CWP Seminar - Fall, 2007

Date

Speaker

Subject

Notes

8-29

Rodrigo

Time-lapse traveltime shifts for compacting reservoirs: 3D solutions for prestack data

SEG rehearsal. The topic of the paper is how to describe analytically traveltime shifts caused by reservoir compaction in 3D prestack seismic data

9-3

Fan

Source distribution for interferometry in diffusion field.

Last semester I showed that for wave problem, we can apply the interferometry game sucessfully if the source angle variation is smooth. Thory says interferometry method can also be applied in the diffusion filed. The difference is that we need sources all over the volume. Question is that if finite number of sources is enough and where should they be for the diffusion interferometry.

Jae

Integration of Seismic and CSEM data with its Uncertainty Analysis

For successful investigation of marine reservoir, it is now needed to investigate electric property of reservoir along with its location and size. Besides, we want to evaluate the uncertainty of the investigation result. For this end, different joint inversion strategy of Seismic and CSEM data is to be studied in Bayesian framework. I will describe the overall research plan, basic assumptions, 1-D model for the research, and CSEM modeling result.

Paul

Imaging with WDFs

Wigner distribution functions (WDFs) belong to the generic family of time-frequency transformations. WDFs have been originally developed in the context of quantum physics, but they have also found applications in signal (image) processing, e.g. in problems requiring attenuation of random signal fluctuations. The noise-attenuating properties WDFs can also be applied to the seismic imaging problem. I describe the basic properties of WDFs and suggest several seismic imaging applications.

9-5

Yong Z.

Rupture durations of Big earthquakes

Rupture durations of Big earthquakes are always quite long and are contaminated by multiple phases, such as PP and PcP in low frequency domain, although in high frequency domain we can remove the multiple contaminations, the coda effects may get serious, which will make it difficult to identify the end location of the P wave so that they cannot get the duration time accurately. In this work we used Emperical Green's Function to remove the coda effects with PLD deconvolution method. We used the preshock or aftershock as emperical Green's Function and deconvolve it with the mainshock to get the relative source time function (RSTF). By RSTFs we can identify the end location clearly from the seismograms and then we can get the accurate ruture duration time of the mainshock.

Weitao

An introduction to the airgun experiment in reservior in china

A field experiment using large volume airguns as active source was held in one reservior in north China, November 2006. Some strongpoints of this source have been validated and with the data from the experiment some results have been attained.

Eduardo

Seismic imaging using isochron rays

Isochron rays are orthogonal lines to isochron surfaces. These lines can viewed as physical rays if we assume the isochron as wavefront propagating in an equivalent velocity media. Both of these concepts can be useful to perform depth migration of common offset gathers.

9-10

Derek

Defining regions with flattening

I will show how to create isochronous surfaces for velocity modeling by flatting a volume of seismic data without picking horizons. I use local structure tensors to compute estimates of local dip. I will discuss how to use these local estimates of dip in a flattening problem. The flattening problem is an overdetermined problem in 3D and we use a least squares inversion to resolve this ambiguity. I will conclude with results from this algorithm.

Rosie

SOLUTION TO A WAVE EQUATION BY A MIXED FINITE ELEMENT WITH FICTITIOUS DOMAIN METHOD

To investigate the capability of fictitious domain methods to simulate the scattering of linear waves by an obstacle whose shape does not fit the mesh. The space-time discretization relies on a combination of a mixed finite element method with a fairly standard finite difference scheme for the time discretization.

Dave

Painting images with anisotropic diffusion

Structure tensors computed from 2-D and 3-D seismic images yield estimates of orientations and anisotropies of image features. We use these attributes to guide painting of seismic images, where paint color may correspond to seismic wave velocity, geologic time, or any other parameter that we wish to specify. We store the painted values in a second image as large as the first. After specifying values for a small number of pixels or voxels, we use an anisotropic diffusion algorithm to paint values elsewhere in the painted image.

9-12

Jyoti

Virtual Real Source

SEG rehearsal: I will talk of a method of extracting source signatures based on seismic interferometry.

9-17

Jia

Elastic reverse time migration

SEG rehearsal: I will talk about elastic reverse time migration with potentials.

9-19

Yong M.

Nonlinearity in Biomedical Ultrasound--Tissue Harmonic Imaging

Nonlinearity phenomena cannot be ignored in biomedical intensity and frequency. In this talk, we first simulate the generation of harmonics by Ultrasound wave equation in attenuation medium, the most significant character of nonlinearity. Then we apply those harmonics to image a tissue model. Due to properties of harmonics, the better results are shown.

Xiaoxiang

Nonhyperbolic moveout inversion for layered orthorhombic media using velocity-independent layer stripping

Last semester, I presented my work on nonhyperbolic moveout inversion for layered VTI media using velocity-independent layer stripping. This time we extended this stripping algorithm to 3D and applied it to the inversion for layered orthorhombic media. From the results of the synthetic test, we found that the velocity-independent layer stripping method remained more accurate and stable for azimuthally anisotropic media than the conventional Dix-type approach.

Jyoti

Computing interval attenuation

I will show a method of computing interval attenuation (can be anisotropic) of a layercake subsurface using the spectral-ratio method.

10-1

Steve

Tunnel Velocity Zone Inversion

I discuss the outcome of applying a regularized Gauss Newton non-linear gradient inversion to a synthetic tunnel data set similar to our Yucca Mt. data. This includes adjustments to the regularization process to account for large forward model compute-times. I will also discuss accuracy/features of forward data using modeling of the inversion zone, application of the source/receiver transfer function, and expected outcome of inverting via genetic algorithm - which should provide a range acceptable models in accordance with the non-uniqueness of the inversion.

Gabi

Wide azimuth angle decomposition

Last semester I discussed and presented wide azimuth angle decomposition for a horizontal reflector in media with different velocity models. Today I am going to discuss this decomposition for a general geometry of the reflector. I am going to adress important computer limitations and show the decomposition for a dip reflector in a homogeneous medium.

Farnoush

Study of the surface-wave and body-wave Green`s function extracted from cross-correlation

In seismic interferometry we do the cross correlation between two receivers seismogram to extract the Green`s function between two receivers. In most cases we won`t get the reflected body waves after cross correlation and we just get the strong amplitude surface waves.The question is where is the body wave?today I will explain some of the reasons that why we do not get the body wave easily.

10-3

Jia

Artifacts associated with potential imaging

I will compare the elastic images for the models I gave in the SEG talk and another model. The elastic images for complicated model with strong reflectors near source show artifacts as strong as true images. I will briefly discuss three possible method to remove those artifacts. (1) angle gathers (2) separation based on wave numbers (3) kill the conversion at conversion surface.

Ilya

Review of seismic fracture detection

I will give an overview of several notable contributions to the recent EAGE/SEG Research Workshop on Fractured Reservoirs in Perugia, Italy (September 3-6, 2007). The technical program of the workshop can be found at "http://www.eage.nl/events".

Yongxia

Acoustic wave propagation in 6-type statistic medium

While seismic wave velocity show large-scale deterministic geological structures in the earth, the ubiquitous micro-fractures often cause small-scale random variations. Although the earth which seismic waves travel through act as a low pass filter and smooth out some such small indterministic components, those statistic features exist in the subsurface. Six types of statistic media are frequently used in seismology such as Gaussian, Exponential, self-similar, white noise, Flicker noise and Brown noise. I compare different statistic models with seismic recordings with finite-difference methods. The results show that various statistical models in amplitude, energy and frequency will cause different synthetic seismograms.

10-8

John M.

Introduction to the steerable pyramid

The steerable pyramid is a filtering process developed in the 1990's in which an image is decomposed in terms of wavenumber and angle. Until now, most applications have to do with medical imaging. I am investigating the use of the steerable pyramid on seismic images, specifically to enhance channel features. In this talk I describe what is the steerable pyramid and give some details on how it works.

Chuck

Wireless sensor networks

Wireless sensor networks (WSNs) are an emerging technology that has many potential applications such as the long term monitoring of infrastructure. We are developing WSNs using various geophysical sensors, including seismic and ground penetrating radar. Many of these sensors are passive, and interferometric techniques are needed to analyze their data. This is talk an overview of our WSN technology, and introduces my plans to develop interferometry algorithms for monitoring infrastructure.

Rodrigo

Is R offset dependent?

During the SEG convention I realized that people are still not convinced that anisotropic velocity changes take place over compacting reservoirs and that such changes should be taken into account properly. Today I will briefly show (again) why they should.

10-10

Fan

Diffusion interferometry using deconvolution

In diffusion interferometry, deconvolution shows different character from cross correlation. For example, in a constant diffusion coefficient medium, we need volume active sources to recover the response between two receivers if cross correlation was used. In contrast, by using deconvolution, only one source is enough to recover the exact response between these two receivers. Why is this?

Jae

Integration of Marine Seismic and CSEM Data with its Uncertainty Analysis

In continuation from the last presentation, I will show 1-D seismic modeling result, and discuss more details about rock-physics model uncertainty and Bayesian model for stochastic inversion. Besides, I'm going to share some questions that arose during the research.

Paul

Imaging with WDFs

I continue the discussion of imaging using Wigner distribution functions (WDFs) and outline two imaging strategies. Both strategies achieve statistical stability by local averaging of extrapolated wavefields. One strategy, advocated by Borcea et al., involves local averaging relative to surface coordinates. This strategy works but it is prohibitively expensive and of no practical value. An effective and much more efficient strategy involves local averaging relative to image coordinates. I illustrate both strategies and discuss their relative advantages and disadvantages.

10-17

Yong Z.

Preliminary Analysis of Velocity Variations caused by Tidal Effect

By long term observation, Yamamura, et al.(2003) found that the changes of seismic velocity and attenuation are in good consistent with the variations of tidal strain, they found the areal strain caused by tidal effect is correlated well with the peaks of velocity changes and the attenuation changes. In 2007 B.S Wang and his coorperators also found similar phenomena in active source seismic experiments. In this presentation I analyze the possible mechanism of these phenomena. Based on Kachanov's theory we make clear the relationship between the stiffness and the extra stress, for this way we can simulate the time lapse of Yamamura's experiment. The main shortcoming of this work is the model does not contain the pore pressure effect, which will be solved in the future.

Weitao

Velocity change measurement using CWI method in the co2 injection experiment

A description on my work and plan in CWP with Roel.I will give some introduction to the co2 injection experiments in Japan and some research on this.Then I will talk about the deconvolution processing on our data to adjust two source signal.

Eduardo

Beam formation for isochron-ray migration

The motivation for using beams in migration algorithms is the reduction of the computational cost. In this talk, I'm going to show how to form suitable beams for isochron-ray migration. Also, I will explain how to perform a data redatuming together the beam formation.

10-22

Derek

Improvements and testing of flatting algorithm

I will present some updates to and analysis of the flattening algorithm that I discussed during my last seminar. Including updates to the way iso-chron surfaces are displayed, analysis of flattening around faults, and a preliminary convergence analysis.

Rosie

Comparisons among Deltas

According to the lectures in Applied Stratigraphy class which I have this semester, I made comparisons among river-dominated delta, wave-dominated delta, and tide-dominated delta. By listing their own dominant processes, mean palecurrent directions, main depositional features, main sedimentary structures and lithology, vertical trends in a transgressive setting, and vertical trends in a regressive setting, it's easier and clearer to tell their features.

Milana

Effective reflection coefficients in 3-D synthetic modeling and AVO inversion

Effective reflection coefficients represent generalization of classical plane-wave reflection coefficients to the case of arbitrary incident wave and finite frequency band. Effective coefficients are smooth, frequency-dependent, and valid for near-critical and post-critical reflections, because they account for the interference of reflected and head waves. Effective coefficients may find useful applications in 3-D synthetic modeling and parameter inversion in AVO study, as I will show on examples.

10-24

Yong M.

Nonlinerity in High Intensity Focus Ultrasound (HIFU)

I continue the talk about the nonlinearity in biomedical ultrasound. This time I will focus on the contribution of nonlinearity to the the thermal effect of HIFU and the generation of thermal lesion in tissue caused by HIFU.

Jyoti

Computing interval Q

Building on my last talk on a new method of extracting interval Q, this time I will show a synthetic data example. Q estimation can be erroneous if the events are weak (low amplitude) and/or if they interfere with other events.

Dave

Why do reservoirs appear to be pulling apart?

Displacements observed in time-lapse seismic images are vectors with three components. Specifically, we have observed lateral (inline and crossline) components of displacements that imply a lateral expansion of time-lapse images (not the reservoir). I speculate that these apparent displacements are caused by lateral velocity variations induced by reservoir compaction. With some simple approximations, this hypothesis can be tested.

10-29

Steve

Automatic waveform alignment and arrival picking on noisy field data.

When working with a misaligned set of field data (i.e. missing trigger information) preprocessing must be used to constrain inversion parameters and process/match model data. The work is complicated by a changing velocity field, a baseline/prior shot record with truncated/missing data, and a source function that is inconsistent from shot to shot. Inversion on a large volume of such data without manual oversight requires precise automatic arrival picking and alignment algorithms. I will discuss methods I've developed for my the field data that are applied to constrain velocity and determine/apply source spectra to model data in waveform inversion.

Gabi

Angle decomposition examples.

I am going to show some dip reflector examples for the angle decomposition following my last talk. I'll try to explain geometrically what happens with the angles as I increase the reflector dip.

Farnoush

Study of the surface-wave and body-wave Green`s function extracted from cross-correlation interferometry

Today I will continue the discussion about why we do not get the body-wave Green`s function from cross-corralating the seismograms between two receivers.I will talk about the stationary point solution of the body-wave term which is one of the four terms of the cross-correlation equation.It seems by putting number of sources around the stationary point we will get better amplitude for the body-wave compare to putting one source at the stationary point.

10-31

Jia

Elastic imaging

(a) Imaging artifacts: I will continue talk about imaging artifacts, mechanism and a possible solution.
(b) Back illumination: I will show why some reflectors behind the source are imaged in the PS image.

Xiaoxiang

Nonhyperbolic moveout inversion for layered orthorhombic media using velocity-independent layer stripping II

This time we make overburden more complicated to be orthorhombic, and I will show some nonhyperbolic moveout inversion results that indicate the velocity-independent layer stripping remains better than the conventional Dix-type approach. Furthermore, I will show the study result of one superbin from 03 Rulison field wide-azimuth data.

Yongxia

Seismic sounding of self-similar random media .

Last time, I talked about 6 types random media and compare them with amplitude, energy and frequency. This time, I will continue to talk about the parameters of random media with self-similar type. We will see four groups of the random media with different variance, a-lateral correlation length, b-vertical correlation length and velocity background. The parameters influence the amplitude, energy and frequency in seismogram.

11-5

John M.

More about the steerable pyramid

My last talk was an introduction to the steerable pyramid. I wish to use this to enhance channel features in 3D seismic images. This presentation covers the work I've been doing in the last month trying to improve my results on a 2D image, and working towards 3D implementation.

Chuck

Imaging with wireless sensor networks

Imaging with wireless sensor networks (WSNs) making passive measurements requires modifications to standard interferometry formulations. The reason is that, in many cases, the conditions needed to apply these standard formulae are not met with WSN measurements. I will present an imaging algorithm that is applicable to WSN measurements called polarized migration. This algorithm is designed to remove the non-physical correlation events that can occur when the ambient sources do not meet the requirements for standard interferometry.

Rodrigo

Symmetries of Third-order elastic tensor

To describe traveltime shifts for compacting reservoirs I used a stress-sensitivity tensor to connect stress changes to seismic velocity perturbations. Today I will talk more about the symmetries of this stress-sensitivity tensor, and how to use these symmetries to further understanding of the stress sensitivity tensor itself.

11-7

Fan

Source distribution in wave and diffusion interferometry

Instead talking something I understand, I would like to present some of the questions which puzzled me in my research. Hope I can get some feedbacks and ideas from the audience. The first one is that why smoothly varing angle function can be used in wave interferometry successfully. The second one is that why we can reconstruct the early time response in diffusion interferometry.

Jae

Integration of Marine Seismic and CSEM Data with its Uncertainty Analysis

In continuation from the last presentation, I will describe two different types of probability inference and explain details of Bayesian Model chosen for this research. Prior, Likelihood, and Posterior probability distributions will be covered in detail. Multi-dimensional Gibbs sampling method, an example of MCMC sampling, which is known to be applicable when we have conditional distributions, will be touched as well. Finally, I will explain current status of research and some desired outcomes.

Paul

Imaging with sparse data

Interferometric imaging condition (a.k.a. imaging with WDFs) reduces imaging artifacts caused by unknown random fluctuations in the image. Random fluctuations, however, can be caused by other factors, for example by sparse sampling of data. I suggest that imaging with WDFs can also be used to address this problem, although it is not yet clear to me how widely applicable this methodology is. However, preliminary results indicate that micro-earthquake location is improved. A potential application of this methodology is for real-time fracture imaging.

11-12

Yong Z.

Analysis of velocity and attenuation changes caused by strain

There's obvious relationship between velocity and attenuation changes with confined stess from sesimic experiments. In this work I analyse the mechanism of velocity change cuased by strain by Third-Order-Elastic tensor method, and further analyse the relationship between the micro-scale structures (especially crack density and crack radius) and the strain, based on this relationship I get a simple relationship between the seismic attenuation and the confined strain.

Ilya

How ``inhomogeneous'' evanescent waves in TI media really are?

It is commonly assumed that the amplitude of evanescent waves decays in the direction perpendicular to the real part of the slowness vector. I will show that for wave propagation outside symmetry directions of non-attenuative transversely isotropic (TI) media, the angle between the real and imaginary parts of the slowness vector (the "inhomogeneity angle") is highly sensitive to the anisotropy parameters and may be smaller than 45 degrees. These results have serious implications in the analysis of surface and nongeometrical waves in tilted TI media, as well as in predicting the attenuation-induced inhomogeneity angle.

Rosie

Paleozoic-Mesozoic Section: Turkey Creek, Morrison, and Dinosaur Ridge, Colorado

It is a overview of the Dakota Group stratigraphy in the area of Turkey Creek, Morrison, and Dinosaur Ridge. By remeasuring and redrawing the Skull Creek and the Muddy J Sandstone, we got the sedimentary structures, lithological changes, paleocurrent directions, trace fossils. Also by correlating with 3 measured sections plus the core section, we could set up some possible depositional models and conclude the possible reasons for the stratigraphic changes from Skull Creek to J Sandstone.

11-14

Derek

Creating Torn Surfaces

I've been working on ways to slice or tear triangulated meshes to model faulting. I will discuss the data structure I use to store and manipulate triangulated meshes. Then I will show one possible way to create tears in this data structure. Finally, I will open the floor for discussion of more intuitive user interfaces.

Eduardo

Weitao

An introduction to the coda wave interferometry

I will give an introduction to the CWI method for the new comers ,hopt it will help.

11-19

Diane

11-26

Yong M.

Detecting HIFU temperature elevation with ultrasound echo signals?

In last talk, I mentioned that detecting (monitoring) the temperature change during the HIFU treatment is a crucial part for HIFU technique. This time I am wanna show a possible way to do this, most in theoretical.

Jyoti

Dave

Finite-difference approximations of symmetric positive-(semi)definite differential operators.

In geophysics we often use finite-difference approximations to partial differential equations. When those equations contain symmetric positive-(semi)definite differential operators, we should preserve this SPD property in our approximations. Finite-difference approximations that do this may be both easier to implement and more accurate than those that do not. I demonstrate these advantages with the simple 1-D elastic wave equation.

11-28

Steve

Math and Algorithm of the Spectral Element Method

Gabi

A stability test and an artifact on the angle gathers

I am going to discuss the errors on the angle gathers due to errors on the dip parameter information. Also, I am going to talk about an artifact that may show up on the angle gathers.

Farnoush

study of the extracted Body-wave and Surface-wave Geern's function from the cross-correlated interferometry using surface source-receiver distribution

Continuing my prevoius talk ,today I will describe more reasons for weak body-wave amplitude that we obtain after cross-correlation.I added some sources under the reflector to see the effect on body-wave amplitude.Also so far the sources that I have been using were correlated sources.I tried uncorrelated sources to see the difference.It seems these sources decrease the cross-terms more than correlated ones..

12-3

Jia

Xiaoxiang

Generation of pseudo-P-wave data for the target layer using velocity-independent layer stripping

In the velocity-independent layer stripping, the reflection moveout data in the target layer can be found by layer stripping the reflection traveltimes from the top and the bottom of the target layer. Although this is feasible in principle, prestack traveltime picking is known to be tediouos, labor-intensive and noise prone. Based on what Vladimir and Pawan did using PP+PS=SS, we find direct traveltime picking can be replaced by computing the integral of the traces' convolutions. The result resembles P-wave reflection data in the target layer kinematicly.

Yongxia

12-5

John M.

Chuck

Rodrigo