Search

This paper presents leanings from the 2015 and 2016 North Sea acquisition season with regards to acquiring data with lots of seismic interference noise.

The 4D global inversion workflow is a multistage process that includes seismic data preconditioning, wavelet estimation, low-frequency model building, 3D inversion, and finally a 4D global inversion. In this work, focusing on a post-salt turbidite reservoir in the Campos Basin, offshore Brazil, we show how this process can help to better understand the 4D seismic data, bringing a significant uplift in the quality of the mapping of the 4D anomalies.

Kaskida is a large three-way reservoir underneath a complex salt body and truncated by a salt weld. Seismic imaging at the reservoir level is impacted by the inhomogeneous illumination from the complex overburden, which distorts the amplitude of reservoir reflectors and generates lots of migration artifacts. The application of least-squares RTM (LSRTM) at Kaskida field demonstrates that it can effectively improve the signal-to-noise ratio (S/N) of the subsalt image by reducing migration artifacts, and it can improve the reservoir amplitude fidelity by compensating for the illumination effects caused by the overburden velocity and acquisition geometry. We further demonstrate that a more correct velocity model derived from reflection FWI (RFWI) improves LSRTM results by providing a better raw RTM image and more accurate illumination compensation. Finally, we compare common-image gathers (CIGs) from well-data synthetic modelling, raw RTM, and LSRTM and conclude that LSRTM improves the AVO response over raw RTM because of the offset-dependent illumination compensation and reduction in migration artifacts.

Time-lapse (4D) seismic surveys are devised to detect subsurface changes resulting from hydrocarbon production and fluid injection. Full-waveform inversion (FWI) of time-lapse seismic data has been reported to provide high-resolution estimates of 4D changes. However, successful applications of 4D FWI on field data have only been seen in a few surveys, either with high repeatability or with vast velocity changes. We propose a new workflow to tackle some challenges in 4D FWI, such as cycle-skipping and amplitude mismatch between modeled synthetic data and recorded field data and water-layer variations between baseline and monitor surveys. We successfully applied this approach to a 4D towed-streamer survey in the Exmouth basin, Western Australia.

The western part of the Gulf of Mexico (WGOM) is often characterized by large and complex salt and shale bodies, making it notoriously challenging to image deep subsalt targets, especially with only wide-azimuth towed-streamer (WATS) data available. In 2018, BHP acquired a large-scale sparse ocean bottom node (OBN) survey in this region with the intention to derive a more accurate velocity through full-waveform inversion (FWI) for WATS data imaging. The relatively large shot spacing in this survey, compounded by the complex overburden and the strong reliance on long-offset data (>40 km) for deeper model updates, pose considerable challenges to the processing of this sparse nodes survey. We present how we mitigated the noise issue accompanying this sparse nodes survey and made better use of the long-offset data to unlock the full potential of this data set, ultimately obtaining improved velocity model and images in this complex area. We also demonstrate possible ways to further improve the sparse nodes data, such as by using a better low-frequency source.

Seismic images obtained through conventional migration methods have limitations as amplitude distortions and migration artifacts. To mitigate these limitations, we can generate FWI Images by computing the reflector-normal derivatives of the high-frequency FWI velocities. Given the resolution, accuracy, and geological consistency of the velocities obtained from TLFWI, FWI Images can be compared with traditional migration images. Thus, we present the advantages of 3D FWI Images over conventional RTM images when dealing with geological obstructions. By computing FWI Images from the individual FWI velocities of baseline and monitor, we can also illustrate the benefits of FWI Imaging for 4D by retrieving more continuous amplitudes at the reservoir level and analyze the effects of using respective baseline and monitor velocities in 4D images.

Compared with towed-streamer acquisitions, ocean bottom nodes (OBN) generally provide fuller-azimuth illumination of the subsurface, longer offsets, higher signal-to-noise ratio (S/N), and improved low frequencies. These advantages provide the necessary ingredients for two key elements of seismic exploration and monitoring: (i) full-waveform inversion (FWI) with better constrained velocity models and (ii) imaging below complex structures with improved illumination. In order to take full advantage of this data, it is imperative to use the best imaging algorithm available. Reverse time migration (RTM) is well suited for imaging deep and complex structures. Moreover, it can be modified to yield angle-domain common-image gathers (ADCIGs). These gathers inherit the usual benefits of RTM, when compared to other imaging algorithms, while providing prestack images with reliable subsurface information. This information can be used for amplitude variation with angle/offset (AVA/O) inversion, migration velocity analysis (MVA), and other prestack domain methods. We will investigate data sampling issues related to the implementation of RTM 3D angle gathers, which are more prominent in the case of OBN acquisitions, and propose a method for sampling the subsurface reflection energy that better represents the full-azimuth nature of OBN data while honoring the amplitudes of the image gathers.

Seismic time-lapse (4D) surveys have been widely used to quantitatively monitor the geophysical property changes within hydrocarbon reservoirs due to production effects. Full-waveform inversion (FWI), which has become one of the most reliable tools for velocity model building (VMB), is a natural technology choice for this purpose. However, robust reconstruction of time-lapse geophysical property changes within reservoirs using FWI remains challenging. Time-lapse signals are weak compared to the seismic response of the background model and thus vulnerable to noise and possible damage by data preprocessing. Imperfect repeatability of the baseline and monitor surveys also introduces uncertainties into the inverted time-lapse changes. To address these issues, we propose a 4D FWI approach, which jointly inverts both baseline and monitor data sets using a time-lag cost function with a target-oriented regularization scheme. This approach enhances the 4D signal within reservoirs while suppressing 4D noise away from them. We demonstrate our method using a synthetic and a field data example. We observe that it not only gives an interpretable 4D velocity difference, but also improves the 4D migration difference when compared with conventional 4D approaches.

The oil-bearing fractured granite basement rocks form a very important and complicated hydrocarbon reservoir in Cuu Long basin offshore Vietnam. However, the poor fractured basement imaging in the conventional tow streamer data makes it hard for detailed interpretation and future well placement. To improve the seismic imaging, the first 3D/4C OBC acquisition over the field was carried out to provide better illumination, better elimination of multiples and broader spectrum with better signal to noise ratio. However, the presence of strong azimuthal anisotropy poses a serious challenge in imaging this OBC data with the full azimuth (FAZ) nature. The steeply dipping fracture imaging can be smeared if the subsurface orthorhombic (ORT) velocity model is not properly derived. In this paper, we present a new orthorhombic velocity model building workflow to estimate the azimuthal anisotropic velocity by incorporating shear wave splitting analysis, well formation microimager (FMI) information and 3D RTM subsurface angle gather based velocity sweeping inside basement. Two geological layers with strong azimuthal anisotropy are identified and incorporated into the final ORT model which results in much shaper imaging not only in shallow classic sediment layers but also in the fractured basement.