CWP Seminar - Spring, 2006

Date

Speaker

Subject

Notes

01/23

Rodrigo

Seismic signatures of reservoir depletion (cont.)

Coming back from my leave of absence from last semester I will present a refresher on the effects of reservoir depletion on the stress distribution around a compacting reservoir. I intend to show some insights I have gained through modeling of such stress changes and how I think this might modify the elastic stiffnesses describing the overburden/underburden of compacting reservoirs.

Ivan

Drill-bit seismic

As I am currently involved in using seismic waves induced by drill-bit noise to image the San Andreas fault at SAFOD, I would like to introduce the basics of "conventional" seismic-while-drilling practice (SWD). We will talk about the character of drill-bit noise, and discuss why most SWD applications rely on recording the so-called pilot trace. This will be an introduction to later discussions on an deconvolution-based SWD processing methodology I am currently working on.

Paul

What is a seismic image?

What is a seismic image? Why do we operate with so many imaging methods? Could it be that most imaging techiniques reduce to simple concepts of wavefield reconstruction and wavefield matching? I will use this and several future seminars to discuss in simple terms imaging methodology. A prime target of investigation is the imaging condition. I intend to make this seminar both a tutorial and an opportunity for discussion, leading to future research projects.

01/25

Yaping

Point-source radiation patterns in attenuative TI media (Part I)

In addition to the velocity anisotropy, the attenuation anisotropy changes the energy distribution along the wavefront and thus influences the AVO behavoir. To understand the wavefields excited from a point source in attenuative anisotropic media, we integrate the spectrum of the particle displacement over the wavenumber (or slowness) components. I will discuss a few observations: 1) Although the inhomogeneity angle of the waves, i.e., the angle between the real and imaginary wavevectors, is a free parameters for plane waves, it is determined for the total wavefield. 2) The inhomogeneity angle depends on the strength of the velocity and attenuation anisotropies. For unbounded media, it reduces to zero in symmetry directions. 3) The relationship between the phase and group velocities, as well as the relationship between the phase and group attenuations, are generalized. 4) Anisotropic attenuation changes the radiation patterns. The discussion helps developing ray-tracing algorithms for attenuative anisotropic media.

Jyoti

Shear wringing of oil shales

In my last talk on shear properties of unconventional hydrocarbon resources, I am going to talk about the influence of temperature on the moduli and attenuation (and their respective anisotropies) in the seismic and sub-seismic band. I will also discuss the most probable attenuation mechanisms in this measurement window.

Suzie

Using coda wave to measure medium attenuation

Seismic coda waves are a natural wonder. Because they are formed by scattered waves from numerous heterogeneities in the lithosphere, nature does the averaging over a large volume of the earth and leads to beautiful simplicity such as the separability of seismic source, progapation medium and recording site effects. In this talk, I will focus on the decay rate of coda amplitudes.

01/30

Roel

Using repeat events at Mount Saint Helens to monitor a moving source in the volcano.

During the recent eruption, Mount Saint Helens excited extremely repeatable seismic events. The correlation of the recorded waveforms decreases with the time lag. I will show that the decorrelation of the waveforms can be explained by a source that moves over a distance of about 100 meter per day.

Xiaoxia

Multicomponent azimuthal AVO synthetic study with moveout-based anisotropic spreading correction (Part 1)

We analyzed PP and PSV reflections for reflection coefficients from a horizontal layered model comprised of isotropic and orthorhombic layers. With moveout-based anisotropic spreading correction, the extracted reflection coefficients agrees fairly well with the exact ones. In contrast, with common gains applied to the synthetics, the recovered reflection coefficients deviates from the exact ones.

02/01

Carlos

Prediction filters and prediction error filters (from 1D to 2D)

Prediction and prediction error filters are used in 1D for deconvolution with the purpose of subtracting the source wavelet and reverberations. I discuss how to calculate prediction filters in 2D by minimizing the least squares error of the filtered signal. The coefficients of the filter in 2D depend on the 2D autocorrelation of the signal.

Jia

sooner field project

an evaluation of several development ideas on the sooner field, weld county, colorado. Kingdom Suit and Eclipse simulation software were used to modelize the scenario.

Kurang

Diagnostic of edge effects in virtual source gathers and using different receivers as the virtual source

The edge effects in the virtual source imaging can be reduced by tapering the correlation gathers before summation. In this presentation, I diagnose the shape of the artifact due to the edge effect. I also show the virtual source gathers generated using 3 different receivers as the virtual source and discuss the possible explanation for the differences between them. This is work in progress.

02/06

Niran

Gaussian-beam migration

The Gaussian-beam migration is an accurate and efficient depth migration method. It has sparked interest, because of the ability to image complex velocity structures.

Steve

Hammer Source Seismic Data: Yucca Mounting Heating Experiment.

Volcanic tuff within the DOE, Yucca Mountain, NV, site has been heated to study the effects of long-term storage of nuclear waste. Hammer seismic array data obtained during this experiment can infer fluid migration (water, leakage) in the region around the experimental borehole (“drift”). This data is of varying quality, and requires extensive preprocessing and time alignment to remove time delays not associated with heating. Interactive alignment techniques/software were used to preprocess the data for correlation analysis, thus determining its suitability for seismic interferometry. These data show dramatic system-wide changes following one year of heating beyond the boiling point of water.

Dave

Local correlations of N-dimensional signals

Consider two N-dimensional digital signals, each with S samples. For some number of lags L (L much less than S), the cost of computing a single cross-correlation of these two signals is proportional to L times S. We show how to compute, not one, but S local cross-correlations, again with computational cost proportional to L times S. By local, we mean cross-correlation of the signals after applying a local Gaussian window. Computational cost is independent of the window width.

02/08

Huub

A hybrid formulation of map migration and wave-equation based migration using curvelets

I present a (blackboard) derivation for the leading order approximation to migration with curvelets, and show that to leading order migration is simply a linear transformation of coordinates. The current derivation is for homogeneous media only, but it is my current understanding that the essence of the derivation applies to heterogeneous media also. Time permitting (why is time always limited when you're having fun), I'll show some synthetic examples.

Ilya

Asymptotic analysis of S*- and P*-waves in TI media

Pseudospherical S*- and P*-waves are nongeometrical modes formed by evanescent energy contained in point-source radiation. When the source is close to the free surface or an internal boundary, such nongeometrical waves can dominate the wavefield. In particular, the S*-wave generates intensive wide-angle shear arrivals that can provide important information for anisotropic parameter estimation. To explain the influence of transverse isotropy (TI) on both the kinematics and decay factor of pseudospherical waves, I will present simple asymptotic expressions obtained by the stationary-phase method.

02/13

Rodrigo

Stress Induced anisotropy

Following up previous talk, I will be commenting on stress induced anisotropy, and how nonlinear effects caused by stress can be accounted for by using 6th-order elastic tensors!!!???

Jyoti

A new method of extracting the seismic source

I will talk about how crosscorrelation can be used to find the seismic source.

Paul

Prestack imaging conditions

In the first-order approximation, an imaging condition amounts to a matching operation between two four-dimensional wavefields from the source and receivers. A conventional imaging condition (for Kirchhoff migration, wave-equation migration, reverse-time migration etc.) cross-correlates the two wavefields at every location in space. More general imaging conditions compare the two wavefields based on shifts in time and space. For perfect imaging, the wavefields show perfect correlation at no shift (in time or space). Imaging inaccuracies lead to correlations that maximize for non-zero shifts. Such imaging conditions can be used for angle-dependent reflectivity decomposition.

02/15

Ivan

SWD and decon interferometry

I will briefly finalize the discussion I started on my previous talk on the nature of seismic-while-drilling noise and the need for deconvolution interferometry for SWD without a pilot trace. As time allows, I will start a blackboard derivation on the meaning of deconvolution interferometry by analyzing the result of performing this type of interferometry on a toy model consisting of a single reflector in a homogeneous medium. I will discuss on the application of the stationary phase method for this example.

Kurang

Deconvolved waveforms: What do they represent?

The talk on Treasure Island earthquake data focused on using deconvolution as a tool for seismic interferometry. Here I establish a connection between the deconvolved waveforms and the elements of the propagator matrix.

Suzie

Coda Wave Modeling

I will continue the topic on coda wave but focus on modeling in this talk.

02/20

Yaping

Effective attenuation anisotropy of layered TI media

Effective attenuation anisotropy could result from fractured media or stack of attenuative layers. For the latter, effective attenuation anisotropy is influenced by various factors such as the heterogeneous isotropic background, intrinsic anisotropy for both the velocity and the attenuation, as well as their coupling effects. Start from snapshots of SH-wave propagation in a thin-layered attenuative model, I'll breifly discuss the approximate anisotropy parameters and possible range of their values.

Xiaoxia

Azimuthal AVO synthetic study of P-waves with moveout-based anisotropic spreading correction (Part 2)

Last time, I covered the full-wavefield modeling and the algorithm to extract wide-azimuth reflection coefficients with moveout-based anisotropic spreading correction. This seminar, I will present the comparison of the azimuthal AVO results on three models with the anisotropic spreading correction and with empirical gain compensations.

Martin

Approximations to the Zoeppritz equation for long offsets

For AVO-analysis of long offset data, it is useful to derive approximations to the Zoeppritz equation that is valid for incidence angles close to the critical angle. It is shown that it is possible to estimate the S-wave velocity contrast, given that the P-wave velocity contrast can be estimated directly from the critical angle.

02/22

Eileen Poeter

2006 Darcy Lecture (Berthoud Hall, Room 108)

All Models are Wrong; How Do We Know Which Are Useful?

02/27

Carlos

LCC for time-shift estimation and the effect of ensemble averaging

I will discuss the improvement introduced by using gaussian windowed autocorrelations in the estimation of time-lapse traveltime change for scattered waves. Despite the improvement, however, the estimation of time-shifts in multiple-scattering media is still impaire due to the fact that we do not have many realizations of the same experiment. I will show the effect of ensemble averaging for an example of a local velocity change in the medium.

Jia

AK Semblance and 2-D application

I will give a brief review on the AK semblance and its application to three VTI models, all of which has polarity reversal but with reversal point at different offsets. AK Semblance and regular semblance were implemented to the three models and AK semblance was found to have more accurate inversion results. Class 2 and Class 1 type AVO were studied and a strong recommendation to apply AK Semblance to Class 2 type AVO was made.

Dave

Local prediction filters

An efficient method for computing local auto-correlations leads to a new method for computing local prediction (or prediction error) filters for multi-dimensional images. Using a conjugate gradient method for least-squares optimization, we compute a different prediction filter for each sample in an image. These adaptive prediction filters preserve locally coherent signals, while attenuating random noise.

03/01

Niran

Geophysical Image Processing With SU

Comparison of different migration methods' results of Viking Graben.

Steve

Water recharge/transients and local cross-correlation analysis hammer seismic calibration shots from Yucca Mountain drift scale heating experiments.

Radioactive waste stored in drift tunnels is expected to generate heat in excess of 200 ºC at drift tunnel walls. Rapid change in hammer seismic calibration waveforms, particularly those following the reduction of experimental drift heating to 50% output (<100 ºC), combined with numerical simulations and lab analysis of volcanic tuff from Yucca Mountain suggest rapid collapse of the “heat pipe” surrounding decaying radioactive waste. This collapse/fracturing causes permanent damage to the surrounding rock, and indicates rapid recharge of water about the drift. Traditional coda wave interferometry analysis techniques must be altered or modified as the waveforms to not change slowly enough, or do not exhibit multiple reflector energy diffusion characteristics common to other cases where the technique has been applied. Measurements of correlation lag times in the waveforms with respect to the baseline event (linear fits) are constrained to smaller and smaller portions of the data due to said rapid changes, and changes due to rock damage.

Ilya

Anisotropic stationary-phase solutions for pseudospherical waves (Part 2)

Existing asymptotic results for P*- and S*-waves are limited to isotropic media and often involve manipulations with integration contours in the complex slowness plane. A much more efficient approach is based on the zero-order stationary-phase method (SPM), which can be extended to anisotropic models in a relatively straightforward fashion. Using the concise zero-order SPM solution, I will discuss properties of the S*-wave in vertical symmetry planes of transversely isotropic and orthorhombic media.

03/06

Rodrigo

Stress Induced Anisotropy (cont.)

Today I hope to share with you some of the results from 2-D modeling. The Model is comprised of a rectangular reservoir experiencing pore-pressure decrease. The host rock and surroundings are all modeled as a isotropic homogeneous background. With the onset of the non-hydrostatic stress due to the pore-pressure decrease the medium becomes inhomogeneous and displays orthorhombic symmetry. Moreover, P-waves will display elliptical anisotropy, while the shear-wave polarization will change from the center to the edges of the reservoir.

Roel

A simple way to derive Lorentz contraction

Lorentz contraction in special relativity usually is derived with thought experiments involving flashes of light that propagate from A to B that are observed by observers moving at different speeds. I will present a much simpler derivation of Lorentz contraction that relies on electrostatics and magnetostatics only. And the puzzle is ..... does this give the right answer for the wrong reason? This is the sort of thing I learn from teaching undergraduate classes!

03/08

Ivan

Theory of deconvolution interferometry

Following up on my last discussion, I will go through the main points of solving the integral over sources for deconvolution using zero-order SPM. This will show that deconvolution interferometry yields the same stationary points as cross-correlation and removes the source function. Deconvolution then does give an estimate of the response between two receivers. Then, using the representation theorem and acoustic reciprocity I will argue that deconvolution gives mostly the causal response between receivers, as opposed to cross-correlation, that gives both a causal and an acausal response. To finalize, I will show numerical examples.

Yaping

Point-source radiation patterns in attenuative TI media (Part II)

A follow-up talk on point-source radiations, in which I will discuss 1) how to determine the inhomogeneity angle for VTI media with VTI attenuation; 2) how this angle influences the attenuation behavior of the waves. In layered attenuative media, such an angle is governed by the Snell's law.

Paul

Angle-decomposition

A generalized imaging condition, discussed in prior seminars, leads to generalized methods for angle decomposition. How many methods are there? Are they all equivalent? Can we operate in different domain (e.g. space-offset, space-time, offset-time)? I discuss imaging conditions from a theoretical standpoint a seek answers to some of those questions.

03/13

David Lumley, 4th Wave Imaging

Some seismic imaging/monitoring R&D problems of interest to 4th Wave

Recent advances in data processing and uncertainty analysis for time-lapse seismic imaging.

03/15

David Lumley, 4th Wave Imaging

More seismic imaging/monitoring R&D problems of interest to 4th Wave

Imaging shallow military targets (bunkers, weapons caches) using high-frequency seismic waves.

Time-lapse GPR monitoring of contaminants in groundwater using technology developed for 4-D petroleum reservoir monitoring.

03/20

Spring Break

Ski Colorado

Complete papers for CWP report!

03/27

Jyoti

Virtual real source

Following up on my talk last time, I will show modeling results of estimation of the source function using cross-correlations.

Kurang

Correlation gathers for dipping reflector & a preliminary attempt to remove source signature from virtual source gathers

I will present the behavior of correlation gather for dipping reflector. This includes change in the location of the source giving stationary phase contribution and change in the slope of the correlation gathers. The second part of my talk will include a preliminary attempt to remove the effect of source signature in the virtual source gathers.

Suzie

Using Coded Seismic Signal to Study the Crustal Structure

In this talk, I will discuss a collaborative research which uses coded seismic signal to study the crustal structure. Unlike a high impact energy source, which is not usually environmentally friendly, this method use a series of small energy sources, which is a coded seismic signal. The analysis uses data having a S/N ratio much smaller than 1. Developing properly coded seismic signals and data processing techniques are the main objectives in this project.

03/29

Jeff Shragge, Stanford University

Generalized Riemannian Wavefield Imaging

Riemannian wavefield extrapolation (RWE) extends seismic wavefield propagation to a class of generalized coordinate systems. Merging these geometric constructs with wavefield propagation allows for some familar seismic imaging issues to be addressed directly through dynamic (e.g. ray-traced) or geometric (e.g. conformally mapped topography) coordinate systems. Examples include steep dip and overturned event imaging and WE migration from topography. However, specify RWE meshes is wrought with new complications, including the potential for coordinate system triplication and grid clustering. In this talk, I discuss a strategy for circumventing these issues based on non-orthogonal wavefield propagation on differentially generated meshes.

04/03

Xiaoxia

???

Did not care to summarize my work.

Carlos

Registration of PP and PS waves

Discuss some of the practical difficulties in doing registration of PP and PS waves using only stacked data. Show how to obtain the long wavelength velocity ratio of the rms velocities using local cross-correlations.

04/05

Jia

AVO sensitive velocity analysis on 3D wide azimuth reflection data

The AK Semblance method applied on 3D data, particularly on polarity reversed data. Difference of conventional semblance and AK semblance will be shown.

Niran

4-D Seismic Data Interpretation at Weyburn Field, Canada

Weyburn Field is an oil field, discovered in Saskatchewan, Canada in 1954, with estimated 1.4 billion barrels of OOIP from a fractured carbonate reservoir of the Midale formation. The reservoir considered thin and below seismic resolution. Some faults are detected from seismic data, and they have a critical influence on the CO2 injection at Weyburn oil field.

08/30

Carlos

Prestack time migration of converted wave data.

In my previous seminars I talked about registration of converted waves using only stacked data. Now I will discuss registration in the context of using prestack time migration of converted wave data. More specifically I will touch in the practical difficulties and limitations of doing prestack time migration of converted wave data compared to the prestack time migration of pure modes. I will show that prestack time migration of converted wave data has more limitations than for pure modes.