HPCN and Atmospheric Flow Problems

This project belongs to CWI's research project MAS1.1: Numerical algorithms for air quality modeling

Project Description:

The mathematical description of atmospheric transport, chemistry and exchange of trace constituents gives rise to a system of time-dependent partial differential equations (PDEs) of the advection-diffusion-reaction type. To simulate with such a model real-life atmospheric chemistry/transport processes on a temporal and spatial scale of interest, first of all efficient and robust numerical algorithms are needed. Computer capacity, however, is also a critical factor, both with respect to computer power and with respect to memory. This induces an unceasing quest for efficient implementations on the most advanced computer systems like vector/parallel supercomputers and massively parallel distributed-memory systems, and on (a cluster of) workstations. An important question is also to what extent a code, written in (a dialect of) HPF (High Performance Fortran), can be efficient on the different platforms.

The goal of this project is the development of software and algorithms for atmospheric flow problems on supercomputers, massively parallel systems and heterogeneous networks.

Subprojects:

NCF/CRG 1994
Algorithm and code development on the shared-memory, vector/parallel supercomputer CRAY C90 for the transport and chemistry of trace constituents in the troposphere.
Main topics:
1. Implementation of the 2D spherical transport equation.
2. The coupled solution of vertical diffusion and chemistry.
3. Implementation of a 3D regional advection-chemistry-diffusion model.
NCF 1995
The implementation of a 3D prototype of an atmospheric transport-chemistry model on the 256-node distributed-memory CRAY T3D of the EPFL at Lausanne.
NCF/CRG 1996
Code development on the shared-memory, vector/parallel supercomputer CRAY C90 for the CIRK model.
NCF/CRG 1997/1998
Implementational aspects of I/O, necessary in off-line AQM, on Cray T3E. AQM benchmark performance on various computer platforms.

People involved:

J.G. Blom
W.M. Lioen (now at DataDistilleries)
E.J. Spee (now at Rijksinstituut voor Kust en Zee)
Dr J.G. Verwer
Dr P.M. de Zeeuw

Go to the MAS1.1: Numerical algorithms for air quality modeling project overview, to the CWI Environmental Modelling and Porous Media Research (MAS1) program overview or to the CWI home page.

Joke Blom (gollum@cwi.nl)

Last update 98/02/19