Numerical weather prediction (NWP) models have grown in complexity through the years. This growth has been driven by the availability of computers that have higher performance capacity. Most major forecast centres in the world have access to parallel supercomputers for operations and research. These supercomputers are very expensive and beyond the budgets of most university research groups and many developing countries. A lower cost option available to such research groups is to establish a cluster of workstations designed to work as a distributed memory parallel computing environment.
In this paper, we describe the eight workstation cluster at the Centre for Environmental Modeling and Prediction (CEMAP), University of New South Wales, Australia. The initial focus of the cluster is in research on tropical cyclone prediction. A tropical cyclone track forecast numerical prediction model has been developed and the performance of the model on the cluster has been evaluated. The model description and performance results are presented in this paper