Estimating extreme daily pollen loads for Szeged, Hungary using previous-day meteorological variables

Abstract

The aim of this paper is to analyse how meteorological elements relate to extreme Ambrosia pollen load on the one hand and to extreme total pollen load excluding Ambrosia pollen on the other for Szeged, Southern Hungary. The data set comes from a 9-year period (1999–2007) and includes previous-day means of five meteorological variables and actual-day values of the two pollen variables. Factor analysis with special transformation was performed on the meteorological and pollen load data in order to find out the strength and direction of the association of the meteorological and pollen variables. Then, using selected low and high quantiles corresponding to probability distributions of Ambrosia pollen and the remaining pollen loads, the quantile and beyond-quantile averages of pollen loads were compared and evaluated. Finally, a nearest neighbour (NN) technique was applied to discriminate between extreme and non-extreme pollen events using meteorological elements as explaining variables. The observed below or above quantile events are compared with events obtained from NN decisions. The number of events exceeding the quantile of 90% and not exceeding that of 10% is strongly underestimated. However, the procedure works well for quantiles of 20 and 80%, and even better for those of 30 and 70%. Using a nearest neighbour technique, explaining variables in decreasing order of their influence on Ambrosia pollen load are temperature, global solar flux, relative humidity, air pressure and wind speed, while on the load of the remaining pollen are temperature, relative humidity, global solar flux, air pressure and wind speed.