Typically, air quality models are steady-state dispersion models, incorporating a simplistic description of the dispersion process and approximating plume behaviour. As a result, some of the fundamental assumptions made do not accurately reflect reality. For example, when calculating hourly ground-level concentrations, air pollutants released during the previous hours are not considered. Another example is that these models usually assume a minimum wind speed, as they ‘break down’ during low wind speed or calm conditions.
In order to improve estimates of ground-level concentrations of air pollutants, a new “box” model was developed.  The new air quality model was developed using a process based modelling approach. This ensures that the model simulates the detailed physical and biological processes that explicitly describe air pollutants dispersion and removal. Furthermore, a Compartment Flow/System Dynamics approach has been employed. This enables conceptualising the system in terms of compartments (stocks) and flows (processes). From a modelling point of view, the main advantage of this approach is that the graphical representation of the system variables and processes helps the model development within an interdisciplinary workgroup. From a mathematical point of view, the use of differential equations means that the air pollutant concentrations calculated in each moment of time depend on concentrations calculated before, allowing for the accumulation of pollutants to be considered even in calm weather conditions.
In order to develop the model, the key elements and processes involved were identified. Experts from various scientific disciplines were consulted to determine which processes are fundamental to capture the dynamics of the system. In essence, the following processes have been incorporated into the model:
Dry deposition of pollutants on vertical surfaces. Consequently, the air quality model takes into account the contribution of plants and vegetation in removing air pollutants, in particular particulate matter;
Wet deposition (through precipitation, for example because of rain);
Simplified chemical reactions that air pollutants undergo;
Movement of pollutants by simple diffusion; and
Movement of pollutants due to wind/air masses.
The model also includes 3D wind flow calculations, which take into account the geometry of buildings, considering advection and turbulent flows, as well as horizontal and vertical flows.
In terms of input, the model uses:
A key input to any air quality model is emissions of air pollutants. The model developed primarily considers traffic emissions (incorporated as line sources), as it has been assumed that air pollution in Portici is mainly from traffic.  However, Portici is the centre of an urban area within the Naples metropolitan area, thus there are other pollution sources within and outside Portici that affect local air quality. Consequently, urban accumulation and the regional background is also incorporated in the model. More specifically, data from the air quality monitoring stations are used to calibrate the model (with night measurements being the starting point of the simulation) in order to account for urban accumulation. The regional background is obtained from the Copernicus Atmosphere Monitoring Service (CAMS). 
Topography, land use and meteorology are important when determining the dispersion of air pollutants, their dry/wet deposition and removal by vegetation. Even more so for the City of Portici that is characterised by complex terrain and meteorological conditions, including sea and volcano breezes.
As such, the terrain of the city has been incorporated in the air quality model with data on vegetation and buildings, derived from a Digital Elevation Model,  along with data on the road network.
Meteorological conditions outside the city of Portici have been provided by the Weather Research and Forecasting model (WRF) to incorporate in the air quality model. More specifically, the model focuses in a 60km x 50 km domain centred on Portici, with a spatial resolution of 1 km x 1 km and an hourly temporal resolution. This uses amongst other the Global Forecast System (GFS), global meteorological data (NCEP), local meteorological data and satellite data. WRF is run once a day, producing four days weather forecasts that are then used in the air quality model, including data on temperature, wind speed, precipitation, relative humidity, and solar radiation.
In summary, the air quality model combines monitored and modelled data. Monitored data provide information on the current situation (including weather conditions, traffic density and air quality concentrations) and models are used to forecast traffic, air quality and assess the effectiveness of policy interventions.
The outputs of the model are two-dimensional maps of pollutant concentration forecasts at a height of 1.5 meters with a spatial resolution between 2 and 5 meters.
A future improvement that may be explored is the use of fixed monitoring stations to undertake real time calibration of the air quality model, including to forecasts as time passes