A methodology for seasonal photochemical model simulation assessment
This paper presents a methodology to evaluate the model-skill assessment in simulating seasonal periods. In the proposed approach, the model evaluation is performed by using statistical and graphical methods. Such indexes and graphics are computed and processed for a limited number of representative domain stations, selected by a procedure based on the Cluster Analysis. Cluster Analysis, processing pollutant time series monitored from the air quality networks, detects similarity in terms of concentration levels and trends. For each homogeneous pattern group, Cluster Analysis allows the selection of the most representative station of the group that best summarises cluster behaviour and peculiarities; otherwise, a virtual station, obtained by an hour-by-hour averaging of the measurements recorded in the stations belonging to the group, can be determined. This methodology has been applied and tested in assessing the seasonal simulations performed by GAMES modelling system over Northern Italy domain in 1999 during the summer season, in the framework of the CityDelta Project.
Keywords: photochemical pollution, seasonal simulation, performance evaluation, clustering process, representative stations, virtual stations, Northern Italy, cluster analysis, pollutant time series, air quality, air pollution, pollutant concentration, atmospheric dispersion modelling