Search

Multi-component data recording from ocean-bottom seismic (OBS) surveys captures both PP and PS (converted wave) events. Processing such data can produce superior images compared to those obtained from conventional streamer acquisitions. In addition, PP and PS images can provide valuable insights into reservoir properties. However, PS imaging needs high-quality and high-resolution P- and S-wave velocity models in depth. While full-waveform inversion (FWI) for P-wave velocity model building is well established, an equivalent tool for updating the S-wave velocity (Vs) is still a challenge. A recent FWI methodology based on PS reflection data (PS-RFWI) has been proposed for the Vs model building. Updates from this technique are typically low wavenumber in nature. In this abstract, we show an application of PS-RFWI to OBS data from the Central North Sea and demonstrate an approach to update the high-wavenumber Vs components. Our real data application produces a high-quality 30 Hz Vs model that reduces the image undulations at the reservoir level and allows to generate a subsequent high-resolution Vs FWI Image.

The Dugong area in the Norwegian North Sea was surveyed by North-South (N–S) orientated, variable depth streamer data, and recently, East-West (E–W) orientated triple source multi-sensor data. By reprocessing the original N-S data in combination with the E–W, we found that a combined dual-azimuth (DAZ) volume can provide significant imaging improvements supporting the de-risking of nearfield exploration targets. The uplift came in the form of enhanced structural imaging, resolution, signal-to-noise ratio and amplitude reliability. These were due to the complimentary illumination, sampling, and cable-varying characteristics of the two surveys, combined with advanced DAZ velocity model building and reprocessing methods. The benefits were found to directly aid in development decisions. Firstly, an inversion study utilizing both azimuths in a joint manner yielded more reliable probabilistic estimates of reservoir-level oil sands when compared to a single azimuth inversion due to the richer illumination and hence amplitude fidelity. Secondly, DAZ full-waveform inversion (FWI) imaging facilitated a substantial improvement to near-surface resolution with the potential for shallow hazard identification.

Convolutional-based neural network (CNN-based) architectures have shown promise in performing denoising tasks. However, it can be demonstrated that their predictions are of limited use for some tasks because they produce signal leakage. For these tasks, a possible improvement is to incorporate CNN-based architectures as one component of, rather than replacement for, the conventional denoising algorithms. In this paper, we formally define a class of denoising problems usually solved iteratively for which using CNN-based predictions as an initial solution can improve efficiency. We illustrate our points using a land data deblending example, for which the CNN-based prediction quality was higher than that of the conventional first iteration but lower than that of the final product. The CNN-complemented conventional deblending leads to satisfactory and efficient results.

Over two million files, containing geochemical information, have been collected from tens of thousands of wells drilled during decades of exploration in the Gulf of Mexico (GOM) and are available to geoscientists in the public domain. While these files represent a vast knowledgebase covering subsurface geology and petroleum systems, data extraction, systematic compilation and quality control was previously too cumbersome to harness the full power of the data to make basin wide correlations, uncover new trends and ultimately opportunities. A novel machine learning approach was employed to automate data classification and extraction across three protraction areas for all public domain geochemistry and PVT documents to provide a single consistent database from un-tagged, legacy formats stored in entirely different subfolders. The resulting database provides the ability to rapidly screen and integrate data from multiple disciplines over a large scale, in terms of data volume as well as geospatial coverage. This in turn opens up petroleum systems analysis work to a wider user base by acting as a bridge between disciplines, such as reservoir engineering and geochemistry. Removing disciplines from silos is critical to enhancing collaboration between teams, improving efficiencies around specific workflows such as fluid property prediction and therefore reducing uncertainty.

A subsurface volume that can be reliably interpreted in terms of geologically-relevant attributes is a reasonable objective for products from depth inversion workflows. Commonly the field geophysics data available are inherently non-unique and deficient (noise, aliasing, etc.), so an implementation of some type of constraint is required to encourage reasonable inversion outputs. We illustrate an implementation of cross-gradient inversion where surface geological information is included in the input data set. The basic application covers the usual structural similarity objective – comparing the gradient fields of distinct property volumes derived from different geophysical domains – but a particular advantage comes when including gradients derived from surface or subsurface geology, or any ancillary property set, providing reference gradient control during single or joint domain inversions of geophysical data.

The development of time-lag FWI (TLFWI) in recent years has enabled the use of the full wavefield (primary reflection, multiple, ghost, and diving waves) in inversion. With this advance it is now possible to include ever more detail in the velocity model, ultimately reaching the point of deriving from the velocity a migration-like reflectivity image, called the FWI Image. When the FWI maximum inverted frequency is increased, velocity model details can reveal superior reservoir information than present in recent conventional imaging results. Two case studies will be discussed, the first in the Greater Castberg area where the 150 Hz FWI Imaging greatly surpassed the Q Kirchhoff pre-stack depth migration imaging from the water bottom level down to the reservoir, located at a depth of around 1.5 km. For the second, over the Nordkapp basin, use of the full wavefield for shallow ultra-high resolution (UHR) imaging run at 200 Hz revealed reverse faulting and pockmark details that were invisible with either KPSDM or RTM. By using additional information present in multiple, ghost and diving waves, a spatial resolution of 2 m was achieved, making it possible to image very thin features without the need for a dedicated high-resolution acquisition design. The current UHR FWI Image obtained in the near-surface can then be used to de-risk and plan well placement as well as the foundations for wind turbines, providing important velocity information in addition to the reflectivity image.

Although the resolution of a seismic image is ultimately bound by the spatial and temporal sampling of the acquired seismic data, the seismic images obtained through conventional imaging methods normally fall far short of this limit. Conventional seismic imaging methods take a piecemeal approach to the imaging problem, with many steps designed in preprocessing, velocity model building, migration, and post-processing to solve one or a few specific problems at each step. The inefficacies of each step and the disconnects between them lead to various issues, such as velocity errors, residual noise and multiples, illumination holes, and migration swings, that prevent conventional imaging methods from obtaining a high-resolution image of good S/N and well-focused details. In contrast, full-waveform inversion (FWI) Imaging, which models and uses the full-wavefield data, including primaries and multiples, reflections, and transmission waves, to iteratively invert for the velocity and reflectivity in one go, is a systemic approach to address the imaging issues. FWI Imaging has proven to be a superior method over conventional imaging methods for providing seismic images of greatly improved illumination, S/N, focusing, and thus higher resolution. We demonstrate with a towed-streamer data set and an OBN data set that FWI Imaging with a frequency close to the temporal resolution limit of seismic data (100 Hz or higher) can provide seismic images of unprecedented resolution from the acquired seismic data, which has been impossible to achieve with conventional imaging methods. Moreover, incorporating more accurate physics into FWI Imaging, for instance, upgrading the modeling engine from acoustic to elastic, can substantially improve the seismic resolution further. Elastic FWI Imaging can further reduce the mismatch between modeled and recorded data, especially around bodies of large impedance contrast, such as salt, and appreciably improves the S/N and therefore the resolution of the inverted images. We show with an OBN data set in the Gulf of Mexico that elastic FWI Imaging further improves the resolution of salt models and subsalt images over its acoustic counterpart.

Standard reflection-based model building for land applications is challenging due to reduced data quality, near-surface heterogeneities, and the low-fold of reflection data at shallow depths. Broadband and large offset data acquisitions have been developed with the aim of investigating full waveform inversion (FWI) as an alternative tool for velocity model building. FWI using minimally-processed refractions and diving rays provides an efficient solution to recover longer spatial wavelength details in the velocity model. In recent studies this method has been proposed as a means to guide and enhance standard reflection-based methods. Rather than follow refraction/diving ray FWI with reflection tomography, we apply FWI to conventionally-processed reflection data, and move to higher frequencies (up to 13Hz). By incorporating reflection data and higher frequencies we obtain higher resolution structural details such as channels and fault structures and obtain improved imaging results compared to those found by migration velocity analysis. We outline a refraction data preprocessing sequence tailored for data quality at low frequencies and long offsets, and describe the FWI workflow which uses both refractions/diving rays and conventionally-processed reflections. We show the resolution uplift over refraction-based FWI and compare migrated stacks generated with the standard tomography model to that generated with the high frequency FWI result.

Conventional Full Waveform Inversion (FWI), mostly based on diving-waves, has become a standard velocity model building tool. Using a dataset from the deep water on the Mexican side of the Gulf of Mexico (GoM), we show that FWI can be effective at resolving different types of complex geological features in the shallow overburden. Unfortunately, it is well known that below diving-wave penetration depth, FWI has to rely solely on reflections. In this case, the velocity update is dominated by high-wavenumber components, and the inversion dependence on the accuracy of the density model increases. However, we show that reflections can still help solve for low wavenumbers of the velocity model when different components of the FWI gradient are properly used for Reflection-wave-based Full Waveform Inversion (RFWI). This allowed reflection data to improve the velocity model at a geologically challenging location within our GoM project area, where traditional ray-based tomography failed and where depths exceed diving wave penetration. We discuss the lessons learned on RFWI from this example.