AL TIME AIR QUALITY MODELING SYSTEMS FOR ROME METROPOLITAN AREA: DESCRIPTION AN D PRELIMINARY PERFOR MANCE EVALUATION

This study describes and evaluates two modelling systems developed to provide air quality forecasts and nowcasts over Rome metropolitan area and its surrounding. The systems have been implemented by the regional environmental protection agency (ARP to satisfy the air quality Directive 2008/50/CE meteorological fields that drive the Eulerian chemi cal transport model FARM. Industrial and domestic emission fluxes are base high resolution emission inventory while emissions from road traffic have been estimated by means of t raffic modelling. The f system is part of Chemical Weather Forecast Network, promoted by COST ES0602 Action, and provides 72 hours pred ARPA Lazio web site, allowing to identify possible exceedancees of EU air quality standards. The nowcasting system includes a assimilation module based on the Successive Correct ion Method (SCM) that considers O monitoring stations (industrial, urban, suburban an d rural) from the regional monitoring network have been used. Air quality available every 3 hours. A statistical analysis has been applied to evaluate the performance predict/reconstruct air pollution episodes. The eva luation has been based on standard air quality mode l evaluation indexes an systems show a good agreement with observed levels for the Rome met assimilation techniques, show a better performance when compared with experimental data suggesting indications on the more co way to get air quality assessment on the fly over R ome me conurbation claiming for an improvement of the emission inventory. : This study describes and evaluates two modelling systems developed to provide air quality forecasts and nowcasts over Rome metropolitan area and its surrounding. The systems have been implemented by the regional environmental protection agency (ARP Directive 2008/50/CE requirements. The meteorological model RAMS has been used to reconstruct 3D meteorological fields that drive the Eulerian chemi cal transport model FARM. Industrial and domestic emission fluxes are base high resolution emission inventory while emissions from road traffic have been estimated by means of t raffic modelling. The f system is part of Chemical Weather Forecast Network, promoted by COST ES0602 Action, and provides 72 hours pred ARPA Lazio web site, allowing to identify possible exceedancees of EU air quality standards. The nowcasting system includes a assimilation module based on the Successive Correct ion Method (SCM) that considers O 3, NO 2, Benzene, CO monitoring stations (industrial, urban, suburban an d rural) from the regional monitoring network have been used. Air quality available every 3 hours. A statistical analysis has been applied to evaluate the performance of the two systems and verify their ability to predict/reconstruct air pollution episodes. The eva luation has been based on standard air quality mode l evaluation indexes an systems show a good agreement with observed levels for the Rome met ropolitan areas. The Near Real Time (NRT) system, that uses data assimilation techniques, show a better performance when compared with experimental data suggesting indications on the more co way to get air quality assessment on the fly over R ome me tropolitan area. Main weaknesses emerged for the ru ral area surrounding Rome conurbation claiming for an improvement of the emission inventory. Air Quality, Forecasting system, Near Real Time sys tem, data assimilation, SCM, Rome. : This study describes and evaluates two modelling systems developed to provide air quality forecasts and nowcasts over Rome metropolitan area and its surrounding. The systems have been implemented by the regional environmental protection agency (ARP A Lazio) requirements. The meteorological model RAMS has been used to reconstruct 3D meteorological fields that drive the Eulerian chemi cal transport model FARM. Industrial and domestic emission fluxes are base d on a local high resolution emission inventory while emissions from road traffic have been estimated by means of t raffic modelling. The f orecasting system is part of Chemical Weather Forecast Network, promoted by COST ES0602 Action, and provides 72 hours pred ictions published on ARPA Lazio web site, allowing to identify possible exceedancees of EU air quality standards. The nowcasting system includes a data , Benzene, CO and SO 2 measurements. 34 monitoring stations (industrial, urban, suburban an d rural) from the regional monitoring network have been used. Air quality analyses are of the two systems and verify their ability to predict/reconstruct air pollution episodes. The eva luation has been based on standard air quality mode l evaluation indexes an d graphics. Both ropolitan areas. The Near Real Time (NRT) system, that uses data - assimilation techniques, show a better performance when compared with experimental data suggesting indications on the more co nvenient tropolitan area. Main weaknesses emerged for the ru ral area surrounding Rome , modelling is considered a powerful tool to assess and manage air quality (AQ). In Italy, to date, only few Regional E nvironmental Protection Agencies (ARPAs) have implemented models to ion coming from air quality monitoring networks and support the definition of measures to reduce healt h impact of air pollution. The aim of this work is to analyse the performances of two modelling systems implemented by orecast and nowcast over Rome metropolitan area and its surrounding. The systems are based on the Eulerian Chemical Transport Model (CTM) FARM (Flexible Air quality Regional Model, Silibello et al. ,2008) assimilation techniques (observational nudging and objective analysis: SCM) in order to estimate an atmospheric status as c lose as ng all the available information: observations, mod el results, The analysis of the results produced by the two sys tems has been performed by means of statistical ind exes such as bias, root rror and absolute error to quantify differences bet ween observations and model predictions regardless of observed or predicted concentrations levels, while other indexes defined as categorical indexes are us ed to measure ctly/incorrectly predicting concentration levels ab ove/below a certain threshold.