Abstract

There is a worldwide gap of knowledge on long-term road traffic contributions to urban air pollution, brought by a lack of long-term receptor modeling studies. To bridge this gap, we have developed a fuzzy clustering analysis of hourly PM10, PM2.5 and NOx concentrations plus surface meteorology, for finding their common spatiotemporal patterns (or clusters). Road traffic patterns are identified as the ones having the largest contributions to diurnal NOx concentrations, a distinctive seasonality and better correlation with available receptor model estimates. By using PM2.5 - PM10 scatterplots for the contributions of these road traffic patterns and the concept of edge lines (or limiting slopes), we apportioned road traffic exhaust and non-exhaust contributions to PM10 and PM2.5. These results must be calibrated with available receptor modeling estimates. We have applied this methodology in Santiago, Chile for the period 2013-2022 and found that estimated contributions to PM10 and PM2.5 are consistent with pre-2013 estimates obtained with receptor modeling studies. Estimated 2013-2022 trends were significant (p < 0.05), except for non-exhaust contributions to PM10. NOx from road traffic has decreased between 2 and 3 ppb/year across the city. PM2.5 and PM10 from exhaust emissions have been reduced between 0.21 and 0.46 (mu g/m(3))/year with similar urban spatial variability. PM2.5 from non-exhaust emissions has mixed trends across the city, and below 0.15 (mu g/m(3))/year in magnitude. These results are consistent with tighter standards for new motor vehicles sold and with the monotonous increase in motor vehicles in the city.