Uncertainty of modelled urban peak O3 concentrations and its sensitivity to input data perturbations based on the Monte Carlo analysis

Abstract

A simple urban air quality model [MODelo de Dispersion Atmosf erica Ubana e Generic Reaction Set (DAUMOD-GRS)] was recently developed. One-hour peak O3 concentrations in the Metropolitan Area of Buenos Aires (MABA) during the summer estimated with the DAUMOD-GRS model have shown values lower than 20 ppb (the regional background concentration) in the urban area and levels greater than 40 ppb in its surroundings. Due to the lack of measurements outside the MABA, these relatively high ozone modelled concentrations constitute the only estimate for the area. In this work, a methodology based on the Monte Carlo analysis is implemented to evaluate the uncertainty in these modelled concentrations associated to possible errors of the model input data. Results show that the larger 1-h peak O3 levels in the MABA during the summer present larger uncertainties (up to 47 ppb). On the other hand, multiple linear regression analysis is applied at selected receptors in order to identify the variables explaining most of the obtained variance. Although their relative contributions vary spatially, the uncertainty of the regional background O3 concentration dominates at all the analysed receptors (34.4 e97.6%), indicating that their estimations could be improved to enhance the ability of the model to simulate peak O3 concentrations in the MABA.