Open AccessOn the parallelisation of weather and climate models
Author(s) -
Sukanta Kumar Dash
Publication year - 2001
Publication title -
mausam
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.243
H-Index - 12
ISSN - 0252-9416
DOI - 10.54302/mausam.v52i1.1687
Subject(s) - supercomputer , computer science , climate model , workstation , data assimilation , downscaling , field (mathematics) , weather forecasting , data science , distributed computing , climate change , parallel computing , meteorology , ecology , physics , mathematics , pure mathematics , biology , operating system
The numerical models used for weather forecasting and climate studies need very large computing resources. The current research in the field indicates that for accurate forecasts, one needs to use models at very high resolution, sophisticated data assimilation techniques and physical parameterisation schemes and multi-model ensemble integrations. In fact the spatial resolution required for accurate forecasts may demand computing power which is prohibitively high considering the processing power of a single processor of any supercomputer. During the last two decades, the developments in computing technology show the emergence of parallel computers with a number of processors which are capable of supplying enormously large computing power as against a single computer. Today, a cluster of workstations or personal computers can be used in parallel to integrate a global climate model for a long time. However, there are bottlenecks to be overcome in order to achieve maximum efficiency. Inter-processor communication is the key issue in case of global weather and climate models. The present paper aims at discussing the status of parallelisation of weather and climate models at leading centres of operational forecasting and research, the inherent parallelism in weather and climate models, the problems encountered in inter-processing communication and various ways of achieving maximum parallel efficiency.