Errors in particle size distributions from Sequential Mobility Particle Sizers due to varying number concentration at an urban roadside location

• We simulate the effect of sudden particle influxes on SMPS data inversion.
• Underestimation of true particle concentrations by ~2% at a roadside site.
• Individual particle diameters may be underestimated by ~8%.
• A decrease in SMPS cycle time increases the severity of the effect.
• Errors introduced are less than typical uncertainties for SMPS measurement.

Scanning or Sequential Mobility Particle Sizers (SMPS) are commonly used to obtain number size distributions (NSD) for submicron aerosols. In SMPS data inversion, correction for particle multiple-charging typically utilises data obtained at earlier times, therefore there is potential for error if concentration varies over the cycle time. We examine this error by using roadside experimental data to define simulated SMPS response distributions, including a time-varying transient peak due to a particle influx event, for input to inversion procedures. Because this distribution is specified for all times, the correct multiple-charging contributions can be used in data inversion. The results are compared with NSD obtained using a standard algorithm. We show that the net effect is to underestimate the true concentration, especially for diameters 80−200 nm (for the instrumentation studied here). For parameters typical of observed roadside transient events, errors in total concentration are ~2%, with up to ~8% error for concentration at individual diameters, and in all but the most extreme cases are less than typical experimental uncertainty estimates for SMPS instruments at roadside sites. Many previously published roadside size distributions may be affected, and for only a modest improvement in time resolution the effect may be exacerbated. The error introduced for given particle influx characteristics (which can be determined experimentally in future work and may be estimated for previous works) and for different instruments can be easily evaluated.