An evolving multi agent system for meteorological alerts: A deployed pilot system

The Australian Bureau of Meteorology has a requirement for complex and evolving systems to manage its weather forecasting, monitoring and alerts. A needs analysis has indicated that due to the nature of the domain, where the environment is constantly changing and there are numerous different goals, intelligent multi agent systems technology is highly appropriate. This paper describes a pilot project that provides the basis on which the rest of the system will be built. The pilot system monitors in real time the current Terminal Area Forecasts (forecasts in areas around airports) and alerts forecasters to inconsistencies between these and observations obtained from the Automatic Weather Station (AWS) data. The contributions of the paper are a description of the overall architecture including legacy components, and the mechanisms that have been used to interface to legacy components; a description of an inferencing mechanism, available in recent versions of the JACK Intelligent Agents toolkit which has been particularly useful in some of the reasoning needed in this application; and a detailed description of the architecture for data sharing and data management. The pilot system is currently deployed and the project is underway to extend this to the much larger system.