3D acoustic wave modeling with a time-space domain dispersion-relation-based Finite-difference scheme

Similar documents
P032 3D Seismic Diffraction Modeling in Multilayered Media in Terms of Surface Integrals

Data-Driven Imaging in Anisotropic Media

Numerical Solution of the MRLW Equation Using Finite Difference Method. 1 Introduction

INTRODUCTION. Residual migration has proved to be a useful tool in imaging and in velocity analysis.

DESIGN OF THE DIE PROFILE FOR THE INCREMENTAL RADIAL FORGING PROCESS *

Experimental Design For Model Discrimination And Precise Parameter Estimation In WDS Analysis

lecture 37: Linear Multistep Methods: Absolute Stability, Part I lecture 38: Linear Multistep Methods: Absolute Stability, Part II

A Nonlinear Sparsity Promoting Formulation and Algorithm for Full Waveform Inversion

Probabilistic micro-earthquake location for reservoir monitoring and characterization

Comparison of Stability of Selected Numerical Methods for Solving Stiff Semi- Linear Differential Equations

NUMERICAL MODELLING OF THE TYRE/ROAD CONTACT

Numerical Studies of a Nonlinear Heat Equation with Square Root Reaction Term

e-companion ONLY AVAILABLE IN ELECTRONIC FORM

Fluid Substitution Model to Generate Synthetic Seismic Attributes: FluidSub.exe

Inspection; structural health monitoring; reliability; Bayesian analysis; updating; decision analysis; value of information

Bernoulli Wavelet Based Numerical Method for Solving Fredholm Integral Equations of the Second Kind

The Wilson Model of Cortical Neurons Richard B. Wells

SPATIAL AND TEMPORAL DISTRIBUTIONS OF MIGRATION IN BIO-RETENTION SYSTEMS

Comparison of Charged Particle Tracking Methods for Non-Uniform Magnetic Fields. Hann-Shin Mao and Richard E. Wirz

COS 424: Interacting with Data. Written Exercises

Analytical solution for the electric potential in arbitrary anisotropic layered media applying the set of Hankel transforms of integer order

P016 Toward Gauss-Newton and Exact Newton Optimization for Full Waveform Inversion

Interferometry in dissipative media: Addressing the shallow sea problem for Seabed Logging applications

Proc. of the IEEE/OES Seventh Working Conference on Current Measurement Technology UNCERTAINTIES IN SEASONDE CURRENT VELOCITIES

Using a De-Convolution Window for Operating Modal Analysis

Multi-Scale/Multi-Resolution: Wavelet Transform

SEISMIC FRAGILITY ANALYSIS

Non-Parametric Non-Line-of-Sight Identification 1

REPRODUCING KERNEL PARTICLE METHOD FOR CONDUCTION RADIATION INTERACTION IN 3D PARTICIPATING MEDIA

New Slack-Monotonic Schedulability Analysis of Real-Time Tasks on Multiprocessors

ABSTRACT INTRODUCTION

ANALYSIS ON RESPONSE OF DYNAMIC SYSTEMS TO PULSE SEQUENCES EXCITATION

AN EXPLICIT METHOD FOR NUMERICAL SIMULATION OF WAVE EQUATIONS

Study of radiation in spherical media using moment method

Dynamic Response of SDOF System: A Comparative Study Between Newmark s Method and IS1893 Response Spectra

16.333: Lecture # 7. Approximate Longitudinal Dynamics Models

Ch 12: Variations on Backpropagation

CHAPTER 19: Single-Loop IMC Control

Extension of CSRSM for the Parametric Study of the Face Stability of Pressurized Tunnels

Ph 20.3 Numerical Solution of Ordinary Differential Equations

REDUCTION OF FINITE ELEMENT MODELS BY PARAMETER IDENTIFICATION

Anisotropic reference media and the possible linearized approximations for phase velocities of qs waves in weakly anisotropic media

Theory and Applications of the Indoor Noise Module (VDI 3760)

ANALYTICAL INVESTIGATION AND PARAMETRIC STUDY OF LATERAL IMPACT BEHAVIOR OF PRESSURIZED PIPELINES AND INFLUENCE OF INTERNAL PRESSURE

A remark on a success rate model for DPA and CPA

A Self-Organizing Model for Logical Regression Jerry Farlow 1 University of Maine. (1900 words)

Q ESTIMATION WITHIN A FORMATION PROGRAM q_estimation

Numerical Study of Tsunami Propagation in Mentawai Islands West Sumatra

An Alternative Discretization and Solution Procedure for the Dual Phase-Lag Equation

Some Perspective. Forces and Newton s Laws

Analyzing Simulation Results

A Boundary Element Model for Underwater Acoustics in Shallow Water

POD-DEIM MODEL ORDER REDUCTION FOR THE MONODOMAIN REACTION-DIFFUSION EQUATION IN NEURO-MUSCULAR SYSTEM

MAKE SURE TA & TI STAMPS EVERY PAGE BEFORE YOU START

Model Fitting. CURM Background Material, Fall 2014 Dr. Doreen De Leon

Research Article Rapidly-Converging Series Representations of a Mutual-Information Integral

A Simplified Analytical Approach for Efficiency Evaluation of the Weaving Machines with Automatic Filling Repair

ANALYSIS OF HALL-EFFECT THRUSTERS AND ION ENGINES FOR EARTH-TO-MOON TRANSFER

Projectile Motion with Air Resistance (Numerical Modeling, Euler s Method)

Recovering Data from Underdetermined Quadratic Measurements (CS 229a Project: Final Writeup)

Intelligent Systems: Reasoning and Recognition. Perceptrons and Support Vector Machines

Grid performance models using Design of Experiments (DoE) methods

An improved self-adaptive harmony search algorithm for joint replenishment problems

A Better Algorithm For an Ancient Scheduling Problem. David R. Karger Steven J. Phillips Eric Torng. Department of Computer Science

Department of Electronic and Optical Engineering, Ordnance Engineering College, Shijiazhuang, , China

Multi-Window Post-Doppler Dimensionality Reduction for Multi-Waveform STAP

Ştefan ŞTEFĂNESCU * is the minimum global value for the function h (x)

Suppress Parameter Cross-talk for Elastic Full-waveform Inversion: Parameterization and Acquisition Geometry

Qualitative Modelling of Time Series Using Self-Organizing Maps: Application to Animal Science

N-Point. DFTs of Two Length-N Real Sequences

Stochastic Subgradient Methods

Ocean 420 Physical Processes in the Ocean Project 1: Hydrostatic Balance, Advection and Diffusion Answers

Adapting the Pheromone Evaporation Rate in Dynamic Routing Problems

Support Vector Machine Classification of Uncertain and Imbalanced data using Robust Optimization

A DESIGN GUIDE OF DOUBLE-LAYER CELLULAR CLADDINGS FOR BLAST ALLEVIATION

Anton Bourdine. 1. Introduction. and approximate propagation constants by following simple ratio (Equation (32.22) in [1]):

Feature Extraction Techniques

Analysis of ground vibration transmission in high precision equipment by Frequency Based Substructuring

An Improved Particle Filter with Applications in Ballistic Target Tracking

Stability Analysis of the Matrix-Free Linearly Implicit 2 Euler Method 3 UNCORRECTED PROOF

This model assumes that the probability of a gap has size i is proportional to 1/i. i.e., i log m e. j=1. E[gap size] = i P r(i) = N f t.

Numerically repeated support splitting and merging phenomena in a porous media equation with strong absorption. Kenji Tomoeda

SPECTRUM sensing is a core concept of cognitive radio

Efficient dynamic events discrimination technique for fiber distributed Brillouin sensors

Celal S. Konor Release 1.1 (identical to 1.0) 3/21/08. 1-Hybrid isentropic-sigma vertical coordinate and governing equations in the free atmosphere

Spine Fin Efficiency A Three Sided Pyramidal Fin of Equilateral Triangular Cross-Sectional Area

Role of Control-Structure Interaction in Protective System Design

An Approximate Model for the Theoretical Prediction of the Velocity Increase in the Intermediate Ballistics Period

Sensorless Control of Induction Motor Drive Using SVPWM - MRAS Speed Observer

Direct numerical simulation of matrix diffusion across fracture/matrix interface

Pattern Recognition and Machine Learning. Learning and Evaluation for Pattern Recognition

Optimization of ripple filter for pencil beam scanning

Bootstrapping Dependent Data

A NEW ROBUST AND EFFICIENT ESTIMATOR FOR ILL-CONDITIONED LINEAR INVERSE PROBLEMS WITH OUTLIERS

Effective joint probabilistic data association using maximum a posteriori estimates of target states

GREY FORECASTING AND NEURAL NETWORK MODEL OF SPORT PERFORMANCE

On Poset Merging. 1 Introduction. Peter Chen Guoli Ding Steve Seiden. Keywords: Merging, Partial Order, Lower Bounds. AMS Classification: 68W40

A model reduction approach to numerical inversion for a parabolic partial differential equation

Hyperbolic Horn Helical Mass Spectrometer (3HMS) James G. Hagerman Hagerman Technology LLC & Pacific Environmental Technologies April 2005

ANALYSIS OF REFLECTOR AND HORN ANTENNAS USING MULTILEVEL FAST MULTIPOLE ALGORITHM

Transcription:

P-8 3D acoustic wave odeling with a tie-space doain dispersion-relation-based Finite-difference schee Yang Liu * and rinal K. Sen State Key Laboratory of Petroleu Resource and Prospecting (China University of Petroleu, Beijing), China The Institute for Geophysics, John A. and Katherine G. Jackson School of Geosciences, The University of Texas at Austin, U.S.A. Suary Spatial finite-difference (FD) stencils designed in the space doain are usually eployed in wave equation odeling. In this paper, we adopt a spatial FD stencil which is devised in the tie-space doain based on a dispersion relation to iprove the accuracy of FD odeling for 3D acoustic wave equation. In addition, we eploy a new hybrid absorbing boundary condition (ABC) to eliinate edge reflection due to finite coputational doain. Dispersion analysis and odeling results deonstrate that the tie-space doain dispersion-relation-based spatial FD stencils can indeed iprove the odeling accuracy. The odeling accuracy can be iproved further by using a truncated FD ethod. The new hybrid ABC can effectively absorb boundaries reflections. The odeling schee presented in this paper is ore accurate and efficient than the conventional one and can be used routinely. Introduction The finite-difference (FD) ethod is an iportant ethod for the nuerical solution of partial differential equations and has been widely utilized in seisic odeling and igration (e.g., Kelly et. al, 976; Dablain, 986; Liu and Sen, 9a). A high-order spatial FD is a coon approach to increase odeling accuracy (e.g., Dablain, 986; Fornberg, 987; Etgen and O Brien, 7). A loworder FD algorith uses a shorter operator but needs ore grid points for discretization. A high-order FD algorith uses a longer operator but needs fewer grid points. Since a conventional explicit high-order teporal FD is usually unstable in the wave equation odeling, spatial derivatives are used to replace high-order teporal derivatives (e.g., Dablain, 986) to increase the accuracy of teporal derivatives with additional coputational cost. Generally, ost FD ethods deterine the FD stencils for spatial derivatives only in the space doain. However, the seisic wave propagation calculation is done both in space and tie doains. To iprove the accuracy of conventional FD ethods, a new tie-space doain FD ethod was proposed to derive the spatial FD coefficients in the joint tie-space doain (Finkelstein and Kastner, 7). The key idea of the ethod is that the dispersion relation is copletely satisfied at designated frequencies and thus the spatial FD coefficients are frequency dependent. This ethod was developed further for D, D and 3D acoustic wave odeling using a plane wave theory and the Taylor series expansion (Liu and Sen, 7b). These new spatial FD coefficients, dependent on the Courant nuber and space point nuber, are frequency independent though they lead to a frequency dependent nuerical solution. In this paper, we report on the nuerical results in 3D using FD that replaces the conventional 3D spatial FD coefficients with these new coefficients to iprove the odeling accuracy without increasing the calculation aount. A truncated FD ethod is also used to enhance the odeling accuracy further. A hybrid absorbing boundary condition (ABC) is also developed for 3D acoustic odeling. Conventional finite-difference schee for the 3D acoustic wave equation We start with the 3D acoustic wave equation given by p p p p + + =, () x y z v t * China University of Petroleu, State Key Laboratory of Petroleu Resource and Prospecting, Changping, Beijing, 49, China. Eail: wliuyang@vip.sina.co.

3D Acoustic odeling with Dispersion-relation-based FD ethod Where, p = p( x, y, z, t) is a scalar wave field, and v is velocity. The following nd-order FD is usually used for the tie derivatives, p δ p = p,, + ( p,, + p,, ) t δt τ, () where, n p, l, j = p( x + h, y + lh, z + jh, t + nτ ), (3) τ is tie step, h is grid size. Generally, the odeling accuracy is iproved by high-order FD on the space derivatives given by p δ p p,, a ( p,, p,, ) x δ x h + +, (4a) p δ p p,, a ( p,, p,, ) y δ y h + +, (4b) p δ p p,, a ( p,, p,, ) z δ z h + +. (4c) A Taylor series expansion is generally used to deterine the FD coefficients (Fornberg, 987; Dablain, 986). The FD coefficients in Equations (4a), (4b) and (4c) are as follows (Liu and Sen, 9a) a a ( ) + n = ( =,,..., ), (5a) n < n, n a =. (5b) It can be proved that when a -order FD space stencil and a nd-order FD tie stencil are used to solve the 3D acoustic wave equation, nuerical odeling has nd-order accuracy. It is noteworthy that increasing ay help reduce the agnitude of the conventional FD error without increasing the accuracy order. Tie-space doain dispersion-relation-based finitedifference schee for the 3D acoustic wave equation For the tie-space doain dispersion-relation-based FD schee, spatial FD coefficients can be derived by using the following steps (Liu and Sen, 9b). First, derive dispersion relation of FD odeling by using plane wave theory and Equations ()-(4). Second, apply the Taylor series expansion for trigonoetric functions in the dispersion relation equation. Then, copare coefficients of power series and obtain the following equations ( cos θ cos φ cos θ sin φ sin θ ) + + a = r j j j j j j j ( j =,,..., ), (6) where, r = vτ / h, θ is the plane wave propagation angle easured fro the horizontal plane perpendicular to z axis, φ is the aziuth of the plane wave. Note that these equations indicate that the coefficients a are a function of θ and φ. We solve Equation (6) to obtain a by using an optial direction with θ = and φ = π /8 (Liu and Sen, 9b), which akes FD odeling attain the highest ()th-order accuracy along 48 directions. Absorbing boundary condition Here, we adopt the new hybrid schee developed by Liu and Sen (9c) to absorb reflections fro the odel boundaries in nuerical solutions of wave equations (WEs). This schee divides the coputational doain into boundary, transition and inner areas. The wavefields within the inner and the boundary areas are coputed by the WE and the one-way wave equation (OWWE) respectively. The wavefields within the transition area are deterined by a weighted cobination of the wavefields coputed by WE and the OWWE to obtain a sooth variation fro the inner area to the boundary via the transition zone. We develop this hybrid ABC for 3D acoustic wave odeling. For 3D odeling, boundaries include six sides, twelve edges and eight corners. Second-order OWWEs are adopted for these sides, and first-order OWWEs for these edges and corners (Clayton and Engquist, 977). Ten grid points are used to define the transition area to absorb boundary reflections. Dispersion analysis The dispersion of FD for 3D odeling is described by the following expression (Liu and Sen, 9b), ( kh θ ) ( ) ( kh θ φ ), + sin cos cos / (7) sin sin / vfd sin sin δ = = r a + kh cos θ sin φ / v rkh

3D Acoustic odeling with Dispersion-relation-based FD ethod Where, k is the wavenuber. If δ equals, there is no dispersion. If δ is far fro, a large dispersion will occur. Since kh is equal to π at the Nyquist frequency, when calculating δ, kh only ranges fro to π. Figure illustrates the variation of dispersion paraeter δ ( φ, θ ) with kh along nine directions, which deonstrates that the accuracy of the tie-space ethod is greater than that of the conventional ethod. v FD /v v FD /v....99.98..5..5.....99 kh (a) Conventional ethod.98..5..5. kh Exact (, ) (, π/8) (, π/8) (π/8, π/8) (π/8, π/8) (π/8, 3π/8) (π/8, π/8) (π/8, π/8) (π/8, 3π/8) Exact (, ) (, π/8) (, π/8) (π/8, π/8) (π/8, π/8) (π/8, 3π/8) (π/8, π/8) (π/8, π/8) (π/8, 3π/8) (b) Tie-space ethod Figure : Plot of dispersion curves of the conventional and the tie-space FD ethods for 3D acoustic wave equation odeling. =, v = 3/s, τ =.s, h =. Nuerical odeling exaples First, both the conventional and the tie-space FD ethods are used to siulate 3D acoustic wave propagation in a hoogeneous acoustic ediu under the sae discretization. The odel and siulation paraeters are listed in the caption of Figure. Coputed snapshots respectively by the conventional and the tie-space FD ethods are shown in Figures (a) and (b). Coparing these two figures, we can see that the tie-space ethod aintains the wavefor better than the conventional ethod and thus has greater precision. To iprove the odeling accuracy further, we introduce the truncated FD ethod into the tie-space FD ethod. It is known that with the increase of the nuber of grid points involved in FD discretization, the accuracy increases but the coputational cost also increases. Liu and Sen (9d) found that there exist soe very sall coefficients for highorder FD coefficients and with the increase of order the nuber of these sall coefficients increases but their values decrease sharply. They also deonstrated that oitting these sall coefficients can aintain approxiately the sae level of accuracy of FD but reduce coputational cost significantly. Figure (c) shows snapshots by the truncated tie-space ethod, it follows that the dispersion decreases further copared with Figure (b). Next, we adopt the truncated tie-space FD ethod to perfor nuerical odeling. Figure 3 displays the seisogras coputed for a horizontally layered acoustic odel with the Clayton-Engquist ABC (Clayton and Engquist, 977) and the new hybrid ABC. The odel and siulation paraeters are given in the figure caption. The figure suggests that the boundary reflections are still strong copared with the reflections fro the true reflectors for the Clayton-Engquist ABC. The new hybrid ABC achieves nearly perfect absorption. Conclusions We have developed a tie-space doain dispersionrelation-based FD schee cobined with the truncated FD ethod and the hybrid ABC for 3D acoustic wave equation odeling. Dispersion analysis and nuerical odeling results deonstrate that this schee has greater accuracy and can effectively suppress dispersion and boundary reflections. Acknowledgents This research is partially supported by NSFC under contract No. 48399 and the National 863 Progra of China under contract No. 7AA6Z8. 3

3D Acoustic odeling with Dispersion-relation-based FD ethod (a) Conventional ethod (a) Snapshots at t=5s by the Clayton-Engquist ABC (left) and (b) Tie-space ethod (b) Snapshots at t=7s by the Clayton-Engquist ABC (left) and (c) Snapshots at t=s by the Clayton-Engquist ABC (left) and 7 3 Trace no. 7 3 7 3 9 Tie (s) (c) Truncated tie-space ethod Figure : Snapshots coputed by finite-difference odeling for a 3D hoogeneous acoustic odel respectively with conventional, tie-space and truncated tie-space FD ethods. Each figure includes panels with snapshots at 5s and 5s fro left to right. The odel velocity is 3/s, grid size is, grid diensions are, grid point coordinates range fro (,,) to (99,99,99), tie step is s, =. The sae discretization is involved in all the three ethods, but their spatial FD coefficients are different. A source pulse of 5Hz sine function with one period length is located at the center of the odel. The free surface condition and ABC are not included here. Top, front and right surfaces of snapshots are recorded at z=, y=5 and x=99 respectively. 4 6 8 (d) Seisogras with the Clayton-Engquist ABC 4

3D Acoustic odeling with Dispersion-relation-based FD ethod Tie (s) 4 6 8 Trace no. 7 3 7 3 7 3 9 (e) Seisogras with the hybrid ABC Figure 3: Snapshots and seisogras coputed by truncated tiespace FD odeling for 3D horizontally layered acoustic odel respectively with the Clayton-Engquist ABC and the hybrid ABC. Each figure of seisogras includes 3 receiver lines. The odel has 6 layers, whose velocities are velocity are 5/s, 3/s, 35/s, 3/s, 36/s and 4/s fro shallow to deep, 5 interfaces depth are 3, 5, 8, and 4. Grid size is, grid diensions are, grid coordinates range fro (,,) to (99,99,99), tie step is s. =. A source pulse of Hz sine function with one period length is located (99, 99, ). Receivers are located on the surface, and 3 receiver lines shown in this figure are located at y=99, 49 and 99 fro left to right. Trace no. of seisogra for each receiver line ranges fro to. Traces whose no. varies fro to 9 are shown since grids of width are used for the hybrid ABC. The free surface condition is included here. Top, front and right surfaces of snapshots are recorded at z=, y= and x=99 respectively. References Clayton, R. W., and B. Engquist, 977, absorbing boundary conditions for acoustic and elastic wave equations: Bulletin of the Seisological Society of Aerica, 6, 59 54. Dablain,. A., 986, the application of high-order differencing to the scalar wave equation: Geophysics, 5, 54 66. Etgen, J. T., and. J. O Brien, 7, Coputational ethods for large-scale 3D acoustic finite-difference odeling: A tutorial: Geophysics, 7, S3 S3. Fornberg B., 987, the pseudospectral ethod - coparisons with finite differences for the elastic wave equation: Geophysics, 5, 483-5. Finkelstein B., and R. Kastner, 7, Finite difference tie doain dispersion reduction schees: Journal of Coputational Physics,, 4 438. Kelly, K. R., R. Ward, W. S. Treitel, and R.. Alford, 976, Synthetic seisogras: A finite-difference approach: Geophysics, 4, 7. Liu Y., and. K. Sen, 9a, A practical iplicit finitedifference ethod: exaples fro seisic odeling: Journal of Geophysics and Engineering, 6, 3-49. Liu Y., and. K. Sen, 9b, A new tie-space doain high-order finite-difference ethod for the acoustic wave equation, Journal of Coputational Physics, doi:.6/j.jcp.9.8.7 Liu Y., and. K. Sen, 9c, A hybrid schee for absorbing edge reflections in nuerical odeling of wave propagation: Geophysics, in revision. Liu Y., and. K. Sen, 9d, Nuerical odeling of wave equation by a truncated high-order finite-difference ethod: Earthquake Science, (): 5-3. 5