Loading...
Please wait, while we are loading the content...
Similar Documents
Finite difference and spectral models for numerical weather forecasting on a massively parallel computer. (1993).
| Content Provider | CiteSeerX |
|---|---|
| Author | Cats, Gerard Gustafsson, Nils Wolters, Lex |
| Abstract | A computationally intensive part in a model for numerical weather forecasting solves a set of partial differential equations. There are several techniques to obtain this solution numerically. In this paper we will discuss the implementation of a numerical weather forecasting model on a massively parallel architecture using two techniques: a finite difference (gridpoint) method and a spectral method. A comparison between the two methods based on their actual performance will be presented. Besides the price/performance ratio of several compute platforms for this forecast model will be discussed. 1 Introduction Numerical models of the atmosphere have much contributed to our general understanding of atmospheric processes. The use of such models has resulted in improved weather forecasts, with important economical impact; also these models are now being used as components in climate simulation models. The horizontal and vertical resolution of atmospheric models is an important factor deter... |
| File Format | |
| Publisher Date | 1993-01-01 |
| Access Restriction | Open |
| Subject Keyword | Finite Difference Numerical Weather Forecasting Massively Parallel Computer Spectral Model Numerical Weather Introduction Numerical Model Partial Differential Equation Atmospheric Process Forecast Model Important Factor Deter Actual Performance Price Performance Ratio Climate Simulation Model Improved Weather Forecast Atmospheric Model Important Economical Impact Several Technique Vertical Resolution Spectral Method Intensive Part Several Compute Platform Parallel Architecture General Understanding |
| Content Type | Text |
| Resource Type | Article |
