- Home
- Conferences
- Conference Proceedings
- Conferences
First EAGE Conference on Seismic Inversion
- Conference date: October 26-29, 2020
- Location: Online
- Published: 26 October 2020
1 - 20 of 40 results
-
-
A Bayesian Approach for Resolving OWC and GOC from 4D Seismic Data
Authors H. Amini and C. MacBethSummary4D seismic technology has proven successful in monitoring the post-production fluid contacts in several reservoirs. Most of these cases include thick reservoirs with reasonably good vertical connectivity in which fluid contacts are shown as isolated events on the seismic data. However, challenges still remain when the contacts’ displacements are around or below seismic tuning thickness. Time-lapse wedge modelling and Sim2seis proved to be useful in providing an insight into the problem, however, it is not an easy task to link the modelling results to the observed 4D response. Here, by application of engineering driven simplifying assumptions on the geometry of the contacts and saturation profile within each fluid zone a Bayesian approach is proposed to invert the 4D seismic data to post-production contacts. As wells as the best case scenario, this method provides the uncertainty associated with the contacts’ locations. This approach could be used to quantify the location and thickness of the displaced oil rim for effective management and development of the field.
-
-
-
Keynote 1: Facies Probabilities – Seismic Inversion Coming of Age
By P. ConnollySummaryMethods for the inversion of seismic amplitudes have been available since the 1970s but their track record of delivering useful results has been, at best, mixed. The realisation of the shortcoming of each method, for example,the failure to account for AVO effects or the ignoring of a key source of uncertainty, has been followed by a change of direction and a new inversion framework, until that one too proved to be inadequate. But perhaps the most recent class of inversion applications, to estimate facies probabilities, will prove to be an approach that can deliver a reliable product. There is still much work to be done to improve the performance of these algorithms; analysis of the inversion problem within this framework more clearly demonstrates the difficulty, but this time the effort could result in successful outcomes.
-
-
-
The Seismic Waveform Indicator Inversion Method and Application in Prediction Sandstone at High Water-Cut Oilfield
More LessSummaryWe applied inversion method of seismic waveform indicator to obtain sandstone distribution in time domain by incorporating both well-log data and seismic data in an area of dense wells in B3D block in Changyuan oil field, to get the accurate distribution of channel-sand between wells. Seismic waveform indicator method is to extract special characteristics of representing certain strata features and combinations patterns from seismic waveforms and well logs, instead of obtaining variogram function based on some wells in geostatistical inversion approach. Comparing the two different inversion methods of waveform indicator and geostatistics, it showed that the former has more reliably high spatial resolution, and the thickness of sand body over 2m could be effectively identified at present well pattern conditions. It proved that this method is effective for the reservoir prediction in continental multilayer heterogeneous thin interbedded reservoir.
-
-
-
Using Stochastic Inversion Results for Reservoir Model Population - a Case Study from the Schiehallion Field
Authors S. Grant, D. Cox, L. Wang and M. Le GoodSummaryBP’s one-dimensional stochastic inversion method (ODiSI) has been widely applied to estimate reservoir properties, facies probabilities and associated uncertainties from seismic data. This paper describes how ODiSI has been applied to the Schiehallion field, discusses how seismic data inaccuracies are affecting the ODiSI products, and shows how the ODiSI reservoir property estimates are being used in reservoir model construction.
-
-
-
Effects of Wavelet Uncertainty on Seismic Inversions
More LessSummaryWe consider the effect of wavelet uncertainties on inversion results in three distinct cases: one synthetic case, one elastic inversion and one time-lapse inversion case. Samples of the posterior distribution of the wavelet extracted in the impedance domain are used to invert test cubes with different realisations of the extracted wavelets. We analyse the effects of wavelet variability on the MAP result of the seismic inversion. Variations from the wavelet alone are in the order of 3-10% of the inversion result, enough to be significant in facies classification, and time-lapse interpretations.
-
-
-
Inversion in the Presence of Anisotropy
Authors P. Mesdag, L. Quevedo and C. TanaseSummarySeismic anisotropy affects both the propagation and the reflection properties of the subsurface. Changes in the travel time need to be accounted for during processing or as a post processing residual alignment. Changes in the AVO reflection coefficients may seriously skew the elastic parameters from seismic inversion and, therefore, the way in which we characterize the reservoir.
To be able to use our isotropic pre-stack modelling and inversion in an anisotropic setting, the concept of effective elastic parameters was introduced previously ( Mesdag et al. 2013 ). These effective elastic parameters account for the bias in the inversion results.
In this paper we show how this works in a vertical transverse isotropy (VTI) and a horizontal transverse isotropy (HTI) case. Knowing the anisotropic parameters, the formulations can also be used to predict the outcome of a seismic experiment. This will be shown based on the SEG Barrett SEAM 2 model.
-
-
-
Forward Modelling and Inversion with 3D Wavelets
More LessSummaryIn the industry, most forward modelling and inversion algorithms use trace-by-trace convolution with a single trace wavelet to generate synthetics. Even for full waveform synthetics, often the reflectivity is first calculated and then a convolution with a single trace wavelet is performed.
It is well-known that the result of real data acquisition, processing and imaging produces a wavelet that is extended into the lateral directions. The imaged result of a point reflector is a three-dimensional point spread function (PSF). In this paper, we discuss the design and implementation of 3D wavelets in both forward modelling and inversion. Using synthetic examples we show the potential benefits of using 3D wavelets over the single trace wavelet.
-
-
-
Probabilistic AVO Inversion of Transversely Isotropic Medium for Better Characterization of North Sea Oxfordian Turbidite Reservoir
Authors E. Aker, J. Thiebaud and P. RøeSummarySedimentary rocks often obey vertical transvers isotropy due to the nature of the sedimentation process or the inherent orientation of rock grain minerals (e.g. clay platelets). Despite several documented cases in the literature, the effect of anisotropy is often neglected in seismic AVO inversion. We study the effect of transverse isotropy in shale encasing isotropic sand using probabilistic AVO inversion. The inversion algorithm is inverting seismic pre-stack data to lithology and fluid probabilities. A synthetic case demonstrates how anisotropy in shale may lead to wrong interpretation of fluid content in underlying sand. A field case in the northern North Sea shows that accounting for anisotropy in shale may improve the discrimination between good and poor sand of an Oxfordian turbidite reservoir. The results are encouraging and consistent with observations of sandy intervals in the wells and the depositional system of the study area. Our results clearly demonstrate that transverse anisotropy in sediments may give a significant contribution to the AVO gradient that otherwise could be misinterpreted. In the study area it becomes especially important, since there is a lack of contrast in acoustic impedance between the encasing shale and the target sand.
-
-
-
Review of Porosity Inversion in Chalk
Authors H. Wagner, H. Klemm and H.P. HansenSummaryIn this note, we share a collection of experiences of porosity prediction in chalk gathered over the last decade from work in the North Sea. Due to the rock properties of chalk, porosity estimation in chalk is very effective, whereas fluid prediction is challenging. Through rock physics we illustrate how the ability to predict porosity in chalk is caused by the strong dependency between acoustic impedance and porosity which tends to overrule fluid contribution.
Thanks to the well-suited nature of chalk to estimate porosity the results can be useful for static geologic modelling, and volume estimation. This stresses the importance to address the uncertainties on the final porosity volumes.
We have found the input prior model to be the main contribution to the uncertainty on volumes compared to lithological variations or seismic noise. Therefore, we suggest adding further constraints to the prior than well data and seismic horizons. Seismic velocities are well suited for this purpose, but their quality especially at depth is often problematic. Going forward it is recommended to include new sources of information e.g. depth effects such as hydrocarbon porosity preservation effects.
-
-
-
Impact of a Markov Random Field Prior Model in Seismic Inversion
Authors T. Fjeldstad, P.Å. Avseth and H. OmreSummaryWe present a simultaneous Bayesian inversion framework for prediction and uncertainty quantification of lithology/fluid classes, petrophysical properties and elastic attributes from seismic amplitude-versus-offset observations. Prediction of these reservoir variables are of interest to generate initial models of the reservoir to model fluid flow. We consider a Gauss-linear likelihood. We compare the result based on two distinct prior models for the lithology/fluid classes. The first prior model is based on a first order neighborhood Markov random field in three dimensions, where both horizontal and vertical spatial coupling is assumed. The second prior model is based on a set of independent vertical first order Markov chains, where only vertical spatial coupling is assumed. We assess the posterior densities by a Markov chain Monte Carlo algorithm. The two models are demonstrated and compared on a gas discovery in the Norwegian Sea. A reduction in prediction error at a blind well location is obtained based on the Markov random field prior.
-
-
-
Prediction of Reservoir Properties Based on Seismic Inversion Technique of Full Stack 2D Seismic Data
Authors G. Akhmetzhanova, R. Abirov and A. ZheksenbayevaSummaryThe methodology for constructing a model for the inversion of the acoustic impedance data was carried out on the area of the field of the Caspian basin.
Inversion method allows us to move from seismic amplitudes to acoustic impedance and, if there are statistical relationships between the elastic parameters of the medium and the petrophysical characteristics of the reservoir, using regression equations, to predict the estimated parameters that can be used as trends in the construction of geological models
Based on the method of inverting the acoustic data of the studied field, the porosity was distributed over the well data and the acoustic impedance a linear relationship was established, which made it possible to recalculate the acoustic impedance sections into porosity sections and use the porosity obtained by the inversion over the area when constructing the geological model as a trend. The obtained sections of acoustic stiffness were used to distinguish reservoirs with improved properties at a high quality level.
-
-
-
Reservoir Characterization of Multi-Stage Valley Fill through the Use of Hit Cube Stochastic Inversion
Authors A. Mustaqeem and V. BaranovaSummaryA probabilistic model matching inversion method, called HITCUBE, is used in the reservoir characterization study. This stochastic workflow can be executed with poststack or prestack (partial angle stack, offset gather, AVO gradient) seismic data while matching the real seismic trace with the modeled synthetic trace (similarity, cross-correlation or amplitude spectrum) generated from an isotropic ray tracing method. The property traces from corresponding models with a correlation beyond threshold are stacked to build the output probability grids.
Based on rock physics analysis of existing well log data, the relationship of the elastic properties (Vp, Vs and Rho) of the target formation and the rock properties (lithology, porosity, water saturation) is built as a physical representative of the geology in the study area which is then used for pseudo-well generation. A number of pseudowells can be generated through Monte Carlo simulation referring to the rock physics analysis result, the geological feature of the study formation and the uncertainty.
This workflow is successfully applied in the Upper Mannville Group clastic reservoir characterization using a public seismic dataset with multiple wells. The seismic gather data are preconditioned with an AVO friendly workflow before the inversion. Optimized reservoir facies with better reservoir quality are characterized.
-
-
-
Anisotropy Analysis of Vaca Muerta Source Rocks and Multicomponent Seismic Inversion, Bandurria Norte Concession, Argentina
Authors D. Curia, U. Strecker and P. VeekenSummaryHydraulic well stimulation requires knowledge of fractures and rock elasticity as these parameters reduce uncertainty attached to shale oil development prospects. Multicomponent 3D-3C seismic data is input for more reliable estimation of rock physical parameters. This info is useful to optimise fracture stimulation of low permeability unconventional reservoirs in the Vaca Muerta formation.
Multicomponent seismics adds value to the anisotropy analysis by considering also shear wave splitting effects. Elastic properties (e.g. Young’s modulus) are retrieved from PP-PS joint inversion. The wide azimuth Bandurria Norte multicomponent survey was successfully acquired, processed and interpreted. Estimation of shear wave splitting effects improve accuracy of the velocities and the PreSTM imaging. Directional dependency of the seismic velocities is thought related to fracture distribution and local stress regime. The method allows to define development ‘sweet spots’ on the Vaca Muerta target level.
-
-
-
Constrained 4D Elastic Inversion, Assessing Workflows Efficiency on Real Data
Authors P. Thore, J. Jorge, J. Pereira and A. LucasSummaryWe perform elastic inversion of time lapse seismic. This is a very difficult task as it induces consistency over different acquisitions and multi angle stacks data. The time-lapse elastic properties can be obtained through several 4D inversion techniques, the purpose of this study is to assess three different workflows (inversion of the differences, uncoupled inversion and coupled inversion) that ultimately deliver reliable information capitalizing on petroelastic model constraints that ensure compatibility with prior geological and dynamical information. The comparison between each workflow demonstrates the key parameters in which the quality of results relies the most as well as the inherent issues related to each methodology. Despite the difference of the results of the different workflows this study show how much reliable information can be obtained from 4D seismic data.
-
-
-
The Use of Elastic Inversion as a Seismic QC-Tool
Authors H.P. Hansen and H. WagnerSummarySeismic elastic inversion is usually used to quantify rock type, porosity and fluid type for reservoir evaluation. And this is most often done during the interpretation stage of a seismic data set which is after the seismic processing is completed.
Because seismic inversion is such a strong integration process of many different data types (well logs, checkshot/VSP, interpreted horizons, seismic sub-stacks and seismic velocity information) where all these various data types need to play together, it is often discovered during the inversion set-up if the seismic data suffer from shortcomings that could have been rectified if discovered earlier.
If trustworthy well information is available, we have good experience in using the elastic inversion as a QC-tool and it should be engaged before the seismic processing project is completed. Inversion should be used with the seismic QC as the objective rather than estimation of final reservoir properties. It therefore mostly does not require a full cycle of inversion setup and QC which makes it fairly time efficient.
-
-
-
Sparse Layer Reflectivity with FISTA for Post-Stack Impedance Inversion
Authors J. Han, B. Russell, J. Downton and T. ColwellSummaryBy sparsely constraining the estimated reflectivity, sparse layer reflectivity inversion improves the resolution and the ability to detect thin layers in seismic data. Historically, the drawback of sparse layer reflectivity inversion is the computational cost. This paper improves the computational efficiency of the inversion by utilizing an optimal optimization tool, fast iterative shrinkage-thresholding algorithm (FISTA) that has proved useful for other hybrid norm problems. Besides improving the computational speed, FISTA keeps the high resolution of sparse layer reflectivity inversion to detect thin layers in seismic data. The method is applied to two separate datasets, both showing improved resolution and reservoir characterization. These results suggest that sparse layer reflectivity inversion with FISTA is a promising tool for reservoir characterization.
-
-
-
Full Wave and Down-Going Elastic Inversion of Deepwater OBN Seismic Data: Example from Niger Delta
Authors S. Amoyedo, E. Tawile, S. Pou-Palome, P. Kakaire, O. Olagundoye, J. Mascomere and E. OwahSummaryEgina Field, which is in water depth of about 1500 m, is located offshore Niger Delta. The reservoir consists of Miocene turbiditic channel complexes. High seismic energy loss in the reservoir zone, due to shallow gas effects and mud volcanoes, necessitated the acquisition of Ocean Bottom Nodes (OBN) survey in order to improve the imaging and reservoir characterization while serving as a baseline for future 4D monitor surveys.
In this paper, we discuss the FWI implementation on the OBN dataset and the added value over legacy narrow azimuth streamer seismic data. Additionally, we discuss the TTI anisotropic Kirchhoff Pre-Stack Depth Migration of the OBN Up-going and Down-going datasets.
Furthermore, we detail our post stack amplitude correction implemented on the dataset prior to the inversion process. This AVO-consistent correction allows for correction of energy loss in the reservoir section due to shallow events.
A model-based elastic inversion of the down-going wavefield yielded a robust and highly consistent elastic properties due in large part to the high-quality low frequency content and high signal-noise ratio. Computation of probability of success allows a more quantitative estimate of the degree of reliability of the inverted and derived properties compared to actual well log data.
-
-
-
Full Elastic Wave-Equation Based Inversion of Pre-Stack Seismic Data and its Differences to Linear AVO Inversion
Authors P. Haffinger, D. Gisolf and P. DoulgerisSummaryUtilising amplitude versus offset (AVO) information of seismic data can have a crucial business impact on projects concerned with the exploitation of resources in the subsurface. Key to success is how accurate pre-stack seismic gathers can be translated into elastic subsurface properties, or ultimately reservoir properties like porosity and saturation. Historically, intercept and gradient techniques and linearized inversions were frequently used to this purpose, while both rely on very simplified approximations of the true seismic wave-propagation. In this abstract, a fundamentally new approach towards inversion of seismic AVO information will be proposed. The technique iteratively solves the full elastic wave equation while computing the subsurface properties but also the seismic wavefield at every point in the inversion domain. In this way not only primaries but also interbed multiples, mode conversions and transmission effect are properly handled. Thereby the technique allows a much more physically realistic inversion of pre-stack seismic data while increasing the accuracy of the recovered subsurface information significantly. The method will be demonstrated on a synthetic dataset simulated in a carbonate setting, and the technical benefits compared to linearised AVO inversion will be discussed and highlighted.
-
-
-
The Impact of Seismic Inversion Methods on Facies Prediction
Authors P. Thore, H. Klemm and L. AzevedoSummaryWe present a simultaneous Bayesian inversion framework for prediction and uncertainty quantification of lithology/fluid classes, petrophysical properties and elastic attributes from seismic amplitude-versus-offset observations. Prediction of these reservoir variables are of interest to generate initial models of the reservoir to model fluid flow. We consider a Gauss-linear likelihood. We compare the result based on two distinct prior models for the lithology/fluid classes. The first prior model is based on a first order neighborhood Markov random field in three dimensions, where both horizontal and vertical spatial coupling is assumed. The second prior model is based on a set of independent vertical first order Markov chains, where only vertical spatial coupling is assumed. We assess the posterior densities by a Markov chain Monte Carlo algorithm. The two models are demonstrated and compared on a gas discovery in the Norwegian Sea. A reduction in prediction error at a blind well location is obtained based on the Markov random field prior.
-
-
-
Sequential Gaussian Conditioning
By A. CherrettSummaryWe propose a new algorithm, improving on existing SGS-type techniques in two main ways. Firstly, it propagates Gaussian uncertainties, rather than individual realisations, so can be used to compute posterior means and variances, if desired, but can also be used stochastically. Secondly, the distribution at each trace location is conditioned to all data, irrespective of where it is positioned in the sequential path. This is done with a two-pass strategy, using Bayes’ theorem to combine the constraints arising from both the backward and forward segments of the path.
-