Sources of birch pollen from April to June 2023 registered at 3 monitoring sites in Sweden. Does a currently used automated forecast model agree with realized pollen concentrations?

Abstract

Allergies caused by birch pollen seem to irritate around 22% of the population in Europe and Sweden, making a priority for creating forecasting models of pollen concentrations to assist people preparing for every year’s allergy season. Lot of studies have been conducted regarding the allergens of the pollen and how they affect human health , while other studies are focusing on different environmental parameters that could affect the release of the pollen in the atmosphere but also, how these parameters can interact with the pollen concentrations while they are transferring from one area to another. This master thesis aims to be a part of a bigger study regarding the pollen existence in the atmosphere. The study is focusing on the pollen season of 2023 , starting on around 3nd week of April and ending around 2nd of June and concerns the areas of Göteborg, Malmö and Umeå. The choice was made due to their geographical areas since Umeå is in the North , Malmö is in the South and Göteborg is South -central. It is a comparison between the measurements of real concentrations of pollen and the model concentrations of the SILAM program with the aim of detecting differences and trying to explain with the use of statistics why the difference between the concentrations occurs. Due to the fact that a number of meteorological parameters are affecting the pollen concentrations in the atmosphere, relative humidity was chosen to be studied, with the aim of detecting correlations of this parameter with the pollen concentrations. The results concluded to a difference between the model and real pollen concentrations, with the model predicted significantly higher than the real for all the areas, with a moderate to strong correlation between them as well. The correlations between real pollen concentrations and real relative humidity concluded to a very weak and negative relationship, while comparisons between model values as well as ,between real relative humidity values with model pollen concentrations, presented weak but both negative and positive relationships. So the study indicated that most likely the relative humidity was not the cause for the differences between the forecasted and the registered pollen concentrations.