Cluster analysis of variations in the diurnal pattern of grass pollen concentrations in Northern Europe (Copenhagen) and Southern Europe (Cordoba)

Abstract

From an allergological point of view, Poaceae pollen is one of the most important types of pollen that the population is exposed to in the ambient environment. There are several studies on intra-diurnal patterns in grass pollen concentrations and agreement on the high variability. However, the method for analysing the different patterns is not yet well established. The aim of the present study is therefore to examine the method of pattern analysis by statistical clustering, and to relate the proposed patterns to time of season and meteorological variables at two highly different biogeographical locations: Cordoba, Spain, and Copenhagen, Denmark. Airborne pollen is collected by Hirst-type volumetric spore traps and counted using an optical microscope at both sites. The counts were converted to 2-h concentrations, and a new method based on cluster analysis was applied with the aim of determining the most frequent diurnal patterns in pollen concentrations and their dependencies on site, season and meteorological variables. Three different well-defined diurnal patterns were identified at both locations. The most frequent pattern in Copenhagen was associated with days having peak pollen concentrations in the evening (maximum between 18 and 20 h), whereas the most frequent pattern at Cordoba was associated with days having peak pollen concentrations in the afternoon (maximum between 14 and 16 h). These three patterns account for 70% of days with no rain and pollen concentrations above 20 grains m−3. The most frequent pattern accounts for 40% and 57% of the days in Cordoba and Copenhagen, respectively. The analysis clearly shows the great variation in pollen concentration pattern, albeit a dominating pattern can be found. It was not possible to explain all the differences in the patterns by the meteorological variables when examined individually. Clustering method is estimated to be an appropriate methodology for studying aerobiological phenomena with high variability.