How chemical kinetics uncertainties affect concentrations computed in an atmospheric photochemical model

Tropospheric photochemical models are used increasingly as predictive tools to assess the chemical response of the lower atmosphere to changes in physical and chemical conditions which influence trace species distributions. Among the many uncertainties in the modeling process are imprecisions in reaction rate data used in formulating model continuity equations. In this paper we evaluate the propagation of these kinetics uncertainties to computed species distributions in a photochemical model. A one-dimensional kinetics-diffusion model having 72 reactions among 24 species is used. Non-chemical sources and initial background concentrations are chosen to be representative of clean continental mid-latitude air. Chemical reaction rate data are mostly those of the NASA Kinetics Evaluation Panel No. 8 (1987) and include imprecisions in photolysis rates and binary and ternary reactions. A Monte Carlo technique is used to estimate uncertainties in computed concentrations due to the given rate uncertainties. We compute uncertainties in odd hydrogen species (the radicals OH and HO₂) and in hydrogen peroxide ranging from 22-41%. Uncertainties for O₃ and CO are, respectively, 17% and 30%. Odd nitrogen uncertainties range from 18% for NO to 72% for N₂O₅. The smallest uncertainty is that for nitric acid at 6%, but this is neglecting uncertainties in physical sources and sinks, such as precipitation scavenging. The uncertainty in OH (31%) is important when using the model to predict tropospheric oxidant levels because OH determines the lifetime of numerous naturally and anthropogenically emitted trace gases.