Forecasting airborne pollen concentrations: Development of local models

摘要

People's sensitivity to allergies may represent one of the most important health factors of the next century to which attention must be paid in order to reduce the incidence of social costs and improve the quality of life. Taking into consideration the earnest requests of the medical-scientific community Emilia-Romagna ARPA (Regional Agency for the Prevention of the Environment) moved the attention from the monitoring to a short and medium term prediction of the concentration of allergenic pollens in the air in order to achieve a more effective therapeutic action. Our main objectives are to improve seasonal forecasts and to interpret anomalous years. A neural network model for grass pollen forecasting has been implemented. Input variables were meteorological situations, i.e., daily temperature (max., min. and average) and rainfall, in addition to combinations of individual variables and their thresholds. The output was daily pollen concentration. The model was able to understand and predict anomalous years. We demonstrate that the relationships between pollen concentrations and meteorological situations are independent from site. This means that such models can understand the differences in different areas.