Back-Propagating into the Future
Author: Andrew Long
Industry Insights - 23 September 2020
I discuss the important similarities and distinctions between RTM (Reverse Time Migration), LS-RTM and FWI, and also relate their fundamental reliance upon back-projection to Deep Neural Network training used in many machine learning pursuits. Hence, the title of this newsletter points to the powerful concept that is paving the ways forward for reducing uncertainty in seismic imaging and seismic characterization in coming years. This newsletter is also written to accompany the TechByte webinars on ‘Least-Squares Migration’ and ‘Illumination vs AVO’.
Full Waveform Inversion (FWI)
Author: Andrew Long
Industry Insights - 3 September 2020
Andrew Long attempts to explain FWI in accessible terms by presenting the critical steps for the methodology to yield useful results, and refers to some common ambitions of the many industry and academic implementations: 1. Setting up a scheme that adjusts the differences between modeled and measured shot gathers so that the subsequent steps in the iterative workflow are stable, 2. Isolating only the seismic information that is relevant to the subsurface parameters being isolated, and 3. Guiding the iterative recovery of an accurate subsurface model in a manner that reduces spurious artifacts.
If you’re a strong mathematical geophysicist this is probably below your competency, but if you can tolerate some basic algebra and have an interest in what FWI is about, then hopefully this might help you complete the initial journey. This article is written to accompany the three-part TechByte webinar series on FWI.
Exploration workflow for real-time modelling of rock property and AVO feasibilities in areas with complex burial history — a Barents Sea demonstration
Author: Per Avseth, Ivan Lehocki, Laurent Feuilleaubois, Tore Nordtømme Hansen, Kristian Angard, Cyrille Reiser
First Break - 1 September 2020
The authors describe an integrated workflow where full waveform inversion velocity data, well log data, seismic stratigraphy, basin analysis, and rock physics modelling are combined to produce 3D feasibility cubes of expected rock properties and AVO signatures for a given geological scenario. They show how these AVO feasibility cubes can be updated in real time and used to guide quantitative interpretation studies, and to run sensitivity analysis to derisk potential prospects.
Integrating Rock Physics and Seismic for Enhanced Understanding of Barents Sea Prospectivity
Author: James Moore and Roberto Puiz
GEO - 28 August 2020
When integrated with the PGS MultiClient seismic data library, rockAVO provides a reliable interactive toolset to understand seismic responses and their relation to geological properties.
HPC and the Future of Seismic
Author: Andrew Long
Industry Insights - 5 August 2020
Andrew Long briefly profiles the world’s largest commercial computer installations, and considers their relevance for the future of high-end seismic imaging and AI pursuits by the oil and gas industry.
Leveraging deep learning for seismic image denoising
Author: Elena Klochikhina, Sean Crawley, Sergey Frolov, Nizar Chemingui, Tony Martin
First Break - 1 July 2020
The authors describe a supervised machine learning approach for attenuation of noise, formed by suboptimal destructive interference within the migration process. They outline the training and validation approach of their deep convolutional neural network, and demonstrate its application on field datasets.
GeoStreamer X in the Viking Graben Illuminates Complex Targets
Author: Kai Fløistad, Julien Oukili
GeoExpro - 8 June 2020
To meet the industry demand for improved subsurface insight at a reduced cost and with faster turnaround time than ocean bottom surveys, PGS deployed a pioneering acquisition configuration in the Viking Graben in the autumn of 2019. Three months after the last shot on the GeoStreamer X survey in the Viking Graben, the early-out multi-azimuth (MAZ) PSDM seismic stack was delivered to pre-funders. From an interpreter’s point of view the results are stunning, and what started as a novel concept now provides new insight to geoscience teams in near-field exploration, appraisal and development.
There are many targets in the area, ranging from Eocene sand injectites to fractured basement, and an approach integrating survey design, imaging solutions and quantitative interpretation has enabled a significant improvement in the resolution of these reservoirs. The main imaging challenges are the large shallow channels and massive sand mounds in the Miocene as these introduce shadow zones beneath them. Eocene cemented sand injectites, called V-brights, have anomalously high velocities which historically have been very difficult to estimate. Deeper, a thin, rugose Cretaceous chalk layer causes multiples and dispersion of the seismic energy.
Seismic processing parameter mining – the past may be the key to the present
Author: Tony Martin, Andrew Long, Suhail Butt, Khairom Baharom
First Break - 1 June 2020
The authors describe a proof-of-concept test, mining an historical database to extract appropriate parameter choices for seismic processing. Using quantitative metrics, they compare the results of using the mined parameters with a full-integrity project, and draw conclusions on whether this approach can bypass testing and reduce turnaround times.
Orphan Basin, Canada | From Regional Prospect Screening to Reliable Reservoir Attributes Estimation
Author: Scott Opdyke, Cyrille Reiser, Tiago Alcantara, Elena Polyaeva
GeoExpro - 4 May 2020
A new, high quality regional dataset covering 22 469 sq. km is enabling the identification of important petroleum system elements in the underexplored Orphan Basin, offshore eastern Canada. Block evaluations for the upcoming license rounds can be performed with confidence with this multi-sensor broadband seismic data and its reliable pre-stack attributes. The most recently acquired part of this regional dataset is the North Tablelands survey, acquired in 2019 as an extension of the Tablelands and Long Range surveys, which were shot in 2017–2018 by PGS and TGS. Fast-track data was delivered just five weeks after the last shot of the North Tablelands survey. A full integrity volume is available now and provides a single, continuous high-resolution seismic dataset available for licensing early next year, encompassing the Long Range, Tablelands and North Tablelands surveys.
The music of marine seismic: A marine vibrator system based on folded surfaces
Author: Okwudili Orji, Mattias Oscarsson-Nagel, Walter Söllner, Endrias Asgedom, Øystein Trætten, Rune Voldsbekk
The Leading Edge - 1 April 2020
Marine vibrators have bespoke geophysical benefits that are yet to be harnessed because of robustness and efficiency issues. We have developed a new marine vibrator source technology that is efficient and stable. The source technology overcomes the historical problems of inefficiency and robustness by using folded surface technology and resonance frequency tuning. We show measured output examples that demonstrate that the folded surface concept combined with small displacements can provide the required output levels. Our source system consists of a low-frequency module covering 1–10 Hz and a high-frequency module covering 10–125 Hz. The source control system has shown high stability and precision and can handle harmonic distortion. With the aid of synthetic data examples, we demonstrate that seismic data acquired using marine vibrators in either intermittent or continuous mode can be processed. Finally, we demonstrate the environmental friendliness of the source in comparison to air gun-based sources.