Forecasting model of , , and pollen concentration levels using spatiotemporal correlation properties of pollen count.

The aim of the study was to create and evaluate models for predicting high levels of daily pollen concentration of , , and using a spatiotemporal correlation of pollen count. For each taxon, a high pollen count level was established according to the first allergy symptoms during exposure. The dataset was divided into a training set and a test set, using a stratified random split. For each taxon and city, the model was built using a random forest method. models performed poorly. However, the study revealed the possibility of predicting with substantial accuracy the occurrence of days with high pollen concentrations of and using past pollen count data from monitoring sites. These results can be used for building (1) simpler models, which require data only from aerobiological monitoring sites, and (2) combined meteorological and aerobiological models for predicting high levels of pollen concentration.