Modelling of Urban Air Pollutant Concentrations with Artificial Neural Networks Using Novel Input Variables

Goulier, L, Paas, B, Ehrnsperger, L, Klemm, O

Abstract

Since operating urban air quality stations is not only time consuming but also costly, andbecause air pollutants can cause serious health problems, this paper presents the hourly predictionof ten air pollutant concentrations (CO2, NH3, NO, NO2, NOx, O3, PM1, PM2.5, PM10 and PN10) ina street canyon in Münster using an artificial neural network (ANN) approach. Special attentionwas paid to comparing three predictor options representing the trac volume: we included acousticsound measurements (sound), the total number of vehicles (trac), and the hour of the day andthe day of the week (time) as input variables and then compared their prediction powers. The modelswere trained, validated and tested to evaluate their performance. Results showed that the predictionsof the gaseous air pollutants NO, NO2, NOx, and O3 reveal very good agreement with observations,whereas predictions for particle concentrations and NH3 were less successful, indicating that thesemodels can be improved. All three input variable options (sound, trac and time) proved to besuitable and showed distinct strengths for modelling various air pollutant concentrations.