Modelling grass pollen levels in Belgium.

Biogenic aerosols such as airborne grass pollen affect the public health badly by putting additional distress on people already suffering from cardiovascular and respiratory diseases. In Belgium, daily airborne pollen concentrations are monitored offline at a few sites only, hampering the timely coverage of the country and short-term forecasts. Here we apply the Chemistry Transport Model SILAM to the Belgian territory to model the spatio-temporal airborne grass pollen levels near the surface based on bottom-up inventories of grass pollen emissions updated with the Copernicus land monitoring Service grassland map of 2015. Transport of aerosols in SILAM is driven by ECMWF ERA5 meteorological data. The emitted grass pollen amounts in SILAM are computed by the multiplication of the grass pollen source map with the release rate determined by the seasonal shape production curve during the grass flowering period. The onset and offset of this period follow a location-dependent prescribed calendar days. Here we optimize the grass pollen seasonal start and end in SILAM by comparing a 2008-2018 time series of daily airborne grass pollen concentrations from the Belgian aerobiological surveillance network with the simulations. The effect of the spatial distribution of grass pollen sources is quantified by constructing pollen source-receptor relations using model simulations with varying grass pollen emissions in five areas of the model domain as input. Up to 33% of the airborne grass pollen in one area was transport from others areas inside Belgium. Adjusting the start and end of the grass pollen season improved the model performance substantially by almost doubling the correlation with local observations. By introducing the temporal scaling of the inter-seasonal pollen amounts in the model, an additional R increase up to 22% was obtained. Further improvements can be made by including more detailed grass pollen sources and more dynamic start and end dates of the pollen season.