Keywords: cloud, grid, parallel computing, forecast systems, numerical models, optimal performance.

Computational forecast systems (CFS) are essential tools for daily and emergency coastal management by providing water dynamics predictions, through integration of numerical models and field data. The reliability of their predictions depends greatly on the parallel models' accuracy, but achieving the adequate resolution in spatial discretizations for port-to-ocean, multi-scale analysis is often hampered by computational costs.

Nowadays forecast systems are processed in dedicated workstations, fulfilling robustness and quality control through automatic comparison with field data and simulation redundancy. Recently, CFS has been ported to HPC grids with significant success, but requiring highly-specialized staff to maintaining them in these complex environments. The need to increase the available resources and to export CFS to coastal managers has promoted the search for simpler approaches. The scalability and flexibility of cloud resources, along with dedicated services for facilitating their use, makes them an attractive option.

In this paper, the performance of operational coastal forecast systems are assessed and compared for the first time in multiple environments, including local workstations, grids and a pilot cloud. The analysis is conducted in a range of resources from the physical processor number available at the cloud to the optimal number of processors for the specific cases.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £75 +P&P)