Shot Gathers Research Articles

Estimation of CO 2 saturation maps from synthetic seismic data using a deep-learning method with a multi-scale approach

In the context of Carbon Capture and Storage (CCS), monitoring the behaviour of CO 2 within subsurface reservoirs is pivotal for efficient and safe storage operations. This study proposes a new approach to enhance the accuracy of predicting CO 2 saturation maps directly from shot gathers, using a combination of deep learning (DL) and feature extraction methods. We employ a U-Net model, specifically tailored to solve regression tasks, and two-dimensional continuous wavelet transform (2D-CWT) to analyse shot gathers at different scales. We introduce a novel approach using multi-channel input data for the DL model, combining shot gathers and CWT images. The DL model was trained and tested using both single channel and multichannel input. The single channel datasets included shot gathers and CWT images at the first scale, while the multi-channel datasets integrated shot gathers with either two or four scales of CWT. We conducted a sensitivity analysis on the number of training epochs to compare the performance of the model with multichannel and single channel input. Additionally, a Transfer Learning approach was implemented to improve performance in noisy conditions by leveraging knowledge from a pre-trained model on noise-free images. Monte Carlo dropout was used to predict pixel-scale variability and provided a prediction variability with a standard deviation ranging from 0.019 to 0.194, enhancing the robustness of CO 2 saturation estimates. Our results suggest that combining feature extraction using 2D-CWT with the U-Net architecture improves the prediction performance of CO 2 saturation in a synthetic CCS model. Additionally, using multi-channel input instead of single channel is more efficient as it requires less training and prediction time to achieve comparable results. This confirms the effectiveness of 2D-CWT as a pre-processing technique, contributing valuable insights and advances in the field of data-driven geophysical inversion problems.

Static Correction of Passive-Source Seismic Reflection Exploration on Undulating Surfaces

Elevation static correction is an important step in seismic data processing used to eliminate distortions in events caused by surface fluctuations. There is a well-established processing sequence for active-source seismic data, with static correction applied first, followed by subsequent processing. For passive-source data, static correction can be carried out using two methods. The first method is similar to the static correction of active-source seismic data and involves performing static correction on virtual shot gathers. The second method involves directly applying static correction to ambient noise and then using the corrected ambient noise to retrieve the virtual shot gathers. In this study, we derived and compared these two methods to perform static correction on passive-source data. We compared the above two methods through the forward modeling of passive-source data and field data from the Daqiao Gold Mine. The results showed that applying static correction to ambient noise and virtual shot gathers can correct passive-source data from an undulating surface to a flat surface. The equivalence of static correction on virtual shot gathers, and ambient noise offers more choices for conducting static correction of passive-source seismic reflection exploration on undulating surfaces.

Observations from the seafloor: ultra-low-frequency ambient ocean-bottom nodal seismology at the Amendment field

SUMMARY Large-scale ocean-bottom node (OBN) arrays of 1000s of multicomponent instruments deployed over 1000s of square kilometres have been used successfully for active-source seismic exploration activities including full waveform inversion (FWI) at exploration frequencies above about 2.0 Hz. The analysis of concurrently recorded lower-frequency ambient wavefield data, though, is only just beginning. A key long-term objective of such ambient wavefield analyses is to exploit the sensitivity of sub-2.0 Hz energy to build long-wavelength initial elastic models, thus facilitating FWI applications. However, doing so requires a more detailed understanding of ambient wavefield information recorded on the seafloor, the types, frequency structure and effective source distribution of recorded surface-wave modes, the near-seafloor elastic model structure, and the sensitivity of recorded wave modes to subsurface model structure. To this end, we present a wavefield analysis of low- and ultra-low-frequency ambient data (defined as <1.0 and <0.1 Hz, respectively) acquired on 2712 OBN stations in the Amendment Phase 1 survey covering 2750 km2 of the Gulf of Mexico. After applying ambient data conditioning prior to cross-correlation and seismic cross-coherence interferometry workflows, we demonstrate that the resulting virtual shot gather (VSG) volumes contain evidence for surface-wave and guided P-wave mode propagation between the 0.01 and 1.0 Hz that remains coherent to distances of at least 80 km. Evidence for surface-wave scattering from near-surface salt-body structure between 0.35 and 0.85 Hz is also present in a wide spatial distribution of VSG data. Finally, the interferometric VSG volumes clearly show waveform repetition at 20 s intervals in sub-0.3 Hz surface-wave arrivals, a periodicity consistent with the mean active-source shot interval. This suggests that the dominant contribution of surface-wave energy acquired in this VSG frequency band is likely predominantly related to air-gun excitation rather than by naturally occurring energy sources. Overall, these observations may have important consequences for the early stages of initial model building for elastic FWI analysis.

Study on the influencing factors of combined processing of active and passive surface-wave data on dispersion imaging

Active and passive surface-wave methods have garnered significant attention in the near-surface geophysical community for their non-destructive, non-invasive, low-cost, and accurate advantages in delineating subsurface shear (S)-wave velocity structures. They are increasingly being utilized to address numerous engineering and environmental problems. Surface waves obtained from actively excited sources such as a hammer and a harmonic shaker, however, lack low-frequency components, resulting in limited investigation depth. Conversely, passive surface waves such as microseisms (< 0.4 Hz, associated with natural ocean wave activity) and microtremor (>1 Hz, generated by cultural and wind sources) retrieved from ambient seismic noise typically lack high-frequency components, which is not conductive to characterizing fine near-surface structures. To overcome these frequency limitations, we employ a “mixed-source data” strategy, imposing active shot gathers into ambient noise data, to widen the frequency range of dispersion images and depth of investigative capabilities. We simulate both active and passive surface-wave data based on a two-layer model, noting that their dispersion images suffer from a mode kissing phenomenon at lower frequencies. By analyzing influencing factors such as the amplitude intensity, the signal-to-noise ratio and the excitation locations of active shot gathers, as well as the length of passive surface-wave data, we better understand their impacts on dispersion images from mixed-source surface-wave data. Simulation tests demonstrate that processing mixed-source data can effectively distinguish the mode kissing phenomenon. Moreover, the effectiveness of this strategy in enhancing the quality of dispersion image is verified, especially when surface-wave dispersion images perform poorly in either the low- or high-frequency bands. Additionally, a real-world example further demonstrates that processing mixed-source data offers significant advantages in improving the quality of dispersion images. This way provides a convenient and efficient measurement strategy for delineating shear-wave velocity profiles in finer shallow layers and deeper penetration depths.

A collocated inversion of sources and early arrival waveforms for credible tomograms: Synthetic and field data examples

AbstractWaveform inversion is theoretically a powerful tool to reconstruct subsurface structures, but a usually encountered problem is that accurate sources are very rare, causing the computation to be unstable or divergent. This challenging practical problem, although sometimes ignored and even imperceptible, can easily create discrepancies in calculated shot gathers, which will then lead to wrong residuals that will be smeared back to the gradients, hence jeopardizing the inverted tomograms. For any real dataset, every shot gather corresponds to its unique source even if some gathers can be transformed alike after data processing. To resolve this problem, we propose a collocated inversion of sources and early arrival waveforms with the two submodules executing successively. Not only can this method reconstruct a decent source wavelet that approaches the ground truth, but also it can produce credible background tomograms with optimized sources. Part of the cycle skipping problems can also be mitigated because it avoids the trial and error experiments on various sources. Numerical tests on a synthetic and a land dataset validate the effectiveness of this method. Restrictions on initial sources or starting velocity models will be relaxed, and this method can be extended to any other applications for engineering or exploration purposes.

Urban subsurface exploration improved by denoising of virtual shot gathers from distributed acoustic sensing ambient noise

SUMMARY Ambient noise tomography on the basis of distributed acoustic sensing (DAS) deployed on existing telecommunication networks provides an opportunity to image the urban subsurface at regional scales and high-resolution. This capability has important implications in the assessment of the urban subsurface’s potential for sustainable and safe utilization, such as geothermal development. However, extracting coherent seismic signals from the DAS ambient wavefield in urban environments at low cost remains a challenge. One obstacle is the presence of complex sources of noise in urban environments, which may not be homogeneously distributed. Consequently, long recordings are required for the calculation of high-quality virtual shot gathers, which necessitates significant time and computational cost. In this paper, we present the analysis of 15 d of DAS data recorded on a pre-existing fibre optic cable (dark fibres), running along an 11-km-long major road in urban Berlin (Germany), hosting heavy traffic including vehicles and trains. To retrieve virtual shot gathers, we apply interferometric analysis based on the cross-correlation approach where we exclude low-quality virtual shot gathers to increase the signal-to-noise ratio of the stacked gathers. Moreover, we modify the conventional ambient noise interferometry workflow by incorporating a coherence-based enhancement approach designed for wavefield data recorded with large-N arrays. We then conduct multichannel analysis of surface waves to retrieve 1-D velocity models for two exemplary fibre subsegments, and compare the results of the conventional and modified workflows. The resulting 1-D velocity models correspond well with available lithology information. The modified workflow yields improved dispersion spectra, particularly in the low-frequency band (&amp;lt;1 Hz) of the signal. This leads to an increased investigation depth along with lower uncertainties in the inversion result. Additionally, these improved results were achieved using significantly less data than required using conventional approaches, thus opening the opportunity for shortening required acquisition times and accordingly lowering costs.

Two‐dimensional SH‐wave and acoustic P‐wave full waveform inversion: A Midland Basin case study

AbstractBuilding a reliable S‐wave velocity model remains challenging from P‐wave land seismic data as multi‐parameter elastic full waveform inversion which updates P‐ and S‐wave velocities simultaneously is not yet widely adopted due to its computational cost, highly nonlinear nature and noise contamination from land seismic data. To overcome this challenge, we propose implementing the SH‐wave full waveform inversion to obtain the S‐wave velocity model. The value of this approach has been examined by applying two‐dimensional time‐domain SH‐ and P‐wave acoustic full waveform inversion to a nine‐component three‐dimensional seismic survey acquired in the Midland Basin. The nine‐component seismic survey was acquired by using both vertical and horizontal component vibrators and receivers, which enables us to apply both SH‐ and P‐wave full waveform inversion to the raw and preprocessed shot gathers along a two‐dimensional line. The SH‐ and P‐wave full waveform inversion applied to the raw shot gathers provide more detailed P‐ and S‐wave velocities compared to the velocities from the stacking velocity analysis. Additionally, there are no problematic artefacts in the inverted S‐wave velocity model, which shows the stability of the SH‐wave inversion. Through the SH‐wave full waveform inversion from the preprocessed shot gathers, we reveal the lateral velocity variations in the Grayburg–San Andres interval, which coincides with the depositional environment changes suggested by the previous regional study mainly based on well log and cores. These variations demonstrate that SH‐wave full waveform inversion can identify additional geological features that are not observed in the inverted P‐wave velocity model. The inverted P‐ and SH‐wave velocities are validated by the flattening of the events observed in the common image gathers after Kirchhoff depth migration. The SH‐wave migration image further reveals the irregular geometries of carbonates in the Spraberry formation. Vp/Vs ratios calculated from the independently inverted P‐ and S‐wave velocity models show strong lateral variations in the Wolfcamp interval (key‐producing interval), which could be caused by the organic content variations between the reservoir and non‐reservoir rocks. The inversion results demonstrate that the SH‐wave full waveform inversion can be used to provide an S‐wave velocity model that is comparable to the P‐wave velocity model derived from acoustic full waveform inversion, which is widely used for P‐wave velocity model building from P‐wave land seismic data.

Deep learning‐based extraction of surface wave dispersion curves from seismic shot gathers

AbstractMulti‐channel analysis of surface waves is a seismic method employed to obtain useful information about shear‐wave velocities in the near surface. A fundamental step in this methodology is the extraction of dispersion curves from dispersion spectra, with the latter usually obtained by applying specific processing algorithms onto the recorded shot gathers. Although the extraction process can be automated to some extent, it usually requires extensive quality control, which can be arduous for large datasets. We present a novel approach that leverages deep learning to identify a direct mapping between seismic shot gathers and their associated dispersion curves (both fundamental and first higher order modes), therefore by‐passing the need to compute dispersion spectra. Given a site of interest, a set of 1D compressional and shear velocities and density models are created using prior knowledge of the local geology; pairs of seismic shot gathers and Rayleigh‐wave phase dispersion curves are then numerically modelled and used to train a simplified residual network. The proposed approach is shown to achieve high‐quality predictions of dispersion curves on a synthetic test dataset and is, ultimately, successfully deployed on a field dataset. Various uncertainty quantification and convolutional neural network visualization techniques are also presented to assess the quality of the inference process and better understand the underlying learning process of the network. The predicted dispersion curves are inverted for both the synthetic and field data; in the latter case, the resulting shear‐wave velocity model is plausible and consistent with prior geological knowledge of the area. Finally, a comparison between the manually picked fundamental modes with the predictions from our model allows for a benchmark of the performance of the proposed workflow.

Near offset reconstruction for marine seismic data using a convolutional neural network

AbstractMarine seismic data is often missing near offset information due to separation between the source and receiver cables. To solve this problem, a convolutional neural network is trained on synthetic seismic data to reconstruct the near offset gap. The synthetic data is created using a two‐dimensional finite difference method within a heterogeneous velocity model. These synthetics are generated with a source‐over‐receiver acquisition geometry so that they contain complete near offset data. The convolutional neural network is then trained on input‐target synthetic pairs where the inputs are common midpoint gathers with the near offset section removed, and the targets are the same gathers with the near offset section retained. Following training, the robustness of the method is investigated with regards to common midpoint data sorting, normal moveout correction and changes in the velocity model. It is found that training on common midpoint‐sorted data results in 2.8 times lower error than training on shot gathers, that normal moveout correction of the training data makes no significant difference in error levels, and that the model can reconstruct realistic near offsets on synthetic data generated 10 km away within the heterogeneous velocity model. In field data testing, first a dataset with source‐over‐cable acquisition geometry from the Barents Sea is used to compare the reconstructed wavefields to ground truth values. Although the reconstructed amplitudes require minor scaling to match the true values, predictions on this dataset yield 2.5 times lower near offset reconstruction error compared to a simple Radon transform interpolation method. Furthermore, amplitude versus offset gradient and intercept sections from the Barents Sea dataset are estimated with half the error when including the convolutional neural network‐predicted near offset data, compared to only using the conventionally‐acquirable portion of the data (beyond 112.5 m of offset). In a secondary field data test, a conventional northern North Sea dataset is used to demonstrate how the method may be applied in practice. Here, the convolutional neural network generates more realistic predictions than the Radon method, and the gradient and intercept sections calculated using the convolutional neural network‐predicted traces have higher signal‐to‐noise ratios than the sections calculated using only the original data. The combination of high‐quality synthetic training data and interpolation in the common midpoint domain enables near offset reconstruction at significant depth (1 s of two‐way traveltime or more), which is demonstrated in both synthetic and field examples.

A method for denoising active source seismic data via Fourier transform and spectrum reconstruction

Denoising is a critical step in signal processing. We develop a method for random noise reduction in active source seismic data using spectrum reconstruction. Two methods are developed for modifying the observed data’s amplitude spectrum: one substitutes it with the source wavelet’s amplitude spectrum, whereas the other involves multiplying the source wavelet’s amplitude spectrum with the observed data’s amplitude spectrum. By reconstructing the modified amplitude spectrum while preserving the observed data’s phase spectrum, noise suppression is achieved. Extensive testing with theoretical models, synthetic shot gathers, and field data indicate a notable improvement in the signal-to-noise ratio (S/N) compared with the traditional band-pass filtering method. This method proves particularly effective for enhancing the S/N in the context of active source wide-angle seismic data used in offshore structural studies, eliminating the need for data segmentation based on offset, and thereby improving processing efficiency. Our method relies solely on a single complete cycle of the source wavelet, making it a purely data-driven solution. It has broad applications in processing active source or controlled source data with consistent source wavelets, including but not limited to seismic exploration, acoustic detection, and signal denoising in various ground-penetrating radars used on Mars, the moon, and earth.

OBC shallow water de-multiple based on the principle of Fresnel diffraction

Abstract In shallow-water ocean-bottom cable (OBC) seismic data, the ineffectiveness of conventional surface-related multiple elimination (SRME) methods due to poor seabed records is addressed. This research utilizes the seismic wavefield received by multiple cables from a single shot gather to predict shallow water multiple models for that shot gather. Initially, the seismic data within a finite aperture around a seismic trace in the time domain shot gather is treated as the known seismic wavefield. This seismic wavefield is then extrapolated along the water layer to this seismic trace, following the Fresnel diffraction principle. The extrapolated data becomes the shallow water multiple model for this seismic trace. This process is repeated for each trace in the shot gather to obtain the shallow water multiple model of the entire shot gather. Forward modeling tests have shown that smaller data apertures can effectively avoid the impact of spatial aliasing on multiple model prediction. To address the overlap of primary waves and shallow water multiples in deep seismic data, which have lower dominant frequencies, the multiple model data is used as a known seismic wavefield and extrapolated along the water layer again. This produces second-order and higher-order multiple models. Applying this model to suppress multiple waves can minimize primary waves loss. This entirely data-driven approach necessitates solely water depth information, imposing no additional conditions. Both forward modeling and real seismic data testing validate the efficacy of this method in shallow water.

Time-lapse seismic inversion for CO2 saturation with SeisCO2Net: An application to Frio-II site

Seismic monitoring of geological CO2 storage (GCS) involves highly nonlinear seismic inversion and petrophysical inversion, making it challenging to estimate CO2 volume efficiently and detect possible early CO2 leakages. Deep learning (DL) using convolutional neural networks (CNNs) has shown promise in solving highly nonlinear seismic inversion problems. However, direct estimation of CO2 plume extent/saturation from time-lapse seismic gathers using DL is still underexplored, with no reported field applications to date. The investigation of field data is primarily hindered by scarcity of field data for neural network training. Other obstacles include highly nonlinear seismic-petrophysics inverse relationship, and presence of noise in field seismic data. We introduce SeisCO2Net, a deep CNN that predicts CO2 saturation maps directly from time-lapse full waveform shot gathers. For training, we use site-specific geological information, fluid flow physics, rock physics, and seismic modeling to generate synthetic datasets that closely resemble the CO2 storage site. Synthetic tests show promising results, inspiring us to apply SeisCO2Net's trained weights on field data collected at Frio-II GCS site by leveraging transfer learning principles. As reference, we compare SeisCO2Net's predicted CO2 saturation maps with results obtained from physics-based inversion. Our analyses show both methods display similar CO2 plume shapes, reasonable CO2 plume characteristics, and comparable saturation values. Our results suggest pre-training CNNs on physics-informed synthetic datasets and then applying the learned weights to field data is a viable approach to estimating field CO2 saturation. This method effectively addresses the scarcity of field training data, thus encouraging the feasibility of long-term GCS monitoring.

The Broadband Virtual Shot Gathers Construction Based on High-Speed Train-Induced Seismic Wave

The Broadband Virtual Shot Gathers Construction Based on High-Speed Train-Induced Seismic Wave

Experiment and Application of A Aparker Source in Seismic Exploration in Coastal Tidal Flats

We investigate a field experiment in coastal tidal flats in Jiangsu Province using a sparker source, to test the seismic data acquisition system and detect the shallow layers in the real scenarios. We also compare the geophones with 60Hz and 10Hz natural frequency in shot gather as well as in stacked profiles. We process the different data separately, and then process them together in a wide-line mode. As predicted, the fold is doubled showing a higher signal to noise ratio. Through the routine multi-domain integrated noise removal and effective data processing, the signal-to-noise ratio and resolution of the seismic data get largely improved, and thus we successfully obtain the high-quality seismic profiles. The practical experiment results are of importance for shallow geological and environmental detections, as well as for geo-hazard prevention study in the related areas.

Detection of Low-Velocity Layer Using Passive Surface-Wave Data and Deep Learning: An Application in The Urban Area of Hangzhou, China

Passive surface-wave methods using dense seismic arrays have gained growing attention in near-surface high-resolution imaging in urban environments. Deep learning (DL) can release a tremendous workload brought by dense seismic arrays. We presented a case study of shear-wave velocity (Vs) structure imaging in the Hangzhou urban area (eastern China) using DL inversion. Noise data were recorded by dense linear arrays with approximately 5 m spacing deployed along two crossing roads for investigating the top 80 m of the subsurface. Phase-velocity dispersion curves are extracted from virtual shot gathers using multichannel analysis of surface waves. We divided the area where the low-velocity layer (LVL) may exist into three layers with a thickness of 5 m. We gave the four layers weak constraints to generate training dataset and adopted a convolutional neural network to directly invert fundamental-mode Rayleigh-wave phase velocity for 1D Vs models. To improve the accuracy, we further applied the sensitivities to weight the loss function in DL inversion. The obtained pseudo-2D Vs profiles correspond to the velocities estimated from logging data and previous survey. The well-trained neural network successfully identified that the LVL is located at 50-60 m deep. And this network was also achieved accurately the inversion of a dense seismic network nearby. The results of this survey demonstrate the accuracy and efficiency of delineating near-surface structures from traffic-induced noise using the DL technique, which has great potential for monitoring subsurface changes in urban areas.

Trace-wise coherent noise suppression via a self-supervised blind-trace deep-learning scheme

Seismic data denoising via supervised deep learning is effective and popular but requires noise-free labels, which are rarely available. Blind-spot networks circumvent this requirement by training directly on noisy data and have been demonstrated to be a powerful suppressor of random noise. In this work, we expand the methodology of blind-spot networks to create a blind-trace network that successfully removes trace-wise coherent noise. An extensive synthetic analysis illustrates the denoising procedure’s robustness to varying noise levels and varying numbers of noisy traces within shot gathers. It is demonstrated that the network can accurately learn to suppress the noise when up to 60% of the original traces are noisy. Furthermore, our procedure is implemented on the Stratton 3D field data set and demonstrates the restoration of the previously corrupted direct arrivals. In addition to trace-wise noise suppression, we adapt the blind-spot networks to the successful suppression of colored Gaussian noise, which exhibits varying coherent properties in time and spatial axes. Our adaptation of the blind-spot networks paves the way for its use in other applications, such as the suppression of coherent noise arising from wellsite activity, passing vessels, or nearby industrial activity.

Relative Amplitude Preservation Analysis on Interpolation Methods of The Unaliased F-K Trace Interpolation and Regularized Nonstationary Autoregression

The seismic data interpolation method has been widely used to increase the fold coverage in seismic data processing. This technique can be applied to convert multi-2D lines into pseudo-3D, which is an alternative to obtaining 3D seismic volume data due to the relatively high acquisition cost. However, the quality of the seismic interpolation results is not the same as the real 3D seismic data acquisition results. This study carefully analyzed these differences to understand how accurate the results were. There are two methods used for data interpolation, namely Unaliased f-k trace interpolation (UFKI) and Regularized Interpolation Nonstationary Autoregression (RNA) methods, which are applied to 2D pre-stack data to increase the fold coverage and 3D data to convert multi-2D lines into pseudo-3D. Then, the interpolation results on the pre-stack data are evaluated on the 2D and 3D data, and an amplitude change is analyzed. It is done to test whether the amplitude of the seismic data from the interpolation results is still relatively preserved based on the evaluation results of the changes in the AVO response. The results show that the interpolation process in the receiver and shot gather domain (UFKI and RNA) could increase the fold coverage and maintain the relative amplitude preservation and AVO response

Spatiotemporal evaluation of Rayleigh surface wave estimated from roadside dark fiber DAS array and traffic noise

Seismic imaging and monitoring of the near-surface structure are crucial for the sustainable development of urban areas. However, standard seismic surveys based on cabled or autonomous geophone arrays are expensive and hard to adapt to noisy metropolitan environments. Distributed acoustic sensing (DAS) with pre-existing telecom fiber optic cables, together with seismic ambient noise interferometry, have the potential to fulfill this gap. However, a detailed noise wavefield characterization is needed before retrievingcoherent waves from chaotic noise sources. We analyze local seismic ambient noise by tracking five-month changes in signal-to-noise ratio (SNR) of Rayleigh surface wave estimated from traffic noise recorded by DAS along the straight university campus busy road. We apply the seismic interferometry method to the 800 m long part of the Penn State Fiber-Optic For Environment Sensing (FORESEE) array. We evaluate the 160 virtual shot gathers (VSGs) by determining the SNR using the slant-stack technique. We observe strong SNR variations in time and space. We notice higher SNR for virtual source points close to road obstacles. The spatial noise distribution confirms that noise energy focuses mainly on bumps and utility holes. We also see the destructive impact of precipitation, pedestrian traffic, and traffic along main intersections on VSGs. A similar processing workflow can be applied to various straight roadside fiber optic arrays in metropolitan areas.

Deghosting dual‐component streamer data using demigration‐based supervised learning

AbstractGhost reflections from the free surface distort the source signature and generate notches in the seismic amplitude spectrum. For this reason, removing ghost reflections is essential to improve the bandwidth and signal‐to‐noise ratio of seismic data. We have developed a novel approach that involves training a convolutional neural network to remove source and receiver ghosts from marine dual‐component data. High‐quality training data is essential for the network to produce accurate predictions on real data. We have used the demigration of a stacked depth‐migrated image to create training shot gathers. Demigrated pressure and vertical velocity data are used to train the network. We apply the trained network on real pressure and vertical velocity data with ghosts. The network's output may be either source deghosting and receiver deghosting, or both. We test our method on synthetic Marmousi and real North Sea data with dual‐component streamers. The method is compared with conventional dual‐component deghosting using the summation of pressure and vertical velocity. Results show that the method can accurately remove the ghosts with only minor errors in synthetic data. Based on a decimation test, the method is less affected by spatially aliased data than a conventional method, which could benefit data with high frequencies and/or large receiver or cable separations. On real data, the results show consistency with conventional deghosting, both within and outside the training area. This indicates that the method is a viable alternative to conventional methods on real data.

Shallow S wave velocity profile from active source seismic data at the Apollo 14 landing site based on virtual multichannel analysis of surface waves

We used active source seismic data at the Apollo 14 landing site to investigate the shallow S wave velocity structure (from the surface to a depth of 10 m) of the Moon. We applied seismic interferometry to data from shots located between two geophones to construct a virtual shot gather for multichannel analysis of surface waves (MASW). We included continuous wavelet transform processing in the MASW analysis to improve the signal-to-noise ratio of the data before estimating the S wave velocity VS profile. The resultant profile showed gradual increase in S wave velocity from the surface (VS = ∼50 m/s) to a depth of 6 m (VS = ∼100 m/s), followed by a harder layer with S wave velocity of 110 m/s. This study demonstrated that future extraterrestrial active seismic surveys can be accomplished faster, with simpler field geometry, and using fewer shots and receivers than past missions, thus reducing the payload and cost of future exploration missions.