Comparison Of Four Dispersion Models And Land Use Regression For Estimating Traffic-Related Ultrafine Particles Concentrations In An Urban Area

Abstract

Introduction: Because the number concentrations of particles (PNC) are usually elevated near busy roadways, comparative evaluation of models used to estimate PNC for epidemiological studies is needed. Our goal was to assess the ability of steady-state Gaussian dispersion models that parameterize atmospheric stability by classes (CALINE 4), or Monin-Obukhov length (R-LINE, AERMOD), a Lagrangian model (QUIC) and land use regression to estimate near-road traffic-related PNCs in urban areas. Methods: We studied two neighborhoods near Interstate 93 (I-93) in MA, USA: a residential area (Somerville) and an urban center (Chinatown). Traffic emission factors were from a local study (1.4x10^14 particles/veh-km) and meteorological and traffic data were obtained from state agencies. Measurements for model evaluation were made with a mobile laboratory over the course of one year in each study area. Performance of hourly models under four test scenarios (hot/cold temperatures, parallel/perpendicular wind directions) was evaluated by R^2 between measurements and predictions, fraction of predictions within a factor of two of measurements (FAC2), normalized mean square error (NMSE), and fractional bias (FB). Results: In Somerville, FAC2 was =45%, NMSE ranged from 0.004 to 0.136, and FB was between –0.76 and 0.37 for all models, and R^2 was generally =0.6. For parallel winds in winter, the R^2 approached 0 for CALINE 4, R-LINE, and AERMOD and was relatively low for the regression (0.35) and QUIC (0.51) models. Model performance was worse in Chinatown, mainly due to the difficulty of estimating background PNC. Discussion: There were minimal differences in the measured performance of the five models because the noise of the measurements was greater than differences between model predictions. Improvement in background PNC estimates could be more important than the choice of model, particularly in urban centers with heavy local traffic.