Two models are used at IRCEL-CELINE to forecast particulate matter (PM10) concentrations :

OVL ovl The computer model OVL is a neural network model. This model generates particulate matter (PM10) concentrations based on meteorological forecasts and historic time series of measured air pollution.  The advantage of neural networks is the speed by which they can produce forecasts at specific locations.   The disadvantage is that forecasts are only possible for locations where sufficient historic air pollution measurements are available.  
More info concencerning the OVL model developped by the VITO can be found here.
CHIMERE chimere The CHIMERE model is a deterministic model that simulates the very complex fysical processes and chemical reactions in the atmosphere based on meteorological forecasts, emissions of air pollutants and geograhical landuse data. The advantage of deterministic models is the possibility to forecast air pollution also at locations where the air quality is not measured. The disadvantage are the complex input and the long computer calculation times (the last becomes less important due to increasing computer calculation power) .
The present resolution of the CHIMERE model is approximately 50x50 km, so the forecasted concentrations are representative for a large area. Real concentrations can be higher on the local scale (near industrial sources, main traffic roads, ...).
More info concencerning the (by IRCEL adapated) CHIMERE model can be found here.
(click on the images for the most recent forecasts)

Together with our own expertise, we use the OVL and CHIMERE model results to inform the public for episodes with enhanced air pollution (wintersmog). These episodes are characterised especially with increased concentrations of particulate matter (PM10, PM2.5) concentrations.

It is possible that the two models will procuce different forecasts. Interpreting the results is not always easy and they must be used carefully.