In each iteration, we have estimated the interference using the blending operator following the proposed formula and then subtracted the interference from the pseudodeblended data. To further eliminate the incoherent interference and constrain the inversion, the data were then transformed to an auxiliary sparse domain for applying a thresholding operator. During the iterations, the threshold was decreased from the largest value to zero following an exponential function. The exponentially decreasing threshold aimed to gradually pass the deblended data to a more acceptable model subspace. Two numerically blended synthetic data sets and one numerically blended practical field data set from an ocean bottom cable were used to demonstrate the usefulness of our proposed method and the better performance of the periodically varying code over the traditional random dithering code.

Recently, a decoupled fractional Laplacian viscoacoustic wave equation has been developed based on the constant-$Q$ model to describe wave propagation in heterogeneous media. We have developed two efficient modeling schemes to solve the decoupled fractional Laplacian viscoacoustic wave equation. Both schemes can cope with spatial variable-order fractional Laplacians conveniently, and thus are applicable for modeling viscoacoustic wave propagation in heterogeneous media. Both schemes are based on fast Fourier transform, and have a spectral accuracy in space. The first scheme solves a modified wave equation with constant-order fractional Laplacians instead of spatial variable-order fractional Laplacians. Due to separate discretization of space and time, the first scheme has only first-order accuracy in time. Differently, the second scheme is based on an analytical wave propagator, and has a higher accuracy in time. To increase computational efficiency of the second modeling scheme, we have adopted the low-rank decomposition in heterogeneous media. We also evaluated the feasibility of applying an empirical approximation to approximate the fractional Laplacian that controls amplitude loss during wave propagation. When the empirical approximation is applied, our two modeling schemes become more efficient. With the help of numerical examples, we have verified the accuracy of our two modeling schemes with and without applying the empirical approximation, for a wide range of seismic quality factor ($Q$). We also compared computational efficiency of our two modeling schemes using numerical tests.

Simultaneous-source acquisition helps greatly boost an economic saving, while it brings an unprecedented challenge of removing the crosstalk interference in the recorded seismic data. In this paper, we propose a novel iterative method to separate the simultaneous source data based on a coherency-pass shaping operator. The coherency-pass filter is used to constrain the model, that is, the unblended data to be estimated, in the shaping regularization framework. In the simultaneous source survey, the incoherent interference from adjacent shots greatly increases the rank of the frequency domain Hankel matrix that is formed from the blended record. Thus, the method based on rank reduction is capable of separating the blended record to some extent. However, the shortcoming is that it may cause residual noise when there is strong blending interference. We propose to cascade the rank reduction and thresholding operators to deal with this issue. In the initial iterations, we adopt a small rank to severely separate the blended interference and a large thresholding value as strong constraints to remove the residual noise in the time domain. In the later iterations, since more and more events have been recovered, we weaken the constraint by increasing the rank and shrinking the threshold to recover weak events and to guarantee the convergence. In this way, the combined rank reduction and thresholding strategy acts as a coherency-pass filter, which only passes the coherent high-amplitude component after rank reduction instead of passing both signal and noise in traditional rank reduction based approaches. Two synthetic examples are tested to demonstrate the performance of the proposed method. In addition, the application on two field data sets (common receiver gathers and stacked profiles) further validate the effectiveness of the proposed method.

The lack of low-frequency signals in seismic data makes the full-waveform inversion (FWI) procedure easily fall into local minima leading to unreliable results. To reconstruct the missing low-frequency signals more accurately and effectively, we have developed a data-driven low-frequency recovery method based on deep learning from high-frequency signals. In our method, we develop the idea of using a basic data patch of seismic data to build a local data-driven mapping in low-frequency recovery. Energy balancing and data patches are used to prepare high- and low-frequency data for training a convolutional neural network (CNN) to establish the relationship between the high- and low-frequency data pairs. The trained CNN then can be used to predict low-frequency data from high-frequency data. Our CNN was trained on the Marmousi model and tested on the overthrust model, as well as field data. The synthetic experimental results reveal that the predicted low-frequency data match the true low-frequency data very well in the time and frequency domains, and the field results show the successfully extended low-frequency spectra. Furthermore, two FWI tests using the predicted data demonstrate that our approach can reliably recover the low-frequency data.

Frequency-dependent amplitude absorption and phase velocity dispersion are typically linked by the causality-imposed Kramers-Kronig (K-K) relations, which inevitably degrade the quality of seismic data. Seismic attenuation compensation is an important processing approach for enhancing signal resolution and fidelity, which can be performed on either prestack or poststack data so as to mitigate amplitude absorption and phase dispersion effects resulting from intrinsic anelasticity of subsurface media. Inversion-based compensation with L₁ norm constraint, enlightened by the sparsity of the reflectivity series, enjoys better stability over traditional inverse Q filtering. However, constrained L₁ minimization serving as the convex relaxation of the literal L₀ sparsity count may not give the sparsest solution when the kernel matrix is severely ill-conditioned. Recently, nonconvex metric for compressed sensing has attracted considerable research interest. In this paper, we propose a nearly unbiased approximation of the vector sparsity, denoted as L_{1 − 2} minimization, for exact and stable seismic attenuation compensation. Nonconvex penalty function of L_{1 − 2} norm can be decomposed into two convex subproblems via difference of convex algorithm (DCA), each subproblem can be solved efficiently by alternating direction method of multipliers (ADMM). The superior performance of the proposed compensation scheme based on L_{1 − 2} metric over conventional L₁ penalty is further demonstrated by both synthetic and field examples.