SINBAD Consortium Meeting - 2012 Fall - Abstracts

Ning Tu, "Fast imaging with multiples"
If correctly used, multiple energy can be mapped to the correct subsurface locations. However, simply applying the cross-correlation imaging condition will introduce non-causal artifacts into the final image. Here we propose an inversion approach to image primaries and multiples simultaneously that yields an artifact-free image. To address the high computational cost associated with inversion, we propose to: i) have the wave-equation solver carry out the multi-dimensional convolutions implicitly, and ii) reduce the number of PDE solves by randomized...

Art Petrenko, "{CARP-CG}: a computational study"
Forward modelling of the wave equation is a key ingredient in seismic full waveform inversion (FWI). Simulation in the time domain and solution of the wave equation in the frequency domain are two competing approaches to modelling. Frequency domain approaches can further be categorized as using either direct or iterative solvers. For 3D FWI, iterative solvers in the frequency domain are attractive, partly because they require less memory than the other methods. This is due to the fact that there is no need to store the wavefield at each...

Rajiv Kumar, "Seismic data interpolation using {SVD} free {Pareto} curve based low rank optimization"
Seismic data acquisition is cursed by missing data caused by physical and/or budget constraints. Aim of interpolation technique is to spatially transform irregularly acquired data to regularly sampled data while maintaining the events coherency. While transform-domain sparsity promotion has proven to be an effective tool to solve this recovery problem, recent developments in Rank penalizing techniques opens new horizon to improved...

Andrew J. Calvert, Ian Hanlon, Mostafa Javanmehri, Rajiv Kumar, Tristan van Leeuwen, Xiang Li, Brendan R. Smithyman, Eric Takam Takougang, Haneet Wason, Felix J. Herrmann, "Our findings on the {Chevron} benchmark dataset"
During this presentation, we will review our findings working with...

Okan Akalin, "Large scale seismic data interpolation with matrix completion"
Seismic surveys amass large and incomplete data sets, and designing algorithms to interpolate the missing data at very large scales poses a daunting and critical challenge. We study how to apply scalable matrix completion methods to such interpolation problems. Recent studies in matrix completion have shown that a matrix that has low rank can be exactly completed when only a small number of observations are available. However, there are two challenges to applying matrix completion to...

Aleksandr Y. Aravkin, "Estimating nuisance parameters in inverse problems"
Many inverse problems include nuisance parameters which, while not of direct interest, are required to recover primary parameters. In this talk, we present the idea of "projecting out" these variables, and how this idea allows us to design methods for solving a broad class of problems with nuisance parameters, such as variance or degrees of freedom. We then discuss several geophysical applications, including including estimation of unknown variance parameters in the Gaussian model for...

Aleksandr Y. Aravkin, "Generalized {SPGL1}: from theory to applications"
The SPGL1 solver has been effectively used for many geophysical applications, including curvelet data interpolation, imaging, and as a subroutine in full waveform inversion. In this talk, we present an overview of the theoretical foundation of the solver, along with a broad generalization of this foundation. We then introduce several applications, including robust & sparse imaging, sparse deconvolution, and data interpolation by matrix completion.

Enrico Au-Yeung, Hassan Mansour, Ozgur Yilmaz, "Compressed sensing, random {Fourier} matrix and jitter sampling"
Compressed sensing is an emerging signal processing technique that allows signals to be sampled well below the Nyquist rate, when the signal has a sparse representation in an orthonormal basis. By using a random Fourier matrix or a Gaussian matrix as our measurement matrix, we can reconstruct a signal from far fewer measurements than required by Shannon sampling theorem. In this...

Curt Da Silva, "Hierarchical {Tucker} tensor optimization - applications to {4D} seismic data interpolation"
There has been a swell of research in the scientific computing community in the last couple of years which tries to extend notions of linear algebra (rank, the SVD, linear systems, etc.) to higher dimensional arrays, or tensors. Much work has been proposed to try to overcome the so called "curse of dimensionality", the O(N^d) storage required for a d-dimensional array, where N is the size of each dimension. The...

Michael P. Friedlander, "Randomized sampling: {How} confident are you?"
At last year's consortium meeting, I described an inexact gradient method and sampling scheme for data fitting. The randomization method has good convergence properties, at least as measured by the distance to the solution––in expectation. But as one insightful critic rightly pointed out, we don't usually observe the expectation, at least not in a single run. In this talk I will characterize the convergence of the method in terms of bounds on the probability of being too far away from the...

Navid Ghadermarzy, "Non-convex compressed sensing using partial support information"
In this talk, we will address the recovery conditions of weighted ℓp minimization for signal reconstruction from compressed sensing measurements when (possibly inaccurate) partial support information is available. First we will motivate the use of (weighted) ℓp minimization with p<1 and point out its advantages over weighted ℓ1 minimization when there is prior information on the support of the signal that is possibly partial and inaccurate. Then...

Brock Hargreaves, "Analysis versus synthesis in weighted sparse recovery"
The synthesis model for compressive sensing has been the model of choice for many years and various weighting schemes have been shown to improve it's performance (see Yilmaz, Mansour, and Ghadermarzy talks). However, there is a counterpart model to synthesis, namely the analysis model, which has been less popular but recently attracted more attention (see Lin's talk). In this talk, weighting in the analysis model is discussed and applied to the seismic trace interpolation problem.

Felix J. Herrmann, "Fast sparsity-promoting imaging with message passing"
To meet current-day challenges, exploration seismology increasingly relies on more and more sophisticated algorithms that require multiple paths through all data. This requirement leads to problems because the size of seismic data volumes is increasing exponentially, exposing bottlenecks in IO and computational capability. To overcome these bottlenecks, we follow recent trends in machine learning and compressive sensing by proposing a sparsity-promoting inversion technique that works on small...

Nathan Krislock, "Wireless sensor network localization"
Locating the position of sensors connected together in a wireless network given only the position of a small number of the sensors and estimates of some of the distances between the sensors is a difficult problem with many modern applications. Within the last few years research in wireless sensor network localization has greatly increased due to the many new applications using wireless sensors, from lightweight sensors used to monitor the environment to ocean-bottom sensors used in geophysical applications. A second reason...

Tristan van Leeuwen, "{3D} frequency-domain waveform inversion using a row-projected {Helmholtz} solver"
3D frequency-domain full waveform inversion relies on being able to efficiently solve the 3D Helmholtz equation. Iterative methods require sophisticated preconditioners because the Helmholtz matrix is typically indefinite. In the first part of the talk I review a preconditioning technique that is based on row-projections. Notable advantages of this preconditioner over existing ones are that it has low algorithmic complexity, is easily...

Tristan van Leeuwen, "Yet another perspective on image volumes"
An extended image is defined as the multi-dimensional cross-correlation of the source and receiver wavefields used for imaging. This extended image will reveal velocity errors by de-focusing and can thus be used for velocity analysis. However, for optimal sensitivity to velocity errors, the subsurface offset has to be aligned with the local dip. As this dip is not known a priori, we consider forming the extended image for subsurface offsets in all directions. However, computing and storing such a large image volume...

Tim T.Y. Lin, "An introduction to cosparse signal reconstruction"
Undersampling techniques in exploration seismology usually relies on the assumption that seismic records and images permit sparse approximations under certain representations, such as Curvelet coefficients. Recent findings have suggested that for redundant representations (of which Curvelet is an example), the analysis operator that maps the physical signal to coefficients may also play a crucial role in recovering data from incomplete observations. In particular, the number of zero-valued coefficients...

Tim T.Y. Lin, "Recent developments on the robust estimation of primaries by sparse inversion"
Robust estimation of primaries by sparse inversion is a next-generation surface multiple removal technique with an objective to truly invert an operator that models the free-surface. Key to the success of this approach is the imposition of a sparsity constraint on the primary impulse response in the time domain. This is accomplished by carefully applying large-scale convex optimization techniques on an extended L1 minimization problem. One of the benefits of...

Hassan Mansour, "Seismic trace interpolation via sparsity promoting reweighted algorithms"
Missing-trace interpolation aims to reconstruct regularly sampled wavefields from periodically sampled data with gaps caused by physical constraints. While transform-domain sparsity promotion has proven to be an effective tool to solve this recovery problem, current recovery techniques make no use of a priori information on the transform-domain coefficients. To overcome these vulnerabilities in solving the recovery problem for large-scale problems, we propose...

Lina Miao, "Accelerating on sparse promoting recovery and its benefits in seismic application"
Sparse promoting recovery problem arises more and more frequently with the broad application of compressed sensing tool in exploration seismology. Because of the curse of dimensionality, the prohibitive computation burden on iteratively evaluating objective functions is one of the key issues that constrain high performance l1 solver. In this paper, we try to further improve the convergence performance of SPGl1, one of the state-of-the-art large...

Felix Oghenekohwo, "Compressed sensing: a tool for eliminating repeatability in acquisition of {4D} (time-lapse) seismic data"
In 4D (time-lapse) seismic data acquisition, a very significant step is the repeatability of the acquisition process. In other words, the geophones must be placed at the exact location as they were, during baseline survey and acquisition. This condition is required to be able to produce an image of the same location over time and this enhances a proper reservoir characterization. The cost of repeating...

Bas Peters, "Frequency domain {3D} elastic wave propagation in general anisotropic media"
Elastic wave propagation in 3 spatial dimensions is modeled using a wave equation containing the full stiffness tensor consisting of 21 independent components. This allows modeling in general anisotropic media. The wave equation is discretized on several Cartesian and rotated Cartesian staggered finite-difference grids (using a 2nd order approximation). The grids are linearly combined and, in combination with a antilumped mass strategy, minimize numerical...

Anais Tamalet, "Variance parameters estimation - application to full waveform inversion"
Many inverse problems include nuisance parameters. While not of direct interest, these parameters are required to recover primary parameters. In order to estimate these nuisance parameters as well as the primary parameters in large-scale inverse problems, a method based on variable projection, which consists in projecting out a subset over the variables, has been developed. We present here the application of this method to the problem of variance parameters...

Mike Warner, "Anisotropic {3D} full-waveform inversion of the {Tommeliten} {Alpha} field"
We have implemented a robust and practical scheme for anisotropic 3D acoustic full-waveform inversion. We demonstrate this scheme on a field data set, applying it to a four-component ocean-bottom survey over the Tommeliten Alpha field in the North Sea. This shallow-water data set provides good azimuthal coverage to offsets of 7 km, with reduced coverage to a maximum offset of about 11 km. The reservoir lies at the crest of a high-velocity antiformal chalk section,...

Haneet Wason, "Ocean bottom seismic acquisition via jittered sampling"
We present a pragmatic marine acquisition scheme where a single (or multiple) vessel sails across an ocean-bottom array firing airguns at ? optimally jittered source locations and instances in time. Following the principles of compressive sensing, we can significantly impact the reconstruction quality of conventional seismic data (from jittered data) and demonstrate successful recovery by sparsity promotion. In contrast to random (under)sampling, acquisition via jittered (under)sampling helps in...

Xiang Li, "Fast {Gauss-Newton} full-waveform inversion with sparsity regularization"
Full-waveform inversion (FWI) can be considered as a controlled data fitting process, in which we approximately fit observed data by iteratively updating the initial velocity model, we expect the final model can reveal subsurface structure till the wavefield misfit can converge to designed tolerance. The conventional FWI approach is expensive since it requires the inversion of a linear system, which involves extremely large multi-experiment data volumes. To overcome...

Ozgur Yilmaz, "Weighted methods in sparse recovery"
In the recent years we have successfully employed "weighted" algorithms to recover sparse signals from few linear, non-adaptive measurements. The general principle here is to use prior knowledge about the signal to be recovered, e.g., approximate locations of large-in-magnitude transform coefficients, if such information is available. An example for this is the use of weighted 1-norm minimization to improve wavefield reconstruction from randomized (sub)sampling. We will review these results and outline some new directions...