Exposure Assessment and Misclassification for Traffic-Related Air Pollutants Using Census Information: Spatial Resolution Needs

摘要

Health surveillance and demographic data are typically collected at the census block or tract level. Is this spatial resolution sufficient for traffic-related pollutants, the most important air pollution source in urban areas? We evaluates the spatial resolution and zonal systems needed to estimate traffic-related air pollutants using the detailed information assembled for the 800 km~2 area encompassing Detroit, Michigan, USA. Concentrations of nitrogen oxides (NOx) due to vehicles are predicted using hourly traffic flows on 9,700 links representing all but the smallest roads, the MOVES2010 emission model, the RLINE dispersion model, local meteorology data, a temporal resolution of 1 hr, and a spatial resolution as low as 10 m. Concentration at 30,000 model receptors were joined with shape files to estimate residential exposures for 700,000 individuals at property parcel, census block, tract, and ZIP code levels. We determined the resolution needed to meet specific error criteria, and quantify exposure misclassification for each zonal systems as shown using maps, distributions, contingency tables, and other statistics. To portray traffic-related air pollutant exposure, interpolations between receptors and points of interest should not exceed 40 m for areas within 60-80 m of major roads, and 100 m at larger distances. In census tracts and ZIP codes, average exposures tend to be overestimated since few individuals live very near major roads, the range of concentrations among zones is compressed, most exposures are misclassified, and the high levels near roads are entirely omitted. The use of smaller zones improves performance, but even block-level data will misclassify a substantial number of individuals. These results are robust in being based on a very comprehensive model. Exposure, health surveillance and demographic data should be geocoded or collected at the highest spatial resolution to estimate exposures and impacts of traffic-related pollutants.