A Data Parallel Hirlam Forecast Model

• Main Authors: Lex Wolters, Robert Van Engelen, High Performance, Computing Division, Royal Netherlands, Nils Gustafsson, Tomas Wilhelmsson
• Other Authors: The Pennsylvania State University CiteSeerX Archives
• Format: Text
• Language: English
• Published: 1994
• Subjects: Norway; Iceland
• Online Access: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.24.1242; http://www.liacs.nl/TechRep/1994/tr94-47.ps.gz

This paper describes the current status of our research on data parallel implementations of the complete HIRLAM forecast model with both gridpoint and spectral dynamics, on massively parallel architectures. It is part of a research project, that investigates the possibilities to apply parallel computer systems for routine numerical weather forecasting. 1 Introduction In our research to investigate the (dis)advantages and the (in)efficiency of data parallel implementations of the HIRLAM forecast model on massively parallel computer systems, we have achieved new results compared to the situation described in [9]. This paper summarizes these results, of which some are preliminary. The HIRLAM system [3, 6] was developed by the HIRLAM-project group, a cooperative project of Denmark, Finland, Iceland, Ireland, The Netherlands, Norway, and Sweden. The model, which is used for producing limited area numerical forecasts, is in use at several of the meteorological services participating in the .