Numerical ragweed pollen forecasts using different source maps: a comparison for France.

Area of Science:

• Aerobiology
• Atmospheric Science
• Computational Modeling

Background:

• Accurate numerical pollen forecasts rely on precise pollen source distribution maps.
• Existing methods for creating these maps include plant inventories, land use data with pollen counts, and ecological modeling.

Purpose of the Study:

• To evaluate the applicability and usefulness of different pollen source distribution map methodologies in numerical pollen forecasting.
• To rank the performance of these maps using simulated ragweed pollen concentrations against observed data.

Main Methods:

• Simulated the 2012 ragweed pollen season in France using the COSMO-ART numerical weather prediction model.
• Employed six exemplary distribution maps from three distinct methodologies (plant inventories, land use data with pollen counts, ecological modeling).
• Statistically compared simulated pollen concentrations with measured values to assess map performance.

Main Results:

• The methodology combining land use data with annual pollen counts yielded the best agreement between simulated and observed pollen concentrations for 2012.
• Ecological modeling without sophisticated plant density estimation showed very low predictive skill.
• Performance of inventory maps and land use-based maps varied significantly by observational site.
• Calibrating maps with pollen counts substantially improved model performance.

Conclusions:

• The integration of land use data and pollen counts is a highly effective approach for generating pollen source distribution maps for forecasting.
• Ecological models require detailed plant density data to achieve reliable predictive skill.
• Map-based pollen forecasting models benefit significantly from calibration using measured pollen data.