An Enhanced Global Spectral Model

Abstract

This study describes the development of a vectorized, multiprocessing global spectral model (GSM) with enhanced physical parameterizations. The starting point was the Phillips Laboratory GSM, in its GL90 version, containing an enhanced suite of physical parameterizations. The latitude tasking scheme for multiprocessing the loop over latitude in the calculation of the spectral tendencies and adjusted model variables was implemented, using the general truncation version of the hydrodynamics code. Wavenumber calculations were vectorized over wavenumber and multiprocessed over vertical level. All gridpoint calculations were vectorized over longitude, and the physics packages were brought into closer compliance with plug-compatibility rules. Speedups due to the optimization were demonstrated in single- and multiprocessing timing tests on a dedicated Cray 2 and Cray Y-MP. The enhanced physics GSM was evaluated in forecast tests, and contrasted with the simpler physics GSM used at GWC. Minimization of errors in the computation of the horizontal pressure gradient force was investigated, using a perturbation temperature instead of full temperature in the integration of the hydrostatic equation. Several schemes were tested in an idealized model atmosphere.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Feb 27, 1992
Accession Number
ADA251242

Entities

People

  • Jean-francois Louis
  • Marina Zivkovic
  • Ross Hoffman
  • Thomas Nehrkorn

Organizations

  • Atmospheric and Environmental Research, Inc

Tags

Communities of Interest

  • Energy and Power Technologies
  • Human Systems
  • Space

DTIC Thesaurus Topics

  • Atmospheric Sciences
  • Boundary Layer
  • Computations
  • Computer Programming
  • Computer Programs
  • Computers
  • Convection
  • Diffusion Coefficient
  • Equations
  • Grids
  • Heat Energy
  • Latitude
  • Longitude
  • Operating Systems
  • Physics
  • Stratified Fluids
  • Weather Forecasting

Readers

  • Atmospheric Science/Meteorology
  • Finite Element Method (FEM) for solving Partial Differential Equations (PDEs)
  • Parallel and Distributed Computing.