Assessment of pollutant emission inventories by principal component analysis of ambient air measurements

Abstract

The emission ratios of [CO]/[NOx] and [SO2]/[NOx] in point and area type sources in the northeastern United States are estimated through a combination of principal component and bivariate regression analyses of concurrent measurements of [CO], [SO2], [NOx], and [NOy] made at a rural Pennsylvania field site during the summer of 1988. Principal component analysis is used to identify time periods when the covariation of [NOy] with either [CO] or [SO2] indicates that the air mass was dominated by emissions from either area or point type sources, respectively. A regression analysis is used to extrapolate the observed pollutant ratios to the emission ratios. The resulting emission ratios compare reasonably well with emission ratios derived from the 1985 NAPAP (National Acid Precipitation Assessment Program) emissions inventory, although this analysis indicates that the NAPAP inventory may underestimate the [CO]/[NOx] emission ratio for area sources by a factor of 1.0 to 2.2 (95% confidence limits).