CWP Seminar - Spring, 2007

Date

Speaker

Subject

Notes

01/17

Kurang

Detection of incoherent seismic features using variational norm

An important seismic analysis tool involves detection of incoherent seismic features such as faults and stratigraphic boundaries which are often associated with the formation of subsurface traps in which hydrocarbons might accumulate. In this talk, I present the application of variational norm for detection of such incoherent seismic features. The variational norm approach allows user-controlled search direction and does not require picking of horizons. This is the work I did during the internship with Apache Corp. in Fall 2006.

John M.

Prediction of 3D Seismic Acquisition Footprint

Land 3-D seismic survey results are frequently contaminated by spatially periodic interference that correlates exactly with the acquisition geometry. Such artifacts are commonly referred to as �footprints� and can hamper seismic interpretation or even invalidate such aspects as inferences drawn from amplitude maps, Amplitude-vs-offset (AVO) measurements and other attributes. In this presentation, different approaches to prediction of 3D data are discussed. Results are shown for pseudo-3D processing of 2D data.

Paul

Interferometric imaging condition

Accuracy of depth seismic imaging depends on the accuracy of the velocity models used for wavefield reconstruction. Models can be decomposed in two components corresponding to large scale and small scale variations. In practice, the large scale velocity model component can be estimated with high accuracy using repeated migration/tomography cycles, but the small scale component cannot. In this seminar, we begin to explore the impact made by the small-scale unknown velocity variations on seismic images and the techniques we can employ to address this problem.

01/22

Laurent Stehly

Laurent Stehly

Laurent Stehly

01/24

Rodrigo

The effect of stress-induced anisotropy on traveltime shifts: insights from stress equations

As I have previously demonstrated, when considering reservoir compaction, traveltime shifts caused by stress-induced velocity changes can be decomposed into two components: one due to isotropic (volumetric) changes and another due to stress-induced anisotropy. Since then I have wondered how those two components compared to each other. It turns out that the stress-induced anisotropic component should dominate over the isotropic one. This conclusion can be reached by examining the equations describing the excess stress generated by the compaction itself. During this seminar, then, I will show you how I stumbled into this result.

Niran

Can we improve Gaussian Beam Migration by migrating different frequency bands with different velocity models? (I)

For answering this question, smoothed velocity models are required. This leads us another question; how much should the velocity model be smoothed? To find an answer of this question has steps. In this seminar, I'll be explaining the first step.

Masatoshi

Application of seismic interferometry to down-hole data at Cold Lake

From down-hole data, we detected seismic signals vertically propagating from surface to bottom, by using a seismic interferometry method with ambient noise data. We found P and S wave arrivals in vertical and horizontal components, respectively, at down-hole stations. Also shear velocity varies for polarization direction, which is probably the first result that shows shear wave splitting found only from noise.

01/29

Norm

Norm

Norm

01/31

Gabi

Angles at the image point

Continuing the discussion about the angles at the image point. Some motivations to extract this angle information and produce Angle Domain Common Image Gathers (ADCIG's or aCIG's). Our first goal is to have a kind of map showing the ilumination we have at each image point in relation to the reflection angle and the azimuth of the reflection plane. The next step would be work on the angle dependent reflection coefficient (which angles?). Comments and sugestions are very wellcome!

Jia

Anisotropic imaging based on wave equation migration

This is the introduction to my thesis project. I will be doing velocity analysis with wave equation migration. The migration should be done before any velocity and anisotropic parameter estimation. So, it is important to decide which wave equation to use and what migration algorithm to use. Here, I will talk about Alkhalifah's acoustic wave equation for VTI media and show the artifact of this AWE (Alkhalifah, 2000). Being aware of the inadequacy, I will instead use elastic wave equation to the modeling and migration. My thesis will also be on imaging condition for the elastic wavefields and paramter estimation based on anisotropic angle decomposition.

Ilya

Conventional and "nongeometrical" components of cuspoidal PS-wavefronts in TI media

As I discussed last semester, cusps (triplications) are quite typical for wavefronts of SV-waves in transversely isotropic (TI) media, and PSV-waves can become cuspoidal as well. This time, I will present a simple analytic description of cusps for mode conversions and show that for relatively hard rocks with low Vp/Vs ratios it should be possible to record at least the low-angle part of the PSV-wave cusp. Furthermore, if the source is located close to the interface and excites nongeometrical modes, the wavefront of the "pseudospherical" S*-wave contains the remaining part of the cusp.

02/05

Fan

Phase problem in the cross correlation using stationary phase method

Last time I showed that the cross correlation of the noise would give us the Green's function between the two receivers. Because stationary phase method was used in the derivation, the phase of the result would not be right. So can we make the phase right? Can we get any useful information even with the wrong phase?

Greg

Multiple-related imaging artifacts in reverse-time migration

The use of the full "two-way" wave equation in reverse-time migration implies that RTM should be able to image using multiples as well as primaries. However, multiples do produce artifacts in RTM images. Incomplete reconstruction of the wavefield due to limited acquisition aperture may cause some artifacts, but the limitations of an imaging condition that assumes Born-approximate scattering would appear to be more important.

Roel

Research questions for the extraction of the Green's function in EM applications

We have shown theoretically that the Green's function for the diffusion equation can be extracted from fluctuations. This opens up the possibility to apply this extraction to low-frequency EM fields near reservoirs. I sketch interesting research questions that I discussed with Shell and that can potentially improve EM measurements near hydrocarbon reservoirs.

02/07

Jyoti

Reflection coefficients in attenuative TI media

Attenuation and attenuation anisotropy can have significant effect of reflection coefficients. Through modeling and using linearized analytic expressions, I show that the effect of attenuation contrast on AVO can be greater than the effect of velocity contrast. For weak attenuation anisotropy, reflection coefficients have isotropic dependence on attenuation, while for moderate/strong attenuation anisotropy (common case in the field), they show anisotropic dependence on attenuation.

Kurang

Removing overburden multiples and other artifacts from the virtual source data

Virtual source method, based on seismic interferometry, has still some challenges related to limited acquisition aperture and energy coming from the overburden. These challenges leads to artifacts (both physical and unphysical) in the virtual source data. I will address these issues, demonstrate them using a complex layered synthetic model and show how wave-field separation suppresses these artifacts.

Laxmidhar

3D Ray-tracing and model building for MVA

For 3D migration velocity analysis (MVA) in VTI media, it is crucial to develop a strategy for 3D model building and ray-tracing through it for traveltime computations. This traveltime table will be suitably used for Kirchhoff depth migration followed by MVA to obtain correct image of the model. Here we address some aspects of the 3D model building, ray-tracing and wavefront (WF) construction by using the method of tetrahedral model building before achieving our goal of 3D MVA. The suitable combination of tetrahedral model building and 3D WF construction ray-tracing is a very powerful tool for fast and accurate computation of traveltimes for 3D depth imaging. The tetrahedral modeling is based on the triangulation of smooth horizons with the continuous variations of sloth (square of slowness) in each layer.

02/12

Steve

Meshing Issues for Spectral Element Wave Propagation

Due to numerical instabilities associated with large transitions in material properties, spectral elements are both faster and necessary for maintaining stability. Meshing issues specific to 3^rd party software (EMC2) used for spectral element wave propagation with spectral element code SEM2DPACK are discussed. Potential solutions are presented and suggestions are solicited from the group.

Farnoush

Application of Anisotropic Geomertical spreading correction on Azimuthaly AVO analysis

This is the first time I am going to talk.So I will introduce myself and talk about my research plan. My plan is to use Xiaoxia`s method of geomertical spreading correction(MASC) on different timelaps from Rulisonfield.After correcting amplitudes MASC will send those corrected amplitudes directly to AVO analysis sequence.The advantage of using this method of AVO analysis is that it is more accurate compare to conventional one.Because geometrical spreading has a considerable effect on reflection amplitudes.The result of this accuracy is giving better idea about change of fracture orientation during four-year production.I also want to give the fracture model for the study zone within the reservoir.I will have a brief talk about the basic concepts of geometrical spreading and how to relate it to AVO analysis . I will also show the flowchart of the steps that are necessary for AVO analysis.That is my plan and I would appreciate any suggestion or new idea that makes my research process more interesting.

Dave

Local causal filters

Local causal filters are linear transforms that can be represented as multiplication by a lower-triangular matrix. By local we mean that such filters are not shift-invariant; their coefficients vary locally with each output sample computed. A lower-triangular matrix representation makes it easy to define corresponding transpose, inverse, and inverse-transpose filters with symmetries that are reflected in efficient algorithms. Combinations of these filters can be used to solve important problems in image processing.

02/14

John M.

Determination of orientation of features in images, and possible applications

My first go at using the Local Orientation Filter (LOF) based on the Gradient Square Tensor, that recently appeared in the Mines java toolkit. The LOF currently calculates attributes in 2D and 3D images including the dip and degree of linearity. This presentation shows the results of application of the LOF to an interesting seismic image. There is a limited discussion of possible applications of this technology.

Xiaoxiang

Velocity-independent layer stripping of PP and PS reflection traveltimes--introduction and the extended work

This is an introduction to my comps. I will make a theoretical review of PP+PS=SS method (Grechka and Tsvankin, 2002) and velocity-independent layer stripping (Dewangan, 2006). I will talk about the assumptions and advantages of the velocity-independent layer stripping method, comparing with traditional methods. The first step for my research plan is to apply this technique in the inversion of P-wave nonhyperbolic moveout in layer-cake VTI media. And maybe, the second step is to set that the overburden of the model in layer stripping method has no horizontal symmetry, i.e. TTI media, then study the theoretical expressoins and the numerical test.

Paul

Interferometric I.C.

I continue the discussion of the interferometric imaging condition started in my preceding seminar. Interferometric imaging condition can be formulated for passive arrays (exploding reflector model) or for active arrays. The discussion in this seminar is limited to the passive array case.

02/19

Gabi

Angles at the image point.

I will continue the discussion about angles at the image point, which is the subject of my comps project. I will show the first model we will work with and discuss how we first intend to approach the problem.

Greg

Wavefield reconstruction and the Kirchhoff integral

A discussion of the implications for wavefield reconstruction of recording pressure, but not the normal derivative.

Ilya

Anisotropic polarization vectors: How non-orthogonal can they be?

One of the main pitfalls in polarization analysis for anisotropic media is inappropriate application of plane-wave theory. Although plane P-waves and split S-waves propagating in the same direction have mutually orthogonal polarization vectors, this orthogonality is not preserved for point-source radiation. I will discuss simple analytic solutions for the polarization vectors of P- and S-waves in TI and orthorhombic media and show how the shape of phase-velocity surfaces can be used to predict anisotropic polarization anomalies.

02/26

Derek

Freeform Deformations using the Mines JTK

I will present an overview of my implementation of Botsch and Kobbelt style deformations in 3D. My code review will give details about the data structures that are used to represent surfaces which are included in the Mines JTK and how these data structures are used in my project. Finally I will show how I calculate the voronoi area and the mean curvature of a node using these data structures.

Niran

Can we improve Gaussian Beam Migration by migrating different frequency bands with different velocity models? (II)

For answering this question, smoothed velocity models are required. This leads us another question; how much should the velocity model be smoothed? To find an answer for this question, I tested beam behaviors on simple models. Then, I implemented velocity smoothing with Gaussian filter. Finally, I compared beam impulse responses.

Roel

Characterizing multi-scale processes

I will report on the DOE-workshop "Basic Research Needs for Geosciences: Facilitating 21st Century Energy Systems". A key overriding issue in the geosciences is the interaction of processes acting on multiple scale. I will present laboratory experiments that illustrate this issue in a stunning way.

02/28

Jia

Elastic wavefield modeling

I will show some test result on the newly coded finite difference elastic modeling codes in madagascar(Paul's contribution). This code is based on elastic wave equations but not with Laplacian operator, but start from displacement-strain relationship, stress-strain relationship and conservation law of momentum. Staggered grids were used to make computation more accurate.

Fan

Source distribution effect in wave interferometry.

Last time I showed that with the right phase shift, we could recover the response between two receivers by putting a source at the stationary point. But practically, we normally do not know where the stationary point is. Theory tells us, if we have the sources all around the two receivers, by integral the crosscorrlation of all the sources, we would also obtain the response between two receivers. But how many sources do we really need? Can we represent the integral by the summation of finite number of sources? If we can, how good or bad it is?

Laxmidhar

3-D model building and synthetic seismogram generation for MVA

Since the key issuse for MVA in 3-D VTI media is model building and synthetic seismogram generation, we present how to generate synthetics in a cost effective manner using wave-front construction technique. The traveltime table computed with this method can suitably be applied for prestack Kirchhoff depth migration and carryout 3-D MVA to produce better imaging results for VTI media.

03/05

Rodrigo

Dealing with Sixth-order tensor contractions

One of the problems I faced when translating stress and strains into seismic velocities was the dauting task of performing a contraction between a sixth-order elastic tensor and the strain tensor. Each contraction demands about 1400 operations (sums and products), making this operation impractical for both analytical derivations and numerical computations. I show you then a workaround I found, by considering Voigt notation, which reduces the number of operations by an order of magnitude. This was benecifial for me not only to speed up codes, but also to derive closed form analytic equations that consider that contraction.

Jyoti

AVO for an inhomogeneous incident P-wave

An inhomogeneous incident wave could have a significant effect on the reflection coefficients given that the medium is strongly attenuative. The effect on PS-waves is particularly interesting as I will show today.

Dave

Local plane filters

Local plane (or plane wave destruction; Fomel, 2002) filters have a variety of uses. These multi-dimensional filters are local in that they vary with each output sample; their coefficients must be computed efficiently from estimates of local dip and azimuth. They should work as well for vertical image features as for horizontal features. They should have efficient and stable inverses (1) to enable the design of high-fidelity notch filters and (2) for use as preconditioners in solutions to inverse problems that require structurally-consistent smoothing.

03/07

Vladimir Grechka

In-situ anisotropy measurements from P-wave VSP data

Vladimir Grechka

In-situ anisotropy measurements from P-wave VSP data

Vladimir Grechka

In-situ anisotropy measurements from P-wave VSP data

03/19

Steve

Genetic Algorithms for Non-Linear Inverse/Parameter Estimation Problems

I discuss the fundamental principles of genetic algorithms, and their intended application to solving the parameter estimation problem for the P and SV wavespeeds in the “dryout-zone” about a heated subterranean tunnel.

Greg

Reverse time migration and boundary value problems

I review the relationship between reverse time migration and boundary value problems, and potential consequences for imaging.

Xiaoxiang

Inversion of nonhyperbolic moveout using velocity-independent layer stripping

I present my recent work on nonhyperbolic moveout inversion using the layer stripping technique that I talked last time. I applied this technique to a model of layer-cake VTI media composed of three layers. The upper two layers are regarded as overburden layer and the lower layer is taken as target layer. Using the layer stripping, I got the interval long-spread traveltimes in the target layer, which was then inverted for the nonhyerbolic moveout coefficients.

03/21

Farnoush

preprocessing steps that should be applied on 3D data before starting gometrical correction

Today my talk is about a few preprocessing steps that I have done on 3D data( understanding the header words,viewing the data,making shot-gathers,making bins and superbins) .Also I will talk about a new challenging problem regarding to the unusual large AVO gradient ratios for the bottom of the reservoir.My next focus would be finding reasonable answer for that.

John M.

More on Local Orientation and Local Plane Filters

Continuing on from my last talk, this presentation covers my current work with local orientation filters based on the gradient-square tensor to highlight features of different types in seismic images. The presentation contains some explanation of the procedure and examples from two different synthetic datasets. The presentation also contains displays showing the effect of applying LocalPlaneFilter on real data. This program uses output of the local orientation filter to calculate local filters that attenuate planar features.

Derek

Freeform deformations in action

I will show a quick demo of my freeform deformation code; discussing the progress I have made. Then I will introduce the half-edge data structure and give my motivations for wanting to use this data structure.

03/26

Gabi

Wide-azimuth angle decomposition

I will show today my very first results for Wide-azimuth angle decomposition in homogeneous media for a flat horizontal reflector.

Niran

Gaussian Beams vs. Finite Difference Modeling Impulse Responses

Can we improve Gaussian Beam Migration by migrating different frequency bands with different velocity models? I am trying to find an answer to this question. In previous presentations, I show the beam behaviors on simple and smoothed -by using Gaussian filters- velocity models. In this talk I will compare beam impulse responses with Finite Difference Modeling impulse responses for various velocity models.

Kurang

Near-field in virtual source data

Virtual source method (seismic interferometry) is a useful tool, based on correlation, for imaging below complex overburden. The distance between the source (on the surface) and the receivers (below the overburden) is quite large compared to seismic wavelength. The wavefield recorded by the receivers hence, mostly contain far-field. Theory states that we get the exact Green's function by correlating the recorded wavefields. So do we actually recover the near-field by correlating the far-fields? I will address this question analytically. This issue has been addressed by Sanchez-Sesma (UNAM), et al., but I just would like to share this interesting result with all of you.

03/28

Jia

Elastic reverse time migration

I will show some preliminary elastic reverse time migration results. The wavefields are downward propagated with the finite difference code I presented in the last seminar. I will show the migration algorithm applied to a simple model and indicate that if the P and S waves are not separated in the data, the image we create have more artifacts than expected.

Rodrigo

Traveltime shifts in shot-gathers

Today I am going to show and discuss some numerical computations regarding time-lapse traveltime shifts caused by reservoir compaction in 2D. In particular I will focus on shifts distribution in shotgathers, showing how they might change as a function of stress distribution, angle of incidence and shot position. Also I will show how the isotropic and anisotropic components contribute to the observed traveltime shifts.

Jyoti

Separating intrinsic attenuation from scattering

It is difficult to separate intrinsic attenuation from scattering effects in reflection seismic data. By analyzing the change in seismic source signature, however, it might be possible to separate the two. I will show some modeling examples and discuss when this is possible.

04/02

Xiaoxiang

Inversion of nonhyperbolic moveout using velocity-independent layer stripping

I present my work in these days on the inversion of nonhyperbolic moveout. After I fixed some bugs in the previous process, I got fine inversion results using velocity-independent layer stripping technique. I tested the random noises, the linear noises and the 'model' noises which were added into the input data and got the inversion results for the target layer. I talk about this process and the comparison of these results. Furthermore, I tested the traditional nonhyperbolic inversion method using semblance analysis for the same input data. I talk about those results and compare them with those of velocity-independent layer stripping method.

Fan

Source distribution effect in wave interferometry.

How does the source distribution affect the result of the interferometry? Last time I showed that equally angle spaced source structure gives much better than the random distribution. This time I would show that the source angle is the most important factor in this problem and when angle variation is smooth, it will give good result.

Farnoush

Isotropic Avo analysis

Today I will talk about the isotropic AVO analysis that I did for checking whether the results from Xiaoxia`s processing are reasonable or not.This is just an approximate estimation to have an idea about AVO gradients.I am going to explain why for bottom of the reservoir the AVO gradient ratios that Xiaoxia derived are not so convincing.

04/04

Derek

Steve

Issues for full waveform inversion of the velocity structure around a heated tunnel

I discuss trends in my data set as a function of receiver position and address the question of what a “Franz” wave is (and why it's important). I discuss sensitivity of forward modeling to potential models, and identify arrivals of importance. Additionally I outline preprocessing and/or at-processing issues such as the source receiver transfer function, and the application of differential velocities to compensate for the lack of a trigger (i.e. poor acquisition technique) for receivers in regions where wavespeed varies with time. I briefly touch on gradient methods for non-linear inverse problems.

Sjoerd

Seismoelectrical Modeling in Stratified Media and Applications of Reciprocity.

I will briefly introduce the theory of seismoelectric coupling in porous media. Then I will show the system decouples in stratified media to SH-TE and P-SV-TM modes of propagation. I will show how the Greens function (in the wavenumber-frequency domain) can be constructed using a scattering matrix formalism. I will show the seismoelectric reciprocity theorems for the SH-TE system in non-flowing time invariant media. From this I will derive the source receiver reciprocity relations. Furthermore, the Greens function matrix can be recovered using crosscorrelations of passive reciprocal surface recordings of a distribution of sources in the subsurface.