- Home
- Conferences
- Conference Proceedings
- Conferences
81st EAGE Conference and Exhibition 2019
- Conference date: June 3-6, 2019
- Location: London, UK
- Published: 03 June 2019
21 - 40 of 1010 results
-
-
Estimation of the Conversion Point Position in Elastic Orthorhombic Media
More LessSummaryIn order to estimate the conversion point in an elastic orthorhombic (ORT) medium, we define an explicit rational form approximation for the radial coordinate of the conversion point for the converted PS1, PS2 and S1S2 waves. In order to obtain the approximation coefficients, the Taylor series approximation in the corresponding vertical slowness for three pure wave modes is applied. The coefficients in our proposed approximation are computed within two vertical symmetry planes. The difference between the acquisition azimuth and the azimuth of the conversion point position is analyzed for different combinations of the wave modes. The accuracy in conversion point position estimation for three ORT models is illustrated in the numerical examples. One can see from the results that for converted PS1 and PS2 waves, the proposed approximation is very accurate in estimating the conversion point position in the tested ORT model. For a converted S1S2 wave, due to the existence of cusps, triplications and shear singularities, the error in conversion point estimation is relatively larger comparing with PS waves in the vicinity of the singularity point.
-
-
-
A Parallel Fast Sweeping Algorithm for Fast and Accurate Traveltime Computation in Anisotropic Media
By U. WaheedSummaryNumerical solution of the eikonal equation has been widely used to compute traveltimes in anisotropic media. Several techniques have been recently proposed to increase the accuracy of these solutions. These include factoring the unknown traveltime to tackle the source-singularity problem, using the Weighted Essentially Non-oscillatory (WENO) approximation of the traveltime derivatives, and the Discontinuous Galerkin method. Although these approaches yield highly accurate traveltimes but they also result in increased computational load. Therefore, I propose a parallel fast sweeping algorithm to compute fast and accurate solution of the anisotropic eikonal equation. High accuracy is achieved by using factorization followed by the WENO approximation of derivatives, whereas computational speed up is obtained by sweeping the computational domain in parallel. With a large number of CPUs, significant reduction in computational cost can be achieved for large 3D models.
-
-
-
A TTI Representation of a Heterogeneous Medium for Finite-Difference Seismic Wave Simulation
More LessSummarySince the computational time of FDM is proportional to the number of grid points, many higher-order and optimized schemes were developed to increase the spatial grid size while maintaining the dispersion error below predefined levels. However, mentioned techniques are applicable only to homogeneous media since large error due to the interfaces may arise when the large grid spacing is used. To solve this problem, various effective medium parameterization methods have been proposed. In our work, We develop a new TTI effective media parameterization method and analyze existing different effective media parameterization methods. Our tests show that the proposed TTI effective media parameterization method is better than other effective media parameterization methods and can be used with the higher-order scheme to utilize 3–4 PPW to produce satisfactory results.spacing is used.
We develop a new TTI effective media parameterization method for
-
-
-
Two-Phase Computational Model for Wave Propagation in Deforming Saturated Porous Medium
Authors G. Reshetova and E. RomenskiSummaryThe new two-phase model for a compressible fluid flow in deforming porous media is presented. The derivation of the model is based on the symmetric hyperbolic thermodynamically compatible systems theory, which is developed with the use of the first principles and fundamental laws of thermodynamics. The governing PDEs form the first order hyperbolic system and can be used for studying a wide variety of processes in saturated porous media, including small amplitude wave propagation. Our theory predicts the three types of waves: fast and slow pressure waves and a shear wave, as it is in Biot's model. The material constants of the model are fully determined by the properties of the solid and the fluid phases, and unlike Biot's model do not contain empirical parameters. The governing PDEs for small amplitude wave propagation in a saturated porous medium are presented, and the efficient numerical method has been developed for solving these PDEs, based on the finite difference staggered-grid scheme of fourth order spatial accuracy with modified coefficients to provide approximation in inhomogeneous media. An application of the developed computational framework to solving a series of test problems confirms the robustness and efficiency of the approach presented.
-
-
-
A Full Wavefield Approach to Survey Planning
Authors C. Lazizi, S. Jetschny, M. Pedersen and A. OrdoñezSummaryTowed streamer acquisition using typical 3D spreads can be problematic for shallow imaging due to strong acquisition footprints effects. These effects can be mitigated by imaging using separated wavefields whereby each receiver position acts as virtual source. We present a full-wavefield modeling and imaging workflow for de-risking of imaging application using separated wavefields in the survey planning phase of a marine seismic acquisition. Each of the applications share a modeling and migration step that considers only a few shots that can be duplicated to represent the migration response of a full 3D acquisition. The workflow allows us to estimate the depth range where separated wavefield imaging provides superior results over primary imaging in terms of sail line related footprint artefacts. We can further analyze the effect of cable separation on separated wavefield imaging and demonstrate how wavefield resampling mitigates these effects. Based on a horizontally layered model we can test and optimize survey geometries by analyzing migration footprint along plane reflectors and the resolution of fault features of various sizes in the image domain. Such analysis can be used to optimize sail line and cable separation to provide sufficient coverage to image the targets of interest as efficiently as possible.
-
-
-
Speeding up a Mass-Lumped Tetrahedral Finite-Element Method for Wave Propagation
Authors W. Mulder, S. Geevers and J. Van der VegtSummaryMass-lumped finite elements on tetrahedra offer more flexibility than their counterpart on hexahedra for the simulation of seismic wave propagation, but there is no general recipe for their construction, unlike as with hexahedra. Earlier, we found new elements up to degree 4 that have significantly less nodes than previously known elements by sharpening the accuracy criterion. A similar approach applied to numerical quadrature of the stiffness matrix provides a speed improvement in the acoustic case and an additional factor 1.5 in the isotropic elastic case. We present numerical results for a homogeneous and heterogeneous isotropic elastic test problem on a sequence of successively finer meshes and for elements of degrees 1 to 4. A comparison of their accuracy and computational efficiency shows that a scheme of degree 4 has the best performance when high accuracy is desired, but the one of degree 3 is more efficient at intermediate accuracy.
-
-
-
Spatially-Optimized Finite-Difference Schemes for Numerical Dispersion Suppression: an Implementation Using Symbolic Computation
By E. CauntSummaryIn this work, a 4th order DRP 2D elastic wave formulation with free surface boundary conditions is presented, extending methods of Tam and Webb (1993 ). Staggered first-derivative stencils are derived and applied to the P-SV formulation of Virieux (1986 ). Performance is compared to the Taylor-series-derived staggered scheme of equal extent, demonstrating the versatility and universal benefits of spatial optimization. Implementation of both FD schemes is carried out using Devito, a domain-specific Python module and compiler for FD applications. Devito allows for model specification with a handful of high-level symbolic Python objects to build an FD operator, used to generate highly optimized C++ code at runtime via a series of intermediate representations, allowing for complex multi-stage optimizations. The high-level, symbolic nature of Devito ensures concise, readable model code and expedites workflow compared to model building with low-level languages, enabling rapid prototyping in hours as opposed to weeks or months without sacrificing underlying code quality. This work showcases the potential of symbolic computation for implementing non-conventional FD stencils, and the straightforwardness of doing so with Devito.
-
-
-
Sparsity of Synthetic Wave Fields in Curvelet Space
Authors M. Henriques, G. Corso, J. Medeiros and L. LucenaSummaryThe Full Waveform Inversion (FWI) is based on the optimization of a physical model in order to fit the generated synthetic data with the empirical data collected by the receivers used in the field. At each iteration of the FWI computational procedure, the wave equation is solved and an update direction of the model is computed. This process requires a large amount of memory, a critical step in the performance of the method. The compression of the wave field data can effectively be performed with the curvelet transform, a modern multi-scale tool. Because of their dependence on orientation, curvelets are suitable to represent the anisotropy of wave patterns. Besides, it has been demonstrated that the solution operators of a wide range of wave equations are optimally sparse. In this work, we explore curvelets for data compression in the seismic FWI context. We compare the memory use of standard FWI processing with similar FWI processing using curvelets decomposition.
-
-
-
Time-Lapse Elastic Properties of Cracked Granite During Deformation Inferred from Laboratory Experiments Using 2D Waveform Modelling
Authors S. Lai, N. Fuji and I. KatayamaSummaryIn order to understand the physics of earthquake generation, we study the microscopic weakening process in a centimetre-scale laboratory experiment. A series of deformation experiment was performed to the granite sample from Aji region, Japan. During the experiments, active seismic wave propagation was measured. This study numerically models the observed waveform in order to infer to the time-lapse elastic structure changes. We first modify the finite difference code on collocated Cartesian introducing zig-zag free-surface discontinuities. We applied response function estimation to the synthetic waveform to consider all the effect (i.e. 3-D effect, instrumental response, attenuation). We use trial-and-error method to find the best-fit velocity models using full waveform synthetics with response. Then we use the same method to make a series of velocity models for the automated Monte Carlo scheme. Comparing automated velocity models using different methods, we find that L1 norm is the best fitting method. We further tested fault models that we find the trend of the changes of the velocity model after the creation and during the thickening of the fault. This method will detail a microscopic weakening process during the nucleation of earthquakes using full waveforms, which have an impact on fundamental understanding of earthquake physics.
-
-
-
Physical Modeling and Seismic Response Characteristics Analysis of Strike-Slip Faults
More LessSummaryIn the northwest exploration area of Sinopec, a series of strike-slip faults are formed due to the multi-stage structure and stress action. These faults are closely related to the formation of oil and gas reservoirs, so the accurate identification of these faults is of great significance to the discovery of the gas and oil reservoirs. Through the development of heterogeneous physical model materials, strike-slip fault physical models with different widths and different degrees of rupture are constructed; seismic data acquisition and response characteristics analysis are also carried out. The results show that the amplitude of the fault reflection is the comprehensive response of fault scale and internal filling characteristics. By summarizing the geological characteristics, a physical model of three typical strike-slip faults forms with positive, negative, and graben are constructed. The data analysis results show that it is difficult to accurately depict the fault by seismic response or single attribute for different strike-slip faults forms, but the low frequency data can obviously improve the ability of recognition.
-
-
-
Optimized and Cost-Efficient Visco-Acoustic Iterative Least-Squares Migration
Authors Ø. Korsmo, A. Osen, A. Pankov and O.A. SandstadSummaryIterative least-squares migration is an expensive process that requires several passes of migration and de-migration. In this paper we focus on how an optimized initial reflectivity model combined with a deconvolution imaging condition can ensure faster convergence. We also incorporate the visco-acoustic effects in the de-blurring process to improve image corrections across and below Q-anomalies. This workflow is demonstrated on a dual-azimuth North Sea dataset, where the aim is to improve the understanding of the Frosk and Bøyla fields, which are characterized by complex and steeply dipping sand systems and areas of weak reflectivity.
-
-
-
Reverse Time Migration with Random Space Shift
More LessSummaryThe classical imaging condition for reverse time migration (RTM) is to cross-correlate the forward propagated source-side wavefield with the backward propagated receiver-side wavefield at zero time- and spatial-lag, under the assumption that an accurate migration velocity model is available to correctly predict the kinematic information of the observed seismic data. When errors are present in the migration velocity model, the conventional RTM image may be defocused. Inspired by the work of the extended RTM with subsurface offset and considering the finite-frequency effects of wave propagation, we propose to introduce a random space shift to the source- and receiver-side wavefield at each time step before applying the cross-correlation imaging condition. The maximum random spatial shift is constrained by the theoretical lateral resolution limit of the half-wavelength after migration. The computational cost of our method is the same as the conventional RTM. Numerical examples on a 2D layered model demonstrate that our method could provide an image with better quality than the conventional RTM even the migration velocity model contains significant errors.
-
-
-
Flank-Preservation De-Primary Reverse Time Migration, Part I: The Missing Flanks and a Modified Imaging Condition
Authors Y. Sun, T. Fei and R. Van BorselenSummaryDe-Primary Reverse Time Migration (dpRTM) is a recently proposed technology, which has the capability to remove incorrectly imaged structures caused by inaccuracies or complexities of the migration velocity model used in RTM. In spite of the success achieved by dpRTM, a bizarre effect is also observed, that is after dpRTM the flank structures in the image have been completely missing. We first point out the root cause leading to the missing flank structures in dpRTM, and next propose a Flank-Preservation De-Primary Reverse Time Migration (FPdpRTM) technology, which not only possesses the capability of dpRTM on removing false structures but also is capable of preserving flank structures in the final image. We use two synthetic examples to demonstrate the effectiveness of FPdpRTM using the true velocity models for migration.
-
-
-
Imaging Enhancement of Reverse Time Migration Using Dynamic Time Warping and Local Similarity
More LessSummaryConventional reverse time migration obtains the final image via simply stacking single-shot imaging result. Many factors such as the inaccurate migration velocity, the mad physics assumption and the imperfection sampling, cause the misalignment, amplitude difference and artifacts in different trace on the shot-index gathers. Mean stacking reduces the imaging quality. In the abstract, we propose a weighted stacking to improve the imaging quality. First, we apply the dynamic time warping method to correct the uneven shot gathers. The dynamic time warping is a nonlinear aligning method that effectively correct the residual misalignment on the shot gathers. Then, we use the local similarity to construct the weighted coefficient on the foundation of the flat shot gathers. The local similarity method can better estimate the correlation between different traces. Finally, we present results on synthetic and real data to illustrate our approach is valid.
-
-
-
A Stereotomography Velocity Inversion Method in Imaging Domain
More LessSummaryGenerally, the data space needed for velocity inversion of stereotomographic is based on the slant stacking of local events on common shot point gathers and common detection point gathers in the data domain. This is a stable but inefficient data space pickup method suitable for 2 dimensional seismic data. Using pre-stack gathers for slant stacking is effective for primary wave in interactive picking, but the diffracted wave in complex structures can not be distinguished. Therefore, in view of the problem of inaccurate data space picking and large workload, this paper chooses a method of obtaining accurate stereotomography data space in the imaging domain under the condition of incorrect velocity. Firstly, extracting the common imaging point gathers by DSR (double square root equation) prestack depth migration method, which includes angle information of underground ray in migration process. Then residual curvature and local half-offset ray parameters are obtained. Then travel time residuals and corresponding ray parameters at shot point and receiving point are obtained. Then, real stereo is calculated by specific correction formula. Model tests shows that stereotomographic velocity inversion can improve the accuracy and stability obviously. This method will have better application in the practice of stereotomography inversion in the future.
-
-
-
Visco-Acoustic Reverse Time Migration Using Lowrank Decomposition
More LessSummaryFor visco-acoustic reverse time migration, low-wavenumber noise removal, precision of extrapolation and stability of backward extrapolation are three important problems. In this paper, we propose an effective image condition for QRTM to suppress the low-wavenumber noise, which utilizes Hilbert transform to procure the upgoing and downgoing wavefield and sequentially uses normalized crosscorrelation to obtain the image value. And for the other two parts, we use lowrank decomposition and adaptive stabilization filter to guarantee the precision of extrapolation and the stability of backward extrapolation. Numerical experiments on a two-layered model and BP gas chimney model can demonstrate that our method can provide accurate and stable image results.
-
-
-
Absolute Acoustic Impedance Inversion Based on Reflectivity Estimated by Image Domain Least Squares Migration
More LessSummaryInstead of solving nonlinear FWI, the acoustic impedance (AI) model can be divided into two parts for linear inversion. One is the smooth background, and the other is the reflectivity.
Assuming the background AI model is obtained, true-amplitude and broadband reflectivity can be estimated by solving least-squares migration (LSM). We can calculate the corresponding broadband AI, and then the background AI is merged to get the absolute AI.
In this paper, we firstly solve the image domain LSM (IDLSM) for true-amplitude and broadband reflectivity inversion. Then the absolute AI model is inverted with the estimated reflectivity and the given the background AI. For IDLSM, we approximate the Hessian matrix with non-stationary matching filters. The image deblurring problem is solved with the sparse (L1 norm) and total variation (TV) regularization. With the estimated reflectivity and given background impedance, we solve a constrained optimization problem with combined first and second order TV regularizations for absolute AI model inversion. Numerical examples on Marmousi model illustrate that the IDLSM results solved by the proposed method have both more balanced amplitudes and higher resolution than conventional migration images. Absolute AI model can be inverted with the estimated reflectivity and given background impedance.
-
-
-
Travel Time Changes in the Groningen Gas Reservoir by Train Noise Interferometry of Borehole Data
Authors W. Zhou and H. PaulssenSummaryIn this study we show that time-lapse measurements of a reservoir can be made by noise interferometry of borehole data. We used borehole geophone array data of monitoring well SDM-1 in the Groningen gas field in the Netherlands. The ambient, anthropogenic noise allows accurate determination of the P and S velocity structure ( Zhou & Paulssen, 2017 ), but is not stable enough in time and space to measure temporal travel time variations. With deconvolution interferometry of high-frequency train signals, however, it is possible to detect small travel time decreases of ~0.05 ms (0.1%) over half a year associated with compaction of the reservoir. Moreover, we identified a strong travel time anomaly over a period of 1.5 months that is caused by drilling of a new borehole in the reservoir at ~5 km distance.
-
-
-
A Novel Array Acquisition and Processing Methodology for Microseismic Monitoring
Authors B. Witten, A. Booterbaugh and R. SegstroSummarySurface microseismic monitoring has converged on a few acquisition designs, notably the “star” and “patch” arrays, to detect and locate low signal-to-noise ratio (SNR) microseismic events. The principal ideas behind both these designs are the same: to be capable of attenuating surface noise and harnessing the power of stacking a large number of sensors to increase the SNR. We present an alternative acquisition design, consisting of hexagonal subarrays, and noise attenuation methodology that is easily scalable, adaptable to local site conditions, and cost effective. We propose this array design as it is easy to deploy in any environment since it does not require lines and each subarray has a very small footprint. In addition, we show that the hexagonal array is capable of removing coherent noise even with a limited number number of sensors through a semblance-weighted stack which attenuates surface wave noise while preserving the signal energy that arrives vertically. We demonstrate this capability through a pilot field test that co-located hexagonal arrays within a patch acquisition.
-
-
-
Selecting a Seismically Safe Formations for Salt Water Disposal in Delaware Basin
Authors Z. Jechumtalova, L. Eisner and D. AnikievSummaryThe recent increased level of the observed seismicity in Delaware Basin, USA, raises an urgent question of how we can safely dispose polluted water (salt water disposal - SWD). There are many active injection wells within this basin and water is injected into many separated formations. We applied a waveform similarity technique to data from regional stations to detect weaker events than those reported by USGS during years 2013-2017. We were able to detect 20 times more events than USGS. We have performed spatio-temporal correlation between this extended catalogue and SWD injected volumes for various injection formations and distances of events from injectors. Four formations targeted for SWD correlated with seismicity, injection rates into two of these formations correlates with the seismicity with less than one month delay between the injection and seismicity and two of these formations show significant delay of seismicity behind the injection (4 and 10 months) which may indicate indirect connection between SWD and seismicity.
-