CWP Seminar - Fall, 2005

Date

Speaker

Subject

Notes

08/29

Kurang

Analysis of Treasure Island earthquake data using seismic interferometry

Seismic interferometry is a powerful tool in extracting the response of ground motion. I show the use of seismic interferometry for analysis of an earthquake recorded by Treasure Island Geotechnical Array near San Francisco, California on 06/26/94. For this data set, I used deconvolution as a tool for doing interferometry as opposed to cross-correlation.

09/05

Rodrigo

Seismic signatures of reservoir depletion

Hydrocarbon reservoirs will experience some degree of compaction due to pressure drop associated with production. The depletion will then impact the effective stress magnitude and direction in the vicinity of the reservoir, bringing about velocity changes. In a time-lapse sense, such velocity changes might be used to monitor production and help to map the reservoir architecture. During this seminar I will try to illustrate those principles and explain how I do intend to go about the problem of relating stress changes to seismic signatures, preferably using analytic expressions.

Roel

Seismic interferometry of the motion in a building

Extract the building response from seismic interferometry

Ivan

Fracture-induced anisotropy

The influence of crack irregularity and partial contact between crack faces on the effective elastic properties of the medium. Overview of the limitations of effective medium theories in describing such heterogeneities. Most models of effective fractured media assume crack circularity and yield an effective orthorhombic medium. Using finite element modeling, we address this issue by looking at the deviations from orthotropy caused by partially-closed, irregular cracks.

09/12

Huub

velocity analysis without picking using curvelets

some preliminary test results and insights

Dave

Improved sinc interpolation

The length-8 sinc interpolation in libcwp (SU) yields less then 1% error for frequencies less than 60% of Nyquist. The alternative proposed here has less error. More importantly, new interpolators can be designed by specifying any two of three parameters: length, maximum error, and maximum frequency.

Steve

Parabolic and sinc interpolation examples

Examples of parabollic and sinc interpolation and associated error surfaces as a function of sample shift function and input bandwidth. Interpolation errors for synthesized seismic horizon data.

09/19

Niran

Introducing myself

Geophysics in Turkey

Ilya

Evanescent waves as a parameter-estimation tool

Evanescent (inhomogeneous) plane waves contained in the plane-wave decomposition of point-source radiation produce not only surface waves but also nongeometrical modes that can travel far away from the interface. Extension of the Christoffel equation to evanescent waves in TI media shows that the relationship between their horizontal velocity and decay factor is strongly influenced by the anisotropy parameters. These results can help to design new anisotropic parameter-estimation algorithms in cross-hole and VSP geometry.

Xiaoxia

Tools for azimuthal AVO analysis in SU

I will discuss two pieces of SU code (suavo and suazvelanavo) I wrote for AVO analysis with anisotropic geometrical-spreading correction.

09/26

Jia

my potential project

AVO and velocity analysis in 3D anisotropic media

Ken

The puzzling ``Vz noise" in vertical-component OBS data

The vertical-component in multi-component OBS data recorded on soft, muddy bottom exhibits events on common-receiver gathers (but not on shot records) with clear S-wave moveout having amplitude that rivals that of events with P-wave moveout, while the pressure-wave data show no evidence of events with S-wave moveout. Recent data obtained with a new multi-component OBS system, developed by Fairfield, contain evidence that might help shed light on the problem. This remains to be demonstrated, probably through appropriate elastic-wave modeling. Any takers?

Carlos

Failure of hyperbolic velocity analysis for the chirp model

Coventional velocity analusis based on the hyperbolic assumption fails when dealing with structurally complex areas. I show results of velocity analysis on two different models with increasing amount of lateral velocity variations.

10/03

Yaping

Effective attenuation anisotropy from thin-layered media

The effective medium theory suggests a possible cause of the attenuation anisotropy. Like the velocity anisotropy, the attenuation anisotropy can result from periodic thin layering (i.e., interbedding of thin attenuative layers with different velocity and attenuation). Depending on the layer properties, the strength of the attenuation anisotropy is generally different from that of the velocity anisotropy. In some instances, the attenuation anisotropy is more pronounced. A special case of alternating thin fractured (attenuative) layers with different fracture azimuth reveals that the velocity and the attenuation have different symmetry. Moreover, the symmetry directions may deviate from each other.

Martin

Simulating 4D seismic signal with noise

A method for simulation of 4D seismic response is presented and applied to data from the Ula Field. A spatial convolutional operator is estimated based on reservoir simulation and applied to the seismic data. In this way we obtain realistic 4D simulated data,and we can adjust noise level to study the effect of improved repeatability. For fields where the 4D signal is marginal (like Ula) this method has been useful to understand the influence of 4D noise on the data. So far the convolutional operator is derived assuming zero-offset reflectivity.

Jyoti

Fourier-Transform Rheology

Low frequency measurements of rock properties can be made using rheometers. From these dynamic measurements, the moduli and attenuation can be calculated using Fourier Transform.

10/10

Yuehua Zeng

Dynamic Processes of Fault Creep along the Chishang Fault, Taiwan

Active fault creep has been observed across one of the most active segments along the Longitudinal Valley Fault system, the present-day plate suture between the Eurasian and the Philippine Sea plates in eastern Taiwan. The data exhibit a distinct seasonal variation, with the fault creeping steadily in the rainy season and remaining locked during the dry season. We modeled the ground water induced Coulomb failure stress change on the creeping section of the fault. Our results suggest a partial hydraulic interconnection of the fault system to its surrounding structure.

Kurang

Virtual Source Gathers from field data - I

This presentation focuses on the virtual source gathers I generated using OBC data set provided by Shell. It will bring out the importance of source locations. Also I present my attempt to attenuate the free-surface multiples which is still work in progress.

Ivan

Seismic characterization of multiple fracture sets.

From previous studies based on effective medium theory and finite element computations, seismic signatures from a medium with complex fracture swarms can be described by only two sets of orthogonal, non-intersecting and non-interacting circular cracks. Using interval NMO ellipse parameters from pure-mode P-, S1- and S2-wave data, we can estimate two fracture densities, background velocities, and the fluid factor. We show a field data example from Rulison Field, CO. In this example there is a poor correlation between EUR data and estimated fracture densities. This suggests that the existence of a cross-property between elasticity and permeability is highly dependent on specific lithological models.

10/24

Roel

Combining seismic interferometry with multiple removal

In seismic interferometry one reconstructs the systems reponse from recorded incoherent signals. This reconstruction process can be carried out in different ways. Specifically, one can reconstruct the response of the system under different boundary conditions that the physical system has. This opens up the opportunity to replace a reflecting free surface by a surface that satisfies radiation boundary conditions. For seismic exploration this means that one can eliminate the reflections of the free surface. I will illustrate this principle with the motion recorded in the Millikan library.

Steve

Surface error characterization.

Improvements to initial research presented in last seminar, including error surfaces for sinc interpolator. Error comparisons vs. angle for sloping horizons.

Niran

Previous studies

1-D inversion of the seismic reflection data, multiples, interpretation of GPR data and earthquakes.

10/31

Dave

Computing amplitudes

Amplitudes of seismic waveforms are commonly estimated by computing mean absolute value, root-mean-square value, or peak value. How we choose among these three values depends on our application. For example, when computing amplitude ratios of noisy signals, all of these yield ratios that are biased towards one. Peak amplitude is least biased, but has the highest variance. If we choose peak amplitude, we should compute it carefully, by interpolation, to avoid systematic errors in amplitude maps.

Xiaoxia

Eikonal and Transport equations in anisotropic inhomogeneous media

I will review the Eikonal and the Transport equation in general anisotropic inhomogeneous media. Important insights can readily be drawn from those equations, for example, the vector reciprocity relation between group velocity and slowness, the relation between group and phase velocity, the geometrical spreading being the solution to the Transport equation, just to name a few. This talk presents part of my effort of putting together a systematic introduction to the theoretical background of my thesis.

Ivan

Active-shot SAFOD data

In an ongoing project with Roel, the objective is to image the San Andreas Fault at SAFOD. This particular dataset is comprised of a deviated, instrumented well with 178 3C receivers, courtesy of Paulsson Geophysical Services. We are currently studying the application of seismic interferometry by deconvolution to learn what information we can extract from a single active shot, placed at the top of the well. The data beautifully displays the reflection from the fault zone. Using the full wavefield at each receiver, we can represent the original wavefield with different boundary conditions by using deconvolution. Because of the acquisition geometry, we can separate the incident wavefield from the reflected wavefield. By deconvolving the full wavefield with the incident wavefield, all incident waves are collapsed to a spike at time 0 and we are left with only reflected waves. This deconvolved wavefield can be potentially used for extracting refeltivity properties of the fault zone.

11/07

SEG

Houston, Texas

Safe travels.

11/14

Jia

the AK semblance code by Debashish

I will show the AK semblance code working on isotropic and VTI media. How it outdoes the traditional semblance on polarity reversed reflections and how it fails for some kinds of anisotropic media.

Carlos

Velocity and seismic resolution: How are they related?

I'll discuss different views about seismic resolution and how it is related to issues of velocity resolution and size of the Fresnel zone.

Yaping

Seismic forward modeling of fractured rocks

Part of my summer work at ExxonMobil (with permission): 1) Calibration of ANRAY (an anisotropic ray-tracing code developed at SW3D, Charles Univ., Czech); 2) Azimuthal AVO effect from anisotropic overburden with the consideration of the fracture azimuth and the fracture dip.

11/16

Martin

TBA

TBA

Jyoti

Heavy Oils: Their shear story - Part I

Heavy oils can behave as solids or liquids or anywhere inbetween depending on the temperature and frequency. They can have significant shear wave dispersion and attenuation within the seismic band.

Kurang

Virtual Source Gathers from field data - II

Continuing my study on virtual source gathers generated using hydrophone component, here, I focus on the effect of tapering and using a larger aperture as compared to restricting the source locations to stationary phase locations. The virtual source gathers generated using vertical component shows strong events with low move-out velocity which are not to be seen in the pressure component.

11/21

Suzie

Nonlinear site response, where and when it will occur?

Nonlinear site effects on ground motion response spectra from both the Chi-Chi and Northridge earthquakes were investigated. We found that nonlinearity affects ground motions from both earthquakes significantly. At large distances, the average soil site shows higher amplitudes in spectra at all frequencies than that of rock sites, but at short distances, in the presence of nonlinearity, the high frequency part of the spectra shows smaller amplitudes on soil sites than that on rock sites. The crossover between response spectra of soil and rock sites occurs at about 1 Hz for both earthquakes. The threshold for nonlinearity to occur is at peak velocity of about 20 cm/s.

Martin

Seismic critical-angle reflectometry - a method to measure anisotropy?

This method has been used in ultrasonic applications for some years. The idea is to investigate whether similar techniques can be applied for seismic applications. For some reservoirs this might be a useful complementary tool for determining the anisotropic properties of the subsurface. A critical limiation is the ground roll type of noise that disturbs the picking of the critical anlge.However, some field data shows that it is possible to determine the critical offset by conventional RMS-analysis.

Xiaoxia

Geometrical-spreading correction of converted waves in anisotropic media

In this talk, I will discuss how to compute geometrical-spreading for converted waves from traveltime observed on the surface.

11/28

Roel

Seismic interferometry and attenuation

One of the ways to derive the extraction of the Green's function from the correlation of noise is based on the time-reversal invariance of the acoustic or elastic wave equation. This symmetry for time-reversal is broken by attenuation. I have tried to incorporate attenuation in the derivation of seismic interferometry and get stuck. I will show what the problem is, and ......... I need help!

Carlos

TBA

TBA

Matt Haney (Sandia National Laboratories)

Stability conditions for heterogeneous finite-difference schemes (Invited talk)

In the numerical analysis of finite-difference schemes, a Courant-Fredrichs-Levy (CFL) condition is usually calculated for the case of an infinitely-extending, homogeneous wholespace. Although this type of condition has been popular in determining the run parameters of full-waveform, finite-difference modeling codes, satisfying it at each point of a heterogeneous medium is no guarantee that the code will be stable. In this brief talk, I will describe a simple (i.e., 1D scalar wave equation) derivation that shows the necessary generalizations needed to make CFL conditions in heterogeneous media. Such conditions should be of practical use when performing large (weeks to months of runtime) finite-difference calculations in heterogeneous models and cannot afford to have the results plagued by numerical instabilities.

11/30

Ilya

Leaking waves in anisotropic media

Leaking modes are similar to conventional head waves but they are generated by the evanescent energy contained in point-source radiation. The strong influence of anisotropy parameters on the properties of evanescent plane waves (discussed in my previous seminar talk) indicates that leaking modes can help to constrain anisotropic velocity models using borehole data. I will present simple linearized expressions for the decay factors and polarization ellipses of leaking waves in VTI media. These solutions show that in some cases multicomponent recording of a single leaking mode can provide enough information for estimation of the relevant Thomsen parameters.

Jia

TBA

TBA

Niran

The Leaning Tower of Pisa

The Leaning Tower of Pisa is one of the seven wonders of Middle Ages. The tower was intended to stand vertically, but began leaning soon after construction started in 1173 What is the reason for leaning? What has been done to get under control?

12/05

Dave

Local estimates of orientations of features in images

Estimates of the orientations of features in digital images are useful in a variety of applications. With 3-D seismic images, for example, we may estimate the dip and azimuth of locally planar features, and then compute other attributes that depend on those estimates.

The structure (gradient-squared) tensor is a well-known tool for estimating orientation. For a continuous function, computing a single global structure tensor is equivalent to computing a quadratic approximation to that function's autocorrelation. This equivalance is easy to prove and it makes sense, but it breaks down for local estimates of orientations from sampled functions.

We wonder, why not simply compute local autocorrelations directly, without approximations? We highlight some properties of Gaussian windows that facilitate the efficient computation of autocorrelations of locally windowed sampled functions.

Steve

Interpolators: Depth vs. Amplitude, Categorizing Surfaces, Maximum Amplitude Extraction on Pinnacle Reefs

Amplitude interpolation is more important than depth when attempting to image structure between samples. Errors are periodic, meaning that no “catalog” of structures discerned by examining error surfaces is possible. Interpolator comparisons for “test-pattern” and synthetic pinnacle reef topography show that sinc interpolation returns accurate amplitudes from the reflector, especially from curved surfaces, where parabolic interpolator error runs up to approx. 4.5%. This means that the sinc interpolator outperforms the parabolic interpolator (and therefore other interpolators, ex. spline), allowing sharper definition of lateral extents, and brighter returns for low amplitude topography.

Jyoti

Dynamic mechanical analysis (DMA) of glass bead-epoxy composites

I will discuss a paper where the authors have used DMA to analyze the effect of interfacial adhesion between the matrix (epoxy) and inclusions (glass beads) on the mechanical properties of the composite as a function of frequency and temperature. Temperature sweep tests indicated that the value of attenuation at the temperature of the \alpha-relaxation peak (i.e. close to glass point) was most sensitive to interfacial adhesion.