- Home
- Conferences
- Conference Proceedings
- Conferences
First EAGE/PESGB Workshop on Velocities
- Conference date: 22 Feb 2018 - 23 Feb 2018
- Location: London, UK
- ISBN: 978-94-6282-242-9
- Published: 22 February 2018
22 results
-
-
Construction Technique Of High Resolution Velocity Field - New Attribute For Seismic Interpretation
Authors R. Øverås, V. Kalashnikova, S. Guidard and I. MeisingsetIn this work, we propose a technique for High-Resolution Velocity construction based on Amplitude Inversion combined with Dynamic Auto Correlation or Dynamic Time Warping. The main aim for this workflow is to provide a High-Resolution velocity field that can be used as an attribute in data interpretation and classification. Synthetic examples show that the algorithm can reconstruct the P-wave velocities from log data in the case of no anisotropy. We demonstrate cases of constructed velocity field’s implementations for lithology, pore pressure and reservoir highlighting. Taking into account that technique does not require big computation resources, that makes it a convenient seismic interpretation attribute.
-
-
-
Azimuthal Anisotropy Resolved By Tilted Orthorhombic Tomography
Authors C.J. Rudling, A. Riaz and J. SmithThis paper explains a qualitative and quantitative approach to identify and resolve P wave TTI and HTI variations in the depth migration velocity model building process. The combination of these effects can be represented by an orthorhombic model with three mutually orthogonal planes of mirror symmetry; the P-waves in each symmetry plane can be described kinematically as an independent TI model (Song and Alkhalifah, 2013). Orthorhombic velocity model building and imaging tools are required to address both HTI and TTI simultaneously in subsurface. The orthorhombic anisotropy has been correctly identified and resolved, which has subsequently resulted in a high resolution velocity model (figure 8). A combination of the accurate velocity model building and advancement in the pre-processing techniques has shown significant uplift in the image quality on the earlier processing. The re processed PreSDM has provided improved coherency, recovered improved bandwidth, enhanced character, and better fault plane definition.
-
-
-
Understanding Delta Anisotropy On A Regional Scale
Authors I. Meisingset, J. Hubred and D. KrasovaComparison of SCF values between wells in shallow and deep water, from a large database of values from Norway and Australia, shows that delta anisortopy must depend primarily on seismic wave propagation angle. The principle for modelling delta should be to use MSL parallell layers, and not seabed parallell. This is different from what many workers do. The large database at hand for our study makes this a firm conclusion.
-
-
-
FWI Velocity Model Building Experiences
Authors M.J. Hart, J. Sheng, S. Baldock and J. MaoFull Waveform Inversion has been used to generate detailed models of shallow, subsurface velocity variations in some challenging areas. An accurate starting model was typically required to avoid cycle skipping. This could be minimised by incrementally increasing maximum offset, depth and traveltime over a number of iterations. We now incorporate dynamic warping into the early iterations of FWI to get around this problem, allowing us to start model building at the same time as time processing. Image-guided smoothing has also been utilised within FWI to minimise the effect of the acquisition footprint on the resulting velocity model and at the same time, provide structural constraints which further enhance the detail preserved within the model. Initially only diving wave energy was used, which limited the maximum update depth, dependent on offset. Recently we have been able to utilise WavePath FWI, which allows us to reliably use reflection events and constrain deeper updates into a more plausible model of the subsurface. Here we present some examples of our approach to model building with FWI over the past few years.
-
-
-
Earth Model Building With Full-Waveform Inversion - A Case Study From A Shallow Reservoir In The Barents Sea
Authors O.J. Lewis, S. Way, G. Apeland, P. Smith, H. Veire, J.R. Granli, L.M. Moskvil and N. StevensThe Wisting field in the southwest Barents Sea is characterized by a shallow reservoir and has a complex geological history. Previous studies in the area showed that imaging the existing surface seismic data in depth with an accurate TTI earth model, and applying compensation for absorption within the migration gave the optimum preservation of AVO for further reservoir characterization. A full-waveform inversion (FWI) model building study was undertaken to derive a high resolution earth model suitable for depth imaging purposes. Anisotropy and absorption quality factor (Q) were derived and calibrated across the available, high quality borehole data in the survey area and incorporated into the starting model. FWI was employed to update the model over the depth range of interest, with reflection tomography used to update below the maximum penetration depth of the FWI. The resulting model is robust, high resolution and consistent with all available well data.
-
-
-
Optimizing The Model Building Approach Using Full-Waveform Inversion And Multilayer Reflection Tomography - A North Sea Workflow
Authors S. Gupta, A. Cooke, M. Steiger-Jarvis, J. Bailey and A. SellarsTwo of the key challenges for velocity model building in the North Sea are a heterogeneous overburden (for example, the presence of fluvial and sub-glacial channels) and the presence of strong velocity contrasts (most often introduced by a chalk layer). Two model building techniques have been developed in recent years which address these challenges. Full-waveform inversion (FWI) has been shown to be highly effective in resolving overburden heterogeneity and multilayer tomography allows us to preserve sharp contrasts in our velocity models. Herein, we show how the combination of these two techniques provides an effective and efficient model building workflow for the North Sea environment
-
-
-
High Resolution Velocity Estimation Using Refraction And Reflection Fwi - The Fortuna Region, Offshore Equatorial Guinea
Authors I.F. Jones, J. Singh, P. Cox, M. Warner, C. Hawke, D. Harger and S. GreenwoodThe primary objective of this project was to improve the understanding of the internal structure of the Viscata and Fortuna reservoirs, and this objective was met via clearer internal imaging of these reservoir intervals and the overlying gas-charged sediments. The underlying geophysical challenge was the presence of extensive, but small-scale low-velocity gas pockets, which gave rise to significant and cumulative image distortion at target level. This distortion had not been resolved in a vintage 2013 broadband preSDM project, as the velocity model was not sufficiently well resolved. But in the initial commercial phase of this project, high-resolution non-parametric tomography using improved broadband deghosted data enabled us to achieve the stated objectives. The follow-on work, considered here, deals with the use of full waveform inversion, to see if we could further delineate small-scale velocity anomalies, associated with the highly compartmentalized reservoir units.
-
-
-
Improving Seismic Image With High Resolution Velocity Model From AWI Starting With 1D Initial Model - Case Study Barents Sea
More LessA feasibility study was carried out over a prospective structure in the south western Barents Sea, Norway. The need of a high fidelity velocity model to solve the complex velocity variations in the overburden was the driving mechanism for this test project. A shallow gas anomaly associated with amplitude dimming is causing distortions in imaging and leading to big uncertainty concerning fault identification within and mapping of this interval. Through a special application of FWI, the so-called adaptive waveform inversion (AWI) which allowed starting the inversion with a very simple velocity model, we solved the strong lateral velocity variations in the near surface leading to an improved image, demonstrating the superior quality provided by an AWI based velocity model.
-
-
-
High Quality Regional Velocity Modelling For Depth Conversion
Authors I. Meisingset, J. Hubred and D. KrasovaRegional high quality 3D velocity models have been constructed for the purpuse of depth conversion, using world leading mature technology which has been developed, and used, over a period of almost 30 years. The models, which are in daily use in many oil companies, will be presented in open session for the first time.
-
-
-
Modelling Geological Layers Into New Velocity Models For Seismic Migration Process - A Brazilian Pre-Salt Case
Authors J.S. Fonseca, L. Teixeira, A. Maul, P. Barros, F. Borges, J. Boechat and M. GonzálezThis paper presents a methodology to build velocity models that reflect geological features identified in previous migrated seismic, particularly within the Brazilian Santos Basin evaporitic salts section. To characterize a geological layer we propose the use of model-based acoustic inversion that contains the geological knowledge from wells and interpretations together with petrophysical correlations. As an outcome, a new migration process was performed coupling this modelled layer into previous tomographic velocities. The results are shown in gathers and stacked migrated sections with improvements in the flattening and the focusing of reflectors. We also defend that this method can be extended to other geological sections, and furthermore, this constructed velocity model could be a good input to FWI acoustic process.
-
-
-
3D Regional Velocity Modeling - North Sea
Authors C. Magneron, H. Doornenbal, I. Van Bever and M. BotzRegional velocity models are used mainly for time-to-depth conversion. At regional scales, where few horizons are available, the accuracy and consistency of the velocity models relies on the integration of seismic velocities which bring additional information in terms of lateral and vertical heterogeneities. Such heterogeneities can only be preserved with a fully 3D geostatistical approach. Besides, the use of a structurally driven merge and calibration process leads to consistent 3D regional velocity models. Finally, the combination of advanced geostatistical techniques coupled with geophysical and geological expertise allows to obtain realistic and accurate depth conversion products.
-
-
-
Quantified Uncertainty Estimation In Depth Conversion
Authors I. Meisingset, J. Hubred and D. KrasovaDepth conversion with quantified base case and uncertainty estimates of depth and gross rock volume is carried out in an exploration area with five structural closeres, using a demo dataset. The study shows how this innovative method is used stand alone in an exploration context. Use in field development studies is very similar. The study investigates some of the common assumptions made in uncertainty estimation, namely the assumption of normal duistribution, and of symmetrical uncertainties in gross rock wolume around a depth base case. These do not hold up well, indicating that there is potential for improving the quality of uncertainty estimation. The solution is to stop guessing, and to start calculating these things.
-
-
-
PSDM Uncertainty Characterization And Calibration Using A Fast Marching Algorithm
Authors I. Van Bever, T. Demongin, C. Magneron, M. Montouchet and C. WardPSDM provides very good imaging results but is often not optimal for time to depth conversion. PSDM model uncertainties quantification and calibration to well data may benefit from geostatistics. Complexity in the geological structures (presence of a salt dome for example) can be very challenging for spatializing data and requires therefore the use of innovative geostatistical techniques such as kriging with a fast marching algorithm.
-
-
-
Enhanced Velocity Model Building For Low Quality Seismic Data
Authors G.L. Eisenberg-Klein, J. Pruessmann and E. SchuenemannDetermining velocities for seismic processing and imaging remains a big challenge, especially for seismic data with poor data quality or from complex geological settings. State-of-the-art methods to determine velocities for depth imaging can often not be applied to seismic data with poor signal/noise ratio(S/N) or data gaps due to restrictions from acquisition challenges. Multi-parameter stacking techniques simultaneously enhance the data quality and use the related wavefront attributes obtained by coherence analysis for velocity model building in the time and depth imaging workflow. We demonstrate examples using the Common Reflection Surface (CRS) method to efficiently derive velocity models for low quality and/or sparse acquisition data examples. Algorithms including CRS-Tomography, iterative Reverse Time Migration (iRTM) and Full Waveform Inversion (FWI) benefit from the CRS processing sequence, e.g. by addressing challenges in the application to onshore field data related to rugged topography as well as to poor data quality. Recently the CRS technology was extended to derive azimuth dependent wavefront attributes for anisotropic velocity model building. Furthermore, recent research on Diffraction-Tomography indicates the improvement of the resolution of CRS-Tomography using the diffracted wavefield energy.
-
-
-
Joint Inversion And Cross-Validation Of Bore-Hole And Surface Seismic Data For Reducing The Uncertainty Of Anisotropic Models
Authors O.K. Zdraveva, A. Ramirez and S. ChenAdding extra data and information in the process of anisotropic Earth model building (EMB) is very important to reduce the uncertainty of the results. Borehole seismic data, either zero-offset, walkaway or 3D vertical seismic profile (VSP) are very valuable but rarely available in large exploration areas. In recent years, with the wider spread of distributed acoustic sensing (DAS), 3D VSP surveys are becoming more common to help interpretation related to advanced development programs and production well placement. In some cases, they are even preferred as a foundation for 4D work and that sets even higher requirements for the accuracy of the Earth models. In this work, we discuss and show how 3D VSP data can be used in two different stages of modern surface seismic anisotropic EMB workflows: model validation and model updating with joint tomography. We demonstrate the value of incorporating 3D VSP information using a real data example from the Green Canyon area of the Gulf of Mexico.
-
-
-
How Safe Are You Drilling? - Novel Velocity Model Building Methods For Mitigating Drilling Hazards
Authors S. Roy and C. EsmersoyA detailed understanding of velocities in the overburden is critical for the planning and successful drilling of wells. An accurate 3D velocity model can play an important role in identifying potential drilling hazards along the well path. It is well known that pre-drill estimates of velocity (and hence predicted formation properties and structural images) have inherent uncertainties. In this paper, advanced model building methods are presented which make use of well measurements and regional geological knowledge as constraints to high-resolution model building techniques. These methods can provide improved confidence in drilling activities and mitigate drilling hazards by reducing uncertainties, identifying drilling hazards ahead of the bit, interpreting high-resolution structural images, and reducing nonproductive time and drilling cost.
-
-
-
Solving Imaging Challenges In A Deep Water, Complex Ooze Regime - A Case Study From The Outer Vøring Area
Authors S. Naumann, Ø. Korsmo and G. RønholtIn this paper we present a case study from the outer Vøring area in the Norwegian Sea where we applied a Full Waveform Inversion (FWI) approach to build a high resolution velocity model. The water depth and the geological complexity in the study area required the use of both reflected and transmitted waves in the inversion process. The use of reflections in the inversion data window allows updating the velocity model beyond the penetration depth of diving waves. In this case study we performed FWI on a 5500km2 dual sensor streamer dataset and we were able to build a high resolution velocity model down to 4km depth. Furthermore we demonstrate improved imaging results by combining the derived velocity model with a detailed attenuation (Q) model in a Q-Kirchhoff pre-stack depth migration algorithm (Q-KPSDM).
-
-
-
Mobile Shale Model Building And Update Using Multi-Disciplines And A Combination Of FWI And Tomography
Authors S. Chen, J. Dai, L. Zhang, K. Glaccum, B.J. Koechner, S. Dasgupta and A. KlebleevaA large expanse of mobile shale of variable sizes exist throughout most of the Perdido Fold Belt, especially on the western side of this project’s survey. Due to the low shear (high ductility) of mobile shale, like mobile salt, they form a variety of intrusive and extrusive structures that disrupt the natural connectivity of the strata. Extremely low velocities and densities, compounded by strong absorption of seismic wave energy, make mobile shales a huge challenge for modern seismic imaging. To overcome this challenge, a model building workflow was devised including a procedure for mobile shale definition in addition to the conventional Deepwater supra-salt and subsalt depth imaging approach. Application of this new model building workflow on the Perdido Fold Belt survey demonstrated significant improvements in the final seismic image.
-
-
-
Joint Seabed And Streamer Model Building In The North Sea Using Multi-Survey FWI And Tomography For Optimal Illumination And Coverage
Authors A. Bullock, N. Zimmerman, L. Zhang, O. Runde and T. SkorveThe North Sea is a mature hydrocarbon resource area that is well covered by seismic data, however there is a strong industry desire to achieve maximum value from existing datasets without having to acquire a new acquisition carpet each time. We present here a case study from the Ivar Aasen field where the simultaneous use of seabed and towed streamer datasets for FWI, tomography and imaging allowed a continuous high resolution model to be derived over a large area spanning multiple discoveries to support current and future development.
-
-
-
PS Velocity Model Building Of Johan Sverdrup OBS Data - When You Expect It To Be Trivial
Authors H. Mehdi Zadeh, E. Sadikhov, M.S. Guttormsen, B. Abbad, N. Ghani, S. Winterstø, S. Madden, S. Gupta and K. RamaniOcean Bottom Seismic (OBS) survey acquired over the Johan Sverdrup field provided a superior image to the streamer data from P-wave imaging. When processing PS data we observed that PS image ties reasonably well with PP image in depth. However, the PS image below chalk is poorer than the PP image. Despite a generally simple overburden without big shallow gas anomalies, the presence of the high velocity contrast injectite intrusions and a high contrast chalk layer, makes PS imaging difficult in the Johan Sverdrup case. The most likely reason for poor PS image below Shetland could be the mode conversion occurring at the chalk interface. Reverse Time Migration (RTM) of PP data shows improvement when compared to Kirchhoff. Similar improvement is expected from RTM migration on PS data, due better handling of multi pathing and mode converted events.
-
-
-
Elastic Reflection Full Waveform Inversion
Authors W. Weibull and F. EbrahimReflection full waveform inversion is a formulation of full waveform inversion designed to extract low wavenumber velocity information from the reflected wavefield. In this works we extend reflection full waveform inversion to an elastic framework. This allows the estimation of both P-wave and S-wave velocities using reflected waveforms. We apply the method to a 3D-3C field dataset. The results clearly show that the velocities estimated with the method improve the quality of the PS image. The method is also able to incorporate in the velocity model information about the location of channel deposits at a depth of about 1200 meters.
-
-
-
Classification And Suppression Of Blending Noise Using CNN
By R. BaardmanIn this abstract a novel machine learning deblending algorithm is introduced. The method uses a convolutional neural netork (CNN) to classify data patches in a "blended" and a "non-blended" class. A second, regression based, CNN deblends the "blended" patches. Results are shown for a synthetic data example.
-