A numerical model of birch pollen emission and dispersion in the atmosphere. Model evaluation and sensitivity analysis

An evaluation of performance of the System for Integrated modeLling of Atmospheric coMposition (SILAM) in application to birch pollen dispersion is presented. The system is described in a companion paper whereas the current study evaluates the model sensitivity to details of the pollen emission module parameterisation and to the meteorological input data. The most important parameters are highlighted. The reference year considered for the analysis is 2006. It is shown that the model is capable of predicting about two-thirds of allergenic alerts, with the odds ratio exceeding 12 for the best setup. Several other statistics corroborate with these estimations. Low-pollen concentration days are also predicted correctly in more than two-thirds of cases. The model experiences certain difficulties only with intermediate pollen concentrations. It is demonstrated that the most important input parameter is the near-surface temperature, the bias of which can easily jeopardise the results. The model sensitivity to random fluctuations of temperature is much lower. Other parameters important at various stages of pollen development, release, and dispersion are precipitation and ambient humidity, as well as wind direction.