HR: 1330h
AN: A42B-0759    [PDF]
TI: 3D Staggered-Grid Finite-Difference Simulation of Acoustic Waves in Turbulent Moving Media
AU: * Symons, N P
EM: npsymon@sandia.gov
AF: Sandia National Laboratories, PO 5800, Albuquerque, NM 87185-0750 United States
AU: Aldridge, D F
EM: dfaldri@sandia.gov
AF: Sandia National Laboratories, PO 5800, Albuquerque, NM 87185-0750 United States
AU: Marlin, D
EM: dmarlin@arl.army.mil
AF: U.S. Army Research Laboratory, White Sands Missile Range, Las Cruces, NM 88004 United States
AU: Wilson, D K
EM: D.Keith.Wilson@erdc.usace.army.mil
AF: U.S. Army Cold Regions Research and Engineering Laboratory, 72 Lyme Rd., Hanover, NH 03755-1290 United States
AU: Sullivan, P
EM: pps@ncar.ucar.edu
AF: National Center for Atmospheric Research, UCAR, Boulder, CO 80301 United States
AU: Ostashev, V
EM: Vladimir.Ostashev@noaa.gov
AF: NOAA Environmental Technology Laboratory, NOAA, Boulder, CO 80310 United States
AB: Acoustic wave propagation in a three-dimensional heterogeneous moving atmosphere is accurately simulated with a numerical algorithm recently developed under the DOD Common High Performance Computing Software Support Initiative (CHSSI). Sound waves within such a dynamic environment are mathematically described by a set of four, coupled, first-order partial differential equations governing small-amplitude fluctuations in pressure and particle velocity. The system is rigorously derived from fundamental principles of continuum mechanics, ideal-fluid constitutive relations, and reasonable assumptions that the ambient atmospheric motion is adiabatic and divergence-free. An explicit, time-domain, finite-difference (FD) numerical scheme is used to solve the system for both pressure and particle velocity wavefields. The atmosphere is characterized by 3D gridded models of sound speed, mass density, and the three components of the wind velocity vector. Dependent variables are stored on staggered spatial and temporal grids, and centered FD operators possess 2nd-order and 4th-order space/time accuracy. Accurate sound wave simulation is achieved provided grid intervals are chosen appropriately. The gridding must be fine enough to reduce numerical dispersion artifacts to an acceptable level and maintain stability. The algorithm is designed to execute on parallel computational platforms by utilizing a spatial domain-decomposition strategy. Currently, the algorithm has been validated on four different computational platforms, and parallel scalability of approximately 85% has been demonstrated. Comparisons with analytic solutions for uniform and vertically stratified wind models indicate that the FD algorithm generates accurate results with either a vanishing pressure or vanishing vertical-particle velocity boundary condition. Simulations are performed using a kinematic turbulence wind profile developed with the quasi-wavelet method. In addition, preliminary results are presented using high-resolution 3D dynamic turbulent flowfields generated by a large-eddy simulation model of a stably stratified planetary boundary layer. Sandia National Laboratories is a operated by Sandia Corporation, a Lockheed Martin Company, for the USDOE under contract 94-AL85000.
DE: 3210 Modeling
DE: 3322 Land/atmosphere interactions
DE: 3379 Turbulence
SC: Atmospheric Sciences [A]
MN: 2003 Fall Meeting