Sensitivity Analysis of the MaTChM Atmospheric Chemistry Model

In a collaborative project involving Drs. Yang Zhang and Richard Easter of Pacific Northwest National Laboratory's(PNNL) Atmospheric Sciences group, and Drs. Christian Bischof and Po-Ting Wu of ANL/MCS, a comprehensive sensitivity study was performed of the Mass Transfer with Chemistry Model (MaTChM). The purpose of this project was to evaluate the sensitivity of the overall model predictions with respect to a variety of model parameters.

ADIFOR was applied to calculate the first-order sensitivity coefficients of 58-145 chemical species concentrations with respect to 147-407 model parameters, including 125 gas-phase reaction rate constants, 120 aqueous-phase kinetic reaction rate constants, 29 Henry's law constants, 29 effective Henry's law constants, 29 mass accommodation coefficients, 29 uptake coefficients, 21 dissociation equilibrium constants, 21 species initial concentrations, temperature, relative humidity, cloud water content, and droplet size. The sensitivity studies were performed under typical atmospheric conditions ranging from rural to heavily polluted atmospheres. Under each condition, we further evaluated four scenarios: clear air, air with aerosols, cloudy conditions, and clouds and aerosols.

The most influential reactions and the most important parameters under each atmospheric conditions have been identified through the sensitivity analysis. Photochemical ozone formation is much more sensitive to the changes of reaction rate constants under polluted atmospheres (with sensitivity 5-20 orders of magnitude higher than that in relatively clean atmospheres). The identified most influential reactions on O3 formation are (i) the photolytic reactions of O3 and its precursor species; (ii) the conversion of nitric oxide (NO) to nitrogen dioxide (NO2); (iii) the reactive radical reactions; (iv) the oxidation of O3 and its precursors by radical species; (v) the formation and dissociation of oxidant and acid; and (vi) the aqueous oxidation of O3, hydroperoxy radical (HO2) and sulfur species.

One of the most interesting findings from this study is that the sensitivity coefficients of photochemical indicators species, such as the ratio of O3 to the difference of odd nitrogen (NOy) and nitrogen oxides (NOx) (O3/(NOy-NOx)), can be used to evaluate O3-NOx-hydrocarbon sensitivity. The figure below shows the sensitivity of O3/(NOy-NOx) with respect to gas-phase reaction rate constants under rural (bar in green) and heavily polluted (bar in red) clear air conditions. The values shown are the sensitivities to the 22 most influential reactions at 120 minutes from a 2-hour simulation starting from local noon time. Reaction 1 through 28 are O3-NO-NO2 reactions, and reaction 38 through 74 are hydrocarbon reactions.

We see that the sensitivities of O3/(NOy-NOx) to hydrocarbon reactions under heavily polluted atmosphere are a factor of 2-25 bigger than that under less polluted atmospheres, indicating a hydrocarbon-sensitive chemistry in heavily polluted regions. Lower sensitivity to hydrocarbon reactions and higher sensitivity to O3-NO-NO2 reactions indicate a NOx-sensitive chemistry. Other good markers for O3-NOx-hydrocarbon sensitivity analysis are the ratio of hydrogen peroxide to nitric acid (H2O2/HNO3), the ratio of formaldehyde to odd nitrogen (HCHO/NOy), and the ratio of hydrogen peroxide to formaldehyde (H2O2/HCHO). These results will aid, for example, regulatory agencies such as the U.S. EPA in the development of more effective O3 abatement strategies in the O3 nonattainment regions.

Our results also show that the presence of clouds and aerosols significantly reduces the total oxidizing capacity, with the sensitivity of O3 lowered by 2 or larger orders of magnitude. More important, these heterogeneous processes alter the O3-precursor relations through changing the reaction influential amplitudes and the signs of the sensitivities. Cloud chemistry is the dominant heterogeneous process under the remote and marine atmospheres. Aerosols are important scavenger for gaseous species and contribute to the heterogeneous perturbation to photochemistry in polluted atmospheres. The overall model predictions are not sensitive to the changes in individual species mass accommodation coefficient, but they are sensitive to changes in uptake coefficients of HNO3, nitrous acid (HNO2), HO2, H2O2, HCHO, NOx, peroxyacyl nitrate (PAN), and dinitrogen pentoxide (N2O5) and O3. These results indicate that the current O3 abatement strategies developed based on clear air chemistry may not be effective in some regions with high cloud frequency and/or large aerosol loading. The heterogeneous chemistry associated with clouds and aerosols and their effects on photochemical O3 formation must be taken into account in these regions.

A detailed analysis of the results will be published in a forthcoming article. For more details on the results, contact Zhang at y_zhang@pnl.gov. This work was also presented at the 1996 DOE Atmospheric Chemistry Program Annual Meeting, Arlington, Virginia, Nov. 19-21, 1996.