Least-squares data-to-data migration: An approach for migrating free-surface-related multiples

Geophysics ◽  
2019 ◽  
Vol 84 (2) ◽  
pp. S83-S94 ◽  
Author(s):  
Yikang Zheng ◽  
Yibo Wang ◽  
Xu Chang

Free-surface-related multiples can provide extra illumination of the subsurface and thus can be usefully included in migration procedures. However, most multiple migration approaches require separation of primaries and free-surface-related multiples or at least prediction of multiples in advance, which is time consuming and prone to errors. The data-to-data migration (DDM) method migrates free-surface-related multiples by forward and backward propagating the recorded full data (containing primaries and free-surface-related multiples). For DDM, there is no need to predict or separate multiples, but the migration results suffer from the crosstalk generated by crosscorrelations of undesired seismic events, e.g., primaries and second-order free-surface-related multiples. We have developed least-squares DDM (LSDDM) for marine data to eliminate the crosstalk generated by DDM. In each iteration, the forward-propagated primaries and free-surface-related multiples are crosscorrelated with the backward-propagated primary and free-surface-related multiple residuals to form the reflectivity gradient. We use a three-layer model and the Marmousi model for numerical tests. The results validate that LSDDM can provide a migrated image with higher signal-to-noise ratio and more balanced amplitudes than DDM. The LSDDM approach might be valuable for general subsurface imaging for marine seismic data when the migration velocity is accurate, and the acquired data have sufficient recording time.

Geophysics ◽  
1984 ◽  
Vol 49 (8) ◽  
pp. 1223-1238 ◽  
Author(s):  
John T. Kuo ◽  
Ting‐fan Dai

In taking into account both compressional (P) and shear (S) waves, more geologic information can likely be extracted from the seismic data. The presence of shear and converted shear waves in both land and marine seismic data recordings calls for the development of elastic wave‐migration methods. The migration method presently developed consists of simultaneous migration of P- and S-waves for offset seismic data based on the Kirchhoff‐Helmholtz type integrals for elastic waves. A new principle of simultaneously migrating both P- and S-waves is introduced. The present method, named the Kirchhoff elastic wave migration, has been tested using the 2-D synthetic surface data calculated from several elastic models of a dipping layer (including a horizontal layer), a composite dipping and horizontal layer, and two layers over a half‐space. The results of these tests not only assure the feasibility of this migration scheme, but also demonstrate that enhanced images in the migrated sections are well formed. Moreover, the signal‐to‐noise ratio increases in the migrated seismic section by this elastic wave migration, as compared with that using the Kirchhoff acoustic (P-) wave migration alone. This migration scheme has about the same order of sensitivity of migration velocity variations, if [Formula: see text] and [Formula: see text] vary concordantly, to the recovery of the reflector as that of the Kirchhoff acoustic (P-) wave migration. In addition, the sensitivity of image quality to the perturbation of [Formula: see text] has also been tested by varying either [Formula: see text] or [Formula: see text]. For varying [Formula: see text] (with [Formula: see text] fixed), the migrated images are virtually unaffected on the [Formula: see text] depth section while they are affected on the [Formula: see text] depth section. For varying [Formula: see text] (with [Formula: see text] fixed), the migrated images are affected on both the [Formula: see text] and [Formula: see text] depth sections.


Geophysics ◽  
2019 ◽  
Vol 84 (6) ◽  
pp. R989-R1001 ◽  
Author(s):  
Oleg Ovcharenko ◽  
Vladimir Kazei ◽  
Mahesh Kalita ◽  
Daniel Peter ◽  
Tariq Alkhalifah

Low-frequency seismic data are crucial for convergence of full-waveform inversion (FWI) to reliable subsurface properties. However, it is challenging to acquire field data with an appropriate signal-to-noise ratio in the low-frequency part of the spectrum. We have extrapolated low-frequency data from the respective higher frequency components of the seismic wavefield by using deep learning. Through wavenumber analysis, we find that extrapolation per shot gather has broader applicability than per-trace extrapolation. We numerically simulate marine seismic surveys for random subsurface models and train a deep convolutional neural network to derive a mapping between high and low frequencies. The trained network is then tested on sections from the BP and SEAM Phase I benchmark models. Our results indicate that we are able to recover 0.25 Hz data from the 2 to 4.5 Hz frequencies. We also determine that the extrapolated data are accurate enough for FWI application.


2017 ◽  
Author(s):  
Raimundo N. C. Carneiro ◽  
Lourenildo W. B. Leite* ◽  
Wildney. W. S. Vieira* ◽  
Cleudilene S. Rufino

Geophysics ◽  
2008 ◽  
Vol 73 (3) ◽  
pp. Q9-Q17 ◽  
Author(s):  
Martin Landrø

In marine seismic acquisition, the typical time interval between two adjacent shots is about [Formula: see text]. This interval is considered sufficient to avoid the signal from one shot interfering with the signal from the next shot. However, when we are looking for very weak signals or weak changes in a given signal (time-lapse seismic), the influence of the shot-generated noise can be of importance. In the present work, shot records with a recording time of [Formula: see text] are used to analyze the influence of the shot-generated noise from the previous shot. Simple decay models are used to match the observed rms decay curves. These calibrated models are used to estimate variations in signal-to-noise ratio versus shot time interval and source strength. For instance, if the source strength is doubled and the time interval between two shots is increased from [Formula: see text], an improvement in the signal-to-shot-generated noise from the previous shot of [Formula: see text] is expected. Especially for time-lapse seismic using permanently installed receivers, this way of increasing the S/N might be useful.


Geophysics ◽  
2021 ◽  
pp. 1-42
Author(s):  
Shaoping Lu

In marine seismic exploration, it has been well known that sea surface-related multiples can be treated as signals to image the subsurface and provide extended illumination. Previous studies on imaging of multiples have been mainly focusing on its algorithm development and implementation. This paper serves as a tutorial where we systematically investigate the fundamental challenges in the process of imaging of multiples. We first examine the impacts of marine seismic data acquisition parameters: such as offset, trace spacing and streamer towing direction, which are all key elements that control the quality of the images of multiples, and illustrate that 3D towed streamer and OBS surveys are preferable acquisition geometries to apply imaging of multiples. In addition, we investigate the challenges in jointly imaging primaries and multiples and the crosstalk problem in the process, and demonstrate that a Least-Squares inversion based algorithm is effective to address these issues. With the proper handling of all those challenges, imaging of multiples can help to mitigate shallow acquisition footprints, improve salt boundary illumination and enhance the imaging resolution, which allow the identification of drilling hazards and reduction in drilling risks. To apply imaging of multiples in practice, the objective is not to replace but to augment imaging of primaries by providing extra illumination.


2015 ◽  
Vol 33 (3) ◽  
pp. 403
Author(s):  
Lourenildo W.B. Leite ◽  
J. Mann ◽  
Wildney W.S. Vieira

ABSTRACT. The present study results from a consistent processing and imaging of marine seismic data from a set collected over sedimentary basins of the East Brazilian Atlantic. Our general aim is first to subsidize geological interpretations with plausible subsurface images for oil and gas exploration. In second place, to verify published schematic geological interpretation for these basins by underlying the sediment/basement contact, from where subvertical faults are projected upwards through the basin followed by folded structures. The data-driven results can be used to trace the reflector boundaries in the time sections, submitted to time-to-depth axis transformation, and to be used as a first model for further basin pressure prediction, where natural pumps necessarily develop for the mechanism of oil and gas accumulation. The applied fundamental techniques are mainly based on the data-driven common reflection surface stack, where it is shown the improvement of the signal-to-noise ratio, the lateral continuity of the reflection events, the resolution, and that time migration collapses the diffraction events. The CRS migration strongly collapsed the diffraction events, allowing some subsurface structures be more evident. The free surface and some shallow internal multiples can be clearly traced for further processing aiming at their attenuation. The interpretation lines are meant to show the geometry of selected reflectors, and to help comparing the results with other similar sections. One can trace some subvertical fault systems starting from the lower part of the section (interpreted as the basement), and their extension upwards through the sedimentary sequence.Keywords: CRS stack, CRS migration, residual static correction, NIP wave tomography. RESUMO. O presente artigo resulta de um processamento e imageamento consistentes de dados sísmicos marinhos de levantamento realizado em bacias sedimentares do Atlântico do Nordeste brasileiro. Nossos objetivos gerais são em primeiro lugar subsidiar as interpretações geológicas com imagens plausíveis do subsolo, e voltadas à exploração de óleo e gás. Em segundo lugar, verificar as interpretações geológicas esquemáticas publicadas para estas bacias, para conferir o delineamento do contato sedimento/embasamento, de onde falhas subvertical são projetadas através da bacia, seguidas de estruturas dobradas. Os resultados baseados em dados reais podem ser usados para delinear interfaces refletoras contidas nas seções tempo, submetidos à transformação da coordenada tempo para profundidade, e que podem ser usados posteriormente como um primeiro modelo para a predição de pressão em bacias sedimentares, onde se desenvolve um bombeamento natural necessário para a acumulação de óleo e gás. As técnicas fundamentais aplicadas baseiam-se principalmente no denominado empilhamento de superfície de reflexão comum, baseado em dados observados, onde se mostra a evolução da relação sinal-ruído, da continuidade lateral dos eventos de reflexão, da resolução, e o colapso dos eventos de difração nas seções de migração do tempo. A migração CRS colapsa fortemente os eventos de difração permitindo que algumas estruturas do subsolo sejam mais evidentes. Múltiplas da superfície livre, e algumas internas rasas, podem ser claramente traçadas para processamento adicional que visam a atenuação. As linhas de interpretação trac¸adas visam mostrar a geometria dos refletores selecionados, e ajudar na comparação com outros resultados de seções semelhantes. Pode-se traçar um sistema de falhas subvertical a partir da base inferior (interpretada como o embasamento) da seção escolhida como referência, e os seus prolongamentos através da sequência sedimentar.Palavras-chave: empilhamento CRS, migração CRS, correção estática residual, tomografia NIP.


2020 ◽  
Vol 8 (4) ◽  
pp. T941-T952
Author(s):  
Jiachun You ◽  
Yajuan Xue ◽  
Junxing Cao ◽  
Canping Li

Because swell noises are very common in marine seismic data, it is extremely important to attenuate them to improve the signal-to-noise ratio (S/N). Compared to process noises in the time domain, we have built a frequency-domain convolutional neural network (CNN) based on the short-time Fourier transform to address swell noises. In the numerical experiments, we quantitatively evaluate the denoising performances of the time- and frequency-domain CNNs, compare the impacts of network structures on attenuating swell noises, and study how network parameter choices impact the quality of the denoised signal based on peak S/N, structural similarity, and root-mean-square-error indices. These results help us to build an optimal CNN model. Furthermore, to illustrate the superiority of our proposed method, we compare the conventional and proposed CNN methods. To address the generalization capability of CNN, we adopt transfer learning by using fine tuning to adjust the weights of the pretrained model with a small amount of target data. The application of transfer learning improves the quality of the denoised images, which further proves that our proposed method with transfer learning has the potential to be deployed in actual seismic data acquisition.


Geophysics ◽  
2009 ◽  
Vol 74 (2) ◽  
pp. Q17-Q25 ◽  
Author(s):  
Thomas Elboth ◽  
Bjørn Anders Reif ◽  
Øyvind Andreassen

Various weather-related mechanisms for noise generation during marine seismic acquisition were addressed from a fluid-dynamic perspective. This was done by analyzing a number of seismic lines recorded on modern streamers during nonoptimal weather conditions. In addition, we examined some of the complex fluid-mechanics processes associated with flow that surrounds seismic streamers. The main findings were that noise in the [Formula: see text] range is mostly the result of direct hydrostatic-pressure fluctuations on the streamer caused by wave motion. For normal swell noise above [Formula: see text] and for crossflow noise, a significant portion of the observed noise probably comes from dynamic fluctuations caused by the interaction between the streamer and fluid structures in its turbulent boundary layer. This explanation differs from most previous work, which has focused on streamer oscillations, bulge waves inside old fluid-filled seismic streamers, or strumming/tugging as the main source of weather-related noise. Although modern streamers are less sensitive to such sources of noise, their ability to tackle the influence on turbulent flow noise has not improved. This implies that noise induced by turbulent flow has increased its relative impact on modern equipment. To improve the signal-to-noise ratio on seismic data, design issues related to flow noise must be addressed.


Sign in / Sign up

Export Citation Format

Share Document