Land Use Regression Models of On-Road Particulate Air Pollution (Particle Number, Black Carbon, PM2.5, Particle Size) Using Mobile Monitoring

• DOI: 10.1021/acs.est.5b01209
• Publication Date: 2015-07-03T14:40:43Z
• Authors: Steve Hankey; Julian D. Marshall
• Keywords: Smoothing

Land Use Regression (LUR) models typically use fixed-site monitoring; here, we employ mobile monitoring as a cost-effective alternative for LUR development. We bicycle-based, measurements (∼85 h) during rush-hour in Minneapolis, MN to build particulate concentrations (particle number [PN], black carbon [BC], fine matter [PM2.5], particle size). developed and examined 1224 separate by varying pollutant, time-of-day, method of spatial temporal smoothing the time-series data. Our base-case had modest goodness-of-fit (adjusted R(2): ∼0.5 ∼0.4 0.35 ∼0.25 [particle size]), low bias (<4%) absolute (2-18%), included predictor variables that captured proximity density emission sources. The our resulted large model-building data set (n = 1101 concentration estimates); ∼25% buffer were selected at scales <100m, suggesting on-road change on small scales. model-R(2) improved sampling runs completed, with diminishing benefits after ∼40 h collection. Spatial autocorrelation model residuals indicated performed poorly where spatiotemporal resolution sources (i.e., traffic congestion) was poor. findings suggest modeling from is possible, but more work could usefully inform best practices.