Formal uncertainty analysis of a Lagrangian photochemical air pollution model

This study applied Monte Carlo analysis with Latin hypercube sampling to evaluate the effects of uncertainty in air parcel trajectory paths, emissions, rate constants, deposition affinities, mixing heights, and atmospheric stability on predictions from a vertically resolved photochemical trajectory model. Uncertainties in concentrations of ozone and other secondary compounds and in predicted changes due to 25% reductions in motor vehicle nonmethane organic compound (NMOC) and nitrogen oxide (NO,) emissions were examined. Surface wind measurements were interpolated over the modeling domain, and uncertainties were quantified using data withholding. The resulting wind fields and uncertainties were used to generate ensembles of trajectories ending at four Southern California air quality monitoring sites. A motor vehicle emissions inventory and associated uncertainties were derived from remote sensing and fuel sales data. Uncertainties in chemical rate parameters were obtained from expert reviews. Estimated uncertainties in O-3 range across the four sites from 24% to 57% (1 standard deviation (la) relative to the mean). Seven variables contribute almost 80% of this uncertainty. Reductions in motor vehicle NMOC reduce O-3 from 10 +/- 10% (1 sigma) to 28 +/- 10%. With reductions in motor vehicle (NO,) emissions, the change in O-3 ranges from an increase of 14 +/- 14% to a decrease of 6.6 +/- 6.2%.