Validation of modelled pollen data on a smartphone app by measured pollen data from pollen sensors

Abstract

Pollen allergy affects a growing number of people. The more and more common use of smartphone apps is a societal feature and pollen information does not derogate from it. Modelled risk data are highly visible to allergic sufferers, they need to reflect reality. The aim of this study is to confront modelled and measured allergic risk data. The app used (MeteoPollen®) is the one developed by Weather Force®. It uses data from CAMS (Atmosphere Monitoring Service by the Copernicus program). The levels provided by Copernicus are levels related to modelled quantities of pollen and not levels of allergic risk. The risk data provided by the RNSA are obtained through the RNSA's pollen network. This network uses pollen sensors, Hirst-type, located in the main cities in France. The analyses are conducted under microscope by trained aerobiologist. RNSA uses also clinical data from doctors’ network, phenological data and meteorological forecast. These data allow to establish for each taxon and each station the allergy risk due to pollen exposure. The taxa retained for this study are birch, grasses, olive and ragweed. The study period is from March to September 2019. We choose a list of cities according to taxa, and as statistical tools correlation graphs between expected and observed data. The different graphs of correlation for the different taxa, for different cities and areas, indicate whether there is agreement between the data obtained with the app and the RNSA measured data. Depending on the taxa, the differences can be significant. Copernicus' use of plant species presence data can cause discrepancies between modelled and measured data. It is important not to confuse modelled risk data related only to modelled pollen quantities, with risk data taking into account measured pollen data and risk levels derived from phenological, clinical and meteorological information.