Quantifying the causes of differences in tropospheric OH within global models

The hydroxyl radical (OH) is the primary daytime oxidant in the troposphere and provides the main loss mechanism for many pollutants and greenhouse gases, including methane (CH4). Global mean tropospheric OH differs by as much as 80% among various global models, for reasons that are not well understood. We use neural networks (NNs), trained using archived output from eight chemical transport models (CTMs) that participated in the Polar Study using Aircraft, Remote Sensing, Surface Measurements and Models, of Climate, Chemistry, Aerosols and Transport Model Intercomparison Project (POLMIP), to quantify the factors responsible for differences in tropospheric OH and resulting CH4 lifetime (τCH4) between these models. Annual average τCH4, for loss by OH only, ranges from 8.0 to 11.6 years for the eight POLMIP CTMs. The factors driving these differences were quantified by inputting 3-D chemical fields from one CTM into the trained NN of another CTM. Across all CTMs, the largest mean differences in τCH4 (ΔτCH4) result from variations in chemical mechanisms (ΔτCH4 = 0.46 years), the photolysis frequency (J) of O3→O(1D) (0.31 years), local O3 (0.30 years), and CO (0.23 years). The ΔτCH4 due to CTM differences in NOx (NO + NO2) is relatively low (0.17 years), although large regional variation in OH between the CTMs is attributed to NOx. Differences in isoprene and J(NO2) have negligible overall effect on globally averaged tropospheric OH, although the extent of OH variations due to each factor depends on the model being examined. This study demonstrates that NNs can serve as a useful tool for quantifying why tropospheric OH varies between global models, provided that essential chemical fields are archived