Development of a user friendly AERMOD interface and evaluation of model performance on a case study in complex environments with multiple sources

A SOX emission source (API Raffineria, Falconara Marittima, Italy) is analyzed using AERMOD air dispersion modelling system, for the purpose of demonstrating newly developed software (CAIRO for AERMOD), made to compile and visualize input, and run analyses. CAIRO for AERMOD (Compile AERMAP, AERMOD and AERPLOT Input and Run Output) is a python-based GUI aimed at streamlining the process of making complex input files, with unique syntax and running them, while obviating the need to manually write input and run the program through Windows Shell. The Windows application features automated features such as the input and conversion of coordinates to UTM through copy operations, while the input is visualized in Google Earth. Input is done through user interface and automatically compiled into correct syntax and project/file structure. It supports point or polygon sources up to 3 averaging periods, maxtable, maxifile, rankfile and plotfile, while the elevation and meteorological data must be third party. A review of air pollution, legislative, planetary boundary layer processes and AERMOD model formulation, introduced the analysis on SOX emissions of 15 point sources in a refinery located near domestic areas. The AERMOD output and real data of 3 monitoring stations, dedicated to monitoring the plant, were further processed using multiple methods to deduce the difference of modelled values compared to actual data. The model performed with an average difference of -1.84 μg SOX/m3 or 5.33%, compared to actual data. The model performance was acceptable by Normalized Mean Square Error (NMSE), Mean Fractional Bias (FB) and Mean Bias (MB) tests to comply with European legislation. Longer averaging periods (month and year) had smaller maximal deviations but on average it exhibited the same deviations as the shorter period (24h), which had up to ± 78% discrepancies from real data. Compared to the regulatory limit it’s a relative deviation of ±3.20% of the regulatory limit on average.