Analysis of Traffic and Meteorology on Airborne Particulate Matter in Munster, Northwest Germany

Gietl JK, Klemm O

Abstract

The importance of street traffic and meteorological conditions on the concentrations of particulate matter (PM) with an aerodynamic diameter smaller than 10 mu m (PM10) was studied in the city of Munster in northwest Germany. The database consisted of meteorological data, data of PM10 mass concentrations and fine particle number (6225 nm diameter) concentrations, and traffic intensity data as counted with tally hand counters at a four- to six-lane road. On working days, a significant correlation could be found between the diurnal mean PM10 mass concentration and vehicle number. The lower number of heavy-duty vehicles compared with passenger cars contributed more to the particle number concentration on working days than on weekend days. On weekends, when the vehicle number was very low, the correlation between PM10 mass concentration and vehicle number changed completely. Other sources of PM and the meteorology dominated the PM concentration. Independent of the weekday, by decreasing the traffic by approximately 99% during late-night hours, the PM10 concentration was reduced by 12% of the daily mean value. A correlation between PM10 and the particle number concentration was found for each weekday. In this study, meteorological parameters, including the atmospheric stability of the boundary layer, were also accounted for. The authors deployed artificial neural networks to achieve more information on the influence of various meteorological parameters, traffic, and the day of the week. A multilayer perceptron network showed the best results for predicting the PM10 concentration, with the correlation coefficient being 0.72. The influence of relative humidity, temperature, and wind was strong, whereas the influence of atmospheric stability and the traffic parameters was weak. Although traffic contributes a constant amount of particles in a daily and weekly cycle, it is the meteorology that drives most of the variability.