Measuring and modelling the local-scale spatio-temporal variation of urban particle number size distributions and black carbon

- Local-scale urban measurements of particles and black carbon.
- Measurements were categorised into different outdoor microenvironments.
- Particle number size distributions were modelled by a multiple regression approach.
- Two models with different input parameters.
- Best agreement for near-traffic microenvironments.

Mobile measurements were performed to study the spatio-temporal variation of particle number size distributions (NSD) in the range 11 < Dp < 365 nm as well as total particle number and black carbon concentrations in Braunschweig, Germany during the winter and summer period 2012/2013. The study area of about 1 km2 consisted of six different outdoor microenvironments (ME) that were classified according to different traffic intensities and dominant land use types along the measurement route.Highest averaged total number concentrations measured at roadside (RO) were 2.5 × 104 pt cm−3 (with a maximum of 7.6 × 104 pt cm−3) during winter and about 1.2 × 104 pt cm−3 on average during the summer campaign. Measurement spots which are more distant to traffic were characterised by lower concentrations of 1.6 × 104 pt cm−3 and 9.0 × 103 pt cm−3 during winter and summer, respectively. Black carbon (BC) concentrations were also clearly related to traffic emissions and resulted in concentrations of 2.8 μg m−3 on average (absolute maximum of 6.2 μg m−3) at RO-sites. The concentrations of particles and BC in the different ME (aggregated from the single measurement spots) documented the concentration of both metrics to be a function of distance of the measurement to fresh traffic emissions.A multiple regression based model was established to identify significant parameters which can be used to model the microscale variation of particle NSD in the outdoor ME. Two models with different numbers of input parameters were calculated. The first contained all measured parameters as input, the second only a reduced number consisting of TNC, BC and wind speed. Both models worked convincingly, even the approach with the limited number of input parameters. The average size integrated (TNC) deviation to observed data in all ME during both seasons was <13%. The best agreement between model and observations is given for the near-traffic ME.