- Home
- Conferences
- Conference Proceedings
- Conferences
76th EAGE Conference and Exhibition 2014
- Conference date: June 16-19, 2014
- Location: Amsterdam, Netherlands
- Published: 16 June 2014
81 - 100 of 1028 results
-
-
Full Waveform Inversion Using the Energy Objective Function
More LessSummaryWe suggest new objective functions for full waveform inversion in the time domain. These objective functions minimize the energy differences between observed and modeled data. We applied sequential time windows to the data and calculated the energy. A numerical example shows that the objective functions can be used to obtain a macrovelocity model. The recovered macro-velocity model can be used as an initial velocity for a subsequent inversion in the frequency domain to obtain high-resolution velocity information. Since we calculate the partial derivative of the energy objective function explicitly, heavy computational requirement is an important limitation of these objective functions.
-
-
-
Two-dimensional Unwrapped Phase Inversion with Damping and a Gaussian Filter
Authors Y. Choi and T. AlkhalifahSummaryPhase wrapping is one of main causes of the local minima problem in waveform inversion. However, the unwrapping process for 2D phase maps that includes singular points (residues) is complicated and does not guarantee unique solutions. We employ an exponential damping to eliminate the residues in the 2D phase maps, which makes the 2D phase unwrapping process easy and produce a unique solution. A recursive inversion process using the damped unwrapped phase provides an opportunity to invert for smooth background updates first, and higher resolution updates later as we reduce the damping. We also apply a Gaussian filter to the gradient to mitigate the edge artifacts resulting from the narrow shape of the sensitivity kernels at high damping. Numerical examples demonstrate that our unwrapped phase inversion with damping and a Gaussian filter produces good convergent results even for a 3Hz single frequency of Marmousi dataset and with a starting model far from the true model.
-
-
-
A New Take on FWI - Wavefield Reconstruction Inversion
Authors T. van Leeuwen, F.J. Herrmann and B. PetersSummaryWe discuss a recently proposed novel method for waveform inversion: Wavefield Reconstruction Inversion (WRI). As opposed to conventional FWI – which attempts to minimize the error between observed and predicted data obtained by solving a wave equation – WRI reconstructs a wave-field from the data and extracts a model-update from this wavefield by minimizing the wave-equation residual. The method does not require explicit computation of an adjoint wavefield as all the necessary information is contained in the reconstructed wavefield. We show how the corresponding model updates can be interpreted physically analogously to the conventional imaging-condition-based approach.
-
-
-
Wave-equation Based Inversion with the Penalty Method - Adjoint-state Versus Wavefield-reconstruction Inversion
Authors B. Peters, F.J. Herrmann and T. van LeeuwenSummaryIn this paper we make a comparison between wave-equation based inversions based on the adjoint-state and penalty methods. While the adjoint-state method involves the minimization of a data-misfit and exact solutions of the wave-equation for the current velocity model, the penalty-method aims to first find a wavefield that jointly fits the data and honours the physics, in a least-squares sense. Given this reconstructed wavefield, which is a proxy for the true wavefield in the true model, we calculate updates for the velocity model. Aside from being less nonlinear--the acoustic wave equation is linear in the wavefield and model parameters but not in both--the inversion is carried out over a solution space that includes both the model and the wavefield. This larger search space allows the algorithm to circumnavigate local minima, very much in the same way as recently proposed model extensions try to accomplish. We include examples for low frequencies, where we compare full-waveform inversion results for both methods, for good and bad starting models, and for high frequencies where we compare reverse-time migration with linearized imaging based on wavefield-reconstruction inversion. The examples confirm the expected benefits of the proposed method.
-
-
-
Explicit High-contrast Surface Refinement Using Full Waveform Inversion
Authors J.W.D. Hobro and J.E. RickettSummaryFull waveform inversion (FWI) provides a powerful mechanism for resolving high-resolution structure within complex seismic velocity models. However, many models used in FWI contain complex bodies (e. g. salt intrusions) that generate a large, sharp contrast in seismic properties. Refining these with FWI can prove extremely difficult when they are represented implicitly within a conventional gridded velocity model. The constraints required to optimize convergence at large contrasts are different from those required elsewhere. It is difficult to allow for this when the surfaces have no explicit representation. This paper describes a new approach to refining complex 3-D surfaces directly using FWI. An explicit representation of the surface is updated during the inversion. The gradient required is obtained by projecting a conventional FWI gradient (for seismic parameters) to a position and shape parameter gradient for the 3-D surface. The method relies upon accurate gridded model generation in the vicinity of the surfaces and a linearization of this process to enable the gradient projection. We present 2-D and 3-D synthetic examples that resolve complex structure within a Gulf of Mexico top salt contrast. The examples demonstrate very rapid and stable convergence that resolves complex features beyond the resolution of the finite-difference grid.
-
-
-
Full Waveform Inversion Based on the Optimized Gradient and Its Spectral Implementation
Authors Z. Wu and T. AlkhalifahSummaryFull waveform inversion (FWI) despite it’s potential suffers from the ability to converge to the desired solution due to the high nonlinearity of the objective function at conventional seismic frequencies. Even if frequencies necessary for the convergence are available, the high number of iterations required to approach a solution renders FWI as very expensive (especially in 3D). A spectral implementation in which the wavefields are extrapolated and gradients are calculated in the wavenumber domain allows for a cleaner more efficient implementation (no finite difference dispersion errors).
In addition, we use not only an up and down going wavefield decomposition of the gradient to access the smooth background update, but also a right and left propagation decomposition to allow us to do that for large dips. To insure that the extracted smooth component of the gradient has the right decent direction, we solve an optimization problem to search for the smoothest component that provides a negative (decent) gradient. Application to the Marmousi model shows that this approach works well with linear increasing initial velocity model and data with frequencies above 2Hz.
-
-
-
Full Wave Inversion Using a Spectral-Element Discontinuous Galerkin Method
Authors J.R. Krebs, S.S. Collis, N.J. Downey, C.C. Ober, J.R. Overfelt, T.M. Smith, B.G. van Bloemen-Waanders and J.G. YoungSummaryWe have developed a flexible Discontinuous Galerkin (DG) toolkit for full-wave inversion (FWI) that operates on unstructured non-affine meshes using a variety of element types (quadrilateral, triangular, hexahedral). The code handles spatially-variable polynomial-order across the mesh, and includes two approaches: modal DG with exact adjoints and gradients, and spectral DG which, though computationally faster, has approximate adjoints and gradients. In this paper, we show that high-quality full wave inversion results are obtained using both the modal DG and the approximate spectral DG approaches. A 3D inversion of field data using spectral-element DG will be presented in the talk.
-
-
-
Source Wavelet Estimation in Full Waveform Inversion
More LessSummaryIn full-waveform inversion (FWI), a small disturbance in the source wavelet may lead to large discrepancies in the inverted model, which becomes larger as the depth increases due to error accumulation. Hence, accurate source wavelet estimation becomes crucial in a successful inversion. On the other hand, an inaccurate model would jeopardize a wavelet estimation based on both simulated and observed data without proper constraints, which may in turn lead to wrong model updates and finally hazard the inversion.
To resolve this inversion dilemma, we propose a shallow-response based variable projection type of strategy to estimate the source wavelet alongside model parameters during FWI. Our approach embeds wavelet estimation naturally into FWI iteration as a standard variable reduction step, and restricts the computation to parts of the data that mainly consist of responses to shallow parts of a model. As the model becomes more and more accurate, one could always use more and more data, which might increase the robustness of wavelet estimation. To demonstrate the feasibility and robustness of our approach, we present inversion experiments with both synthetic and real data-sets, which suggest that shallow responses suffice to yield robust wavelet estimation that facilitates FWI.
-
-
-
Scattering Angle Base Filtering of the Inversion Gradients
More LessSummaryFull waveform inversion (FWI) requires a hierarchical approach based on the availability of low frequencies to maneuver the complex nonlinearity associated with the problem of velocity inversion. I develop a model gradient filter to help us access the parts of the gradient more suitable to combat this potential nonlinearity. The filter is based on representing the gradient in the time-lag normalized domain, in which low scattering angles of the gradient update are initially muted. The result are long-wavelength updates controlled by the ray component of the wavefield. In this case, even 10 Hz data can produce near zero wavelength updates suitable for a background correction of the model. Allowing smaller scattering angle to contribute provides higher resolution information to the model.
-
-
-
Low Frequency De-Noising and De-Ghosting Marine Full Waveform Inversion
Authors P.P. Milcik, R.E. Plessix, K.H. Matson and V.L. GohSummaryIn this paper we shall present a number of processing steps that can be applied to improve the low-frequency signal-to-noise ratio such that the effects of the ghosts are taken into consideration in a consistent manner between the observed and modeled data. The discussion will include examples from wide azimuth marine streamer data (WAZ) and ocean bottom node (OBN) data sets and the impact these steps have on the FWI results.
-
-
-
Full-waveform Inversion of Conventional Vibroseis Data Using Preconditioning Focused on Low Frequency Enhancement
Authors A. Adamczyk, M. Malinowski and A. GórszczykSummaryDespite the popularity that full-waveform inversion (FWI) has gained in recent years, its application to Vibroseis data is still challenging, due to the problematic low-frequency content. We present the results of acoustic frequency-domain FWI of Vibroseis data with sweep starting at 6 Hz, acquired with standard 10 Hz geophones in 2010 in south-east Poland. 4.5 Hz matching filter and curvelet denoising in the frequency domain are used to enhance the low-frequency content of the data. This, together with dense sampling of the frequencies in the first inversion group, resulted in a geologically plausible P-wave velocity model, which correctly reproduces the data, including the far-offset arrivals and wide-angle reflections.
-
-
-
Effective Cycle Skipping Reduction through Adaptive Data Selection for Full Waveform Inversion
More LessSummaryFull waveform inversion (FWI) has proven that it has potential to provide high-resolution velocity parameters. However, for most cases, FWI suffers from an objective function with a local minimum instead of a global minimum due to cycle skipping between real data and predicted data. To avoid this issue, researchers have proposed an FWI work flow that uses offset stripping and inner mutes to limit the input for FWI to near offsets for the initial inversion iteration, and then gradually incorporates farther offsets in subsequent iterations as the velocity model accuracy improves with depth. However, this work flow is computationally expensive and cannot effectively avoid the cycle skipping issue. We propose a data selection algorithm that assures all input data for FWI is within a half-cycle difference compared with the predicted data. This data selection process is implemented in each iteration of the inversion to generate a velocity model with higher accuracy and fewer artefacts.
-
-
-
A Probabilistic QC for Cycle-skipping in Full Waveform Inversion
Authors A.M.S. Martinez-Sansigre and A.R. RatcliffeSummaryFull waveform inversion (FWI) is now well established in the toolbox of velocity model building. That said, the cycle-skipping problem remains one of the primary practical limitations of FWI, especially in industry applications. A simple, effective and robust QC would be of benefit to determining the suitability of any given velocity model as a good starting point for FWI. Here we present a QC toward this goal based on a probabilistic interpretation of cycle-skipping. We discuss the theory and rationale behind our idea and demonstrate its effectiveness in a controlled experiment. We then show an application to a real data example from the Central North Sea, where it was able to distinguish between starting velocity models that were free from cycle-skipping and lead to an improved FWI update, and those that were cycle-skipped and gave an FWI update that degraded the migrated image.
-
-
-
Statistical Sampling Enabled Full Waveform Inversion
Authors K. Jiao, A. Schiemenz and R. CoatesSummaryFull Waveform Inversion (FWI) has recently emerged as a promising method for refining seismic velocity models to achieve enhanced imaging. The algorithm involves iteratively updating the velocity model to improve the match between the recorded seismic data and the simulated waveforms. Each iteration typically requires multiple wavefield extrapolations. As a result the technique places significant computational burdens on even the largest computers when applied to large three-dimensional surface seismic datasets.
This paper discusses the application of two statistical sampling strategies to a time-domain FWI algorithm, with the aim of minimizing the computation costs while still ensuring that all the information in the data is utilized. Results are shown for a synthetic model and for a real data set acquired with a multi-vessel coil geometry, both of which show significant computational savings.
-
-
-
Fast Uncertainty Quantification for 2D Full-waveform Inversion with Randomized Source Subsampling
Authors Z. Fang, F.J. Herrmann and C.D. SilvaSummaryUncertainties arise in every area of seismic exploration, especially in full-waveform inversion, which is highly non-linear. In the framework of Bayesian inference, uncertainties can be analyzed by sampling the posterior probability density distribution with a Markov chain Monte Carlo (McMC) method. We reduce the cost of computing the posterior distribution by working with randomized subsets of sources. These approximations, together with the Gaussian assumption and approximation of the Hessian, leads to a computational tractable uncertainty quantification. Application of this approach to a synthetic leads to standard deviations and confidence intervals that are qualitatively consistent with our expectations.
-
-
-
Temporal Variation in Subsurface Stress Estimated from Seismic Scattering
Authors K. Okamoto, H. Mikada, T. Goto and J. TakekawaSummaryWe focus on seismically scattered waves that bring the information of the earth crust where the seismic waves travel through to estimate stress field in the subsurface. Since seismic scattering is strongly related to the crustal inhomogeneities such as faults, cracks, etc., which are also created by the stress in the crust, we could be able to estimate stress field in the deep subsurface using the seismic scattering.
We employ coda-Q (Qc) as a parameter to detect the variation of stress field. Qc is a parameter reflecting the inhomogeneities.
Here we hypothesize that Qc could be used to estimate regional-scale stress accumulation in the crust without local and shallow disturbances since Qc is estimated from the scattered seismic waves travelling over a wide range of the crust. We first obtain a relationship between Qc and the stress change using numerical simulations. We calculate the static stress changes associated with the Iwate-Miyagi Nairiku earthquake in 2008 (Mw 6.9) in the subsurface using earthquake dislocation model and the surface deformation for comparison. As the result, it is found that the stress change inferred from Qc shows the similarity of stress change in the deep subsurface calculated by the earthquake dislocation model.
-
-
-
Passive Seismic Monitoring at the Ketzin CCS Site - Magnitude Estimation
Authors B.F. Paap and T.P.H. SteeghsSummaryIn order to allow quantification of the strength of local micro-seismic events recorded at the CCS pilot site in Ketzin in terms of local magnitude, earthquake data recorded by standardized seismometers were used. Earthquakes were selected that occurred in Poland and Czech Republic and that were detected both on the Ketzin array and on nearby situated regional seismometers operated by BGR. Analysis of the waveforms of both types of data suggests that they could be combined to broaden the useful spectral bandwidth for earthquake waveform analysis. By identifying and isolating the overlapping frequency spectrum of both data types (1–3 Hz), linear relations were obtained that were used to define local magnitude for a given peak amplitude value. Extrapolation of these relations to smaller local magnitudes can serve as an estimate for local magnitudes of local high frequency micro-seismic events that are detected at the Ketzin site.
-
-
-
Observation and Signatures of Injection-induced Repeating Earthquake Sequences
Authors J. Kummerow, C. Dinske, H. Asanuma and M. HäringSummaryWe detect and analyze repeating earthquake sequences in the Basel 1 microseismic data set. Using a combination of waveform similarity and estimates of source radius we identify 144 repeating earthquake sequences with a total of 422 events. The specific spatio-temporal behaviour of the repeating events indicates clearly that they are more sensitive to pore pressure changes than the total induced seismicity, which is at larger distances (more than about 250–300m from the injection source in the Basel reservoir) not only affected by pore pressure variations, but presumably also by stress transfer from the larger (Mw>2.0) events. The results suggest that repeating earthquakes may contribute to identifying and differentiating the dominant triggering effects of recorded microseismicity.
Furthermore, this study shows that a significant percentage of the fractures stimulated in the Basel reservoir ruptured repeatedly: More than 15% of all located events are repeaters. We anticipate that the occurrence of repeating events is a common feature in injection induced microseismic data sets, and that it is useful to incorporate repeating events in future statistical analyses of microseismic data, which so far generally assume that each potential source in the stimulated reservoir can only fail one-time.
-
-
-
Rupture Propagation Imaging at Microseismic Scale at the Basel EGS Project
Authors J. Folesky, J. Kummerow, S.A. Shapiro, M. Häring and H. AsanumaSummaryWe further develop the Back Projection technique for tracing the rupture propagation of microseismic events. We lay out the basic idea of Back Projection imaging and show the results for three synthetic datasets obtained using finite-difference modeling. The synthetic rupture models in use where generated according to microseismic events that occurred at the Basel EGS. They help us to understand the influences of the station geometry and the station weighting process which need to be applied in Back Projection imaging. The focus of the work lies in the analysis of real events and the extraction of their respective properties. We show the corresponding results for the four largest real events of local magnitudes M=3.1–3.4 from the Basel EGS site and discuss the validity and interpretation of the outcome. We find that the obtained rupture dimensions are consistent with the independent magnitude derived estimates. The rupture directions which are obtained fit reasonably well to the shape of the microseismic cloud and to one of the respective fault planes obtained from source mechanism analysis. This supports the validity of our approach. In addition we are hereby capable of solving the fault plane ambiguity.
-
-
-
Seismic Efficiency and Overshoot of Fractures Associated with Stimulation in Heavy Oil Reservoirs
Authors L.N. Meighan, T.I. Urbancic and A.M. BaigSummaryTo better understand the different fracturing process of two passive seismic datasets collected over a 30 month period for two heavy oil reservoirs, utilizing steam injection and water flooding, we investigate radiated energy and seismic efficiency. We calculate the Savage-Wood Efficiency (ŋsw) and overshoot of microseismic events recorded in two reservoirs with different stimulations and find under similar geological conditions and time period, the seismic efficiency, for steam injection compared to water flooding are unique across the datasets implying that temperature and fluid pressure has a different impact on the fracturing dynamics. Reservoir A (stream injection), has 4069 events (Mw = −2.2 top −0.2) with low efficiency (ŋsw=0.01 to 0.48) and considered in overshoot (ε=0.1 to 0.5). Reservoir B (water flooding), has, 1763 events, but higher magnitude range (Mw = −1.4 to 1.8); with a combination of low efficiency (overshoot ε=0.002 to 0.49) and high efficiency (undershoot ε=−415 to 0.49) events. Overshoot occurs when the dynamic strength of the fault is relatively higher than the final stress acting on the fault and less energy is radiated; where undershoot occurs when the dynamic strength is relatively lower than the final stress on the fault and in a state of dynamic weakening.
-