G-AQFS (Grid Air Quality Forecast System): an experimental system based on GRID computing technologies to forecast atmospheric dispersion of pollutants

The emerging Grid technology offers the resources needed to perform complex atmospheric simulations. The virtualization of Grid resources such as CPU processing, network bandwidth and storage capacity allows creating a single system image to manage a Air Quality Forecast System. Aim of this work is the implementation of an integrated system for modelling air pollutants dispersion at regional scale using computational Grid technology. The system includes meteorological models, emission preprocessors and dispersion models. Both meteorological and dispersion models use for the initial and boundary conditions models which run at European scale. We have used a small experimental computational Grid for testing G-AQFS. Two meteorological models are coupled in cascade: RAMS, a mesoscale meteorological model, and the micrometeorological model CALMET. RAMS is initialized and driven using the ECMWF data (European Centre for Medium-Range Weather Forecasts). The emission data are computed following a top down approach, through disaggregation, modulation and splitting of CORINAIR national inventary, and a bottom up approach, for the industrial, road transport and biogenic emissions. The output provided by the meteorological and the emission models are used to drive the dispersion models CALPUFF and/or the photochemical model CALGRID. Initial and boundary conditions for concentrations are obtained by the Chimere model (www.prevair.org) which runs at continental scale. As a test the modelling system is applied over Salento Peninsula, located in the south-east corner of Italy, in the Mediterranean central area.