Scaling Down Modelled Airborne Birch and Grass Pollen Levels

Allergenic airborne pollen in Europe affect the health of a quarter of the adult population and a third of all children badly. Due to climate change even more people might suffer from pollen allergies in the future. If timely information on forthcoming pollen episodes is available, however, mitigation measures can be taken for easing off the allergy symptoms. This requires forecasting systems at the scale of the citizens that may alert people who are vulnerable for these pollen. In order to achieve this, we aim at providing modelled and forecasted airborne birch and grass pollen levels near the surface at the one by one kilometer scale.

The pollen transport model SILAM (System for Integrated modeLling of Atmospheric coMposition) is used as backbone for modelling and forecasting airborne birch and grass pollen in Belgium. SILAM is driven by ECMWF ERA5 meteorology in a bottom-up emission approach. The dynamic vegetation component in the pollen transport model is determined by pollen emission source maps which have to be ingested every pollen season in the model. To date, these maps have 0.10° x 0.10° and 0.05° x 0.05° gridcells for birch trees and grasses, respectively. Here, we combine monthly MODIS Land Surface Temperature (LST) data on a one by one kilometer grid with vegetation maps from earlier research on top of a pollen footprint analysis. We apply daily pollen footprints produced by SILAM running in a 3-day backward mode for five locations in Belgium, coupling the fraction of air to the pollen levels monitored by the devices of the aerobiological network. The down-scaled pollen emission source maps are then applied into SILAM in the forward mode to obtain modelled birch and grass pollen concentrations near the surface for Belgium more tailored towards the scale of citizens.

Preliminary analysis indicates that late winter/early spring MODIS LST is a good proxy of the severity of the grass pollen season. The added value of LST for the birch pollen season is small. By ingesting the new maps with down-scaled pollen emissions sources into SILAM and by comparing modelled and measured time series for the 2013-2018 pollen seasons a substantial improvement (up to 210% increase in R² values for grass pollen) is found for the monitoring stations, especially at the North Sea side. This can be mainly attributed to a better separation between sea and land characteristics in the 0.01 x 0.01° grid of the pollen emission source maps compared to the coarser native gridcells.