Articles | Volume 17, issue 5
Research article
17 Mar 2017
Research article |  | 17 Mar 2017

Global sensitivity analysis of the GEOS-Chem chemical transport model: ozone and hydrogen oxides during ARCTAS (2008)

Kenneth E. Christian, William H. Brune, and Jingqiu Mao

Abstract. Developing predictive capability for future atmospheric oxidation capacity requires a detailed analysis of model uncertainties and sensitivity of the modeled oxidation capacity to model input variables. Using oxidant mixing ratios modeled by the GEOS-Chem chemical transport model and measured on the NASA DC-8 aircraft, uncertainty and global sensitivity analyses were performed on the GEOS-Chem chemical transport model for the modeled oxidants hydroxyl (OH), hydroperoxyl (HO2), and ozone (O3). The sensitivity of modeled OH, HO2, and ozone to model inputs perturbed simultaneously within their respective uncertainties were found for the flight tracks of NASA's Arctic Research of the Composition of the Troposphere from Aircraft and Satellites (ARCTAS) A and B campaigns (2008) in the North American Arctic. For the spring deployment (ARCTAS-A), ozone was most sensitive to the photolysis rate of NO2, the NO2 + OH reaction rate, and various emissions, including methyl bromoform (CHBr3). OH and HO2 were overwhelmingly sensitive to aerosol particle uptake of HO2 with this one factor contributing upwards of 75 % of the uncertainty in HO2. For the summer deployment (ARCTAS-B), ozone was most sensitive to emission factors, such as soil NOx and isoprene. OH and HO2 were most sensitive to biomass emissions and aerosol particle uptake of HO2. With modeled HO2 showing a factor of 2 underestimation compared to measurements in the lowest 2 km of the troposphere, lower uptake rates (γHO2 < 0. 055), regardless of whether or not the product of the uptake is H2O or H2O2, produced better agreement between modeled and measured HO2.

Short summary
To better understand the sources of uncertainty in modeled Arctic tropospheric oxidants, we created and analyzed an ensemble of chemical transport model runs with inputs perturbed according to their respective uncertainties. Ozone and OH were most sensitive to various emissions and chemical factors. HO2 was overwhelmingly sensitive to aerosol particle uptake. When compared to airborne measurements, better agreement was found when the model used lower aerosol particle uptake rates.
Final-revised paper