A Novel Model for Predicting Ambient Ultrafine Particle Number Concentrations in Urban Neighborhoods
Traffic-related ultrafine particles (UFP; <100 nanometers diameter) are ubiquitous in
urban air. Epidemiological evidence of health effects, which is needed to inform risk
assessment at the population scale, is limited due to challenges of accurately
estimating UFP exposures. Model improvements are needed to better predict UFP
concentrations for use in epidemiological studies. This ... read morestudy used a novel methodology
to combine mobile and stationary monitoring to inform models of UFP in Chelsea and
Boston (MA, USA). The objectives were to build hourly particle number concentration
(PNC; a proxy for UFP) models, compare modeled to measured concentrations at residential
sites in both study areas, and assess the ability of the models to predict PNC outside
the data-collection period. Our results show differences between monitoring strategies:
mean one-minute PNC on roads were higher (64,000 and 32,000 particles/cm3 in Boston and
Chelsea, respectively) compared to central-site measurements (23,000 and 19,000
particles/cm3) and both were higher than at residences (14,000 and 15,000
particles/cm3). In both study areas, PNC was highest during winter and lowest during
summer. The combined mobile-and-stationary modeling approach was an improvement over a
mobile-monitoring-only model when compared to ambient PNC at residences: Pearson
correlations of modeled and measured natural log-transformed PNC [ln(PNC)] were 10-50%
higher with the combined models; and combined models increased model precision by as
much as 32%. We also showed that adding a proxy variable to the models for secondary
particle formation explained an additional 2-3% of PNC variability. Models overpredicted
hourly PNC at residences, but adding an intercept into the models corrected for this,
resulting in models with an under/overprediction interquartile range of -3,700-2,900
particles/cm3 (-37-27%) in Boston and -4,100-3,600 particles/cm3 (-29-49%) in Chelsea.
Additionally, these models demonstrated they can be applied to time periods outside of
the original monitoring window (as much as 9 years) and maintain high correlations with
ambient measurements. Sensitivity analyses were conducted to explore model limitations.
These results suggest that PNC models informed by both mobile and stationary monitoring
reduce exposure error and can be applied to longitudinal epidemiological
Thesis (Ph.D.)--Tufts University, 2017.
Submitted to the Dept. of Civil Engineering.
Advisor: John Durant.
Committee: Doug Brugge, Jonathan Levy, and Elena Naumova.
Keyword: Environmental engineering.read less