Implications of Carbon Monoxide Bias for Methane Lifetime and Atmospheric Composition in Chemistry Climate Models

A low bias in carbon monoxide (CO) at northern high and mid-latitudes is a common feature of chemistry climate models (CCMs) that may indicate or contribute to a high bias in simulated OH and corresponding low bias in methane lifetime. We use simulations with CO tagged by source type to investigate the sensitivity of the CO bias to CO emissions, transport, global mean OH, and the hemispheric asymmetry of OH. We also investigate how each of these possible contributors to the CO bias affects the methane lifetime. We find that the use of specified meteorology alters the distribution of Deleted: Implications of Model Bias in Carbon Monoxide for Methane Lifetime


Introduction
Carbon monoxide (CO) is an ozone precursor and a major sink of the hydroxyl radical (OH) in the troposphere [Logan et al., 1981;Spivakovsky et al., 2000].
Consequently, CO indirectly impacts climate by increasing tropospheric ozone where sufficient NO x is present and increasing the lifetimes of methane and other short-lived greenhouse gases (GHGs), as well as eventually oxidizing to CO 2 (e.g. Prather, 1996;IPCC, 1990). The effect of CO on OH also leads to impacts on oxidation of SO 2 to sulfate, providing another climate forcing (Shindell et al., 2009). Previous studies calculated global warming potentials due to these effects using box models (Daniel andSolomon, 1998), 2-dimensional models (Fuglestvedt et al., 1996;Johnson and Derwent, 1996), or 3-dimensional models (Derwent et al., 2001;Fry et al., 2013;Berntsen et al., 2005;Shindell et al., 2009;Fry et al., 2012). Since neither CO nor ozone is well mixed in the atmosphere, the location of the CO perturbation affects its climate impact. CO emissions in the tropics have a greater impact on ozone radiative forcing than emissions at high latitudes (Bowman and Henze, 2012) due the intense photochemistry in the tropics as well as the presence of deep convection, which can loft ozone precursors to the upper troposphere where the ozone radiative forcing is greatest (Fry et al., 2013;Naik et al., 2005).
While a large number of modeling studies have investigated the sources, transport, and distribution of CO, global models often show major biases compared to observations. A study of 26 atmospheric chemistry models found that the simulated CO was biased low in the extratropical northern hemisphere (Shindell et al., 2006) compared to satellite observations from MOPITT (Emmons et al., 2004) and surface observations, especially during spring. Shindell et al. (2006) attribute this bias primarily to an underestimate in northern hemisphere (NH) CO emissions, particularly from Asia. Monks et al. (2015) found that arctic CO is biased low in the multi-model POLARCAT Model Intercomparison Project (POLMIP), and identified differences in global OH concentrations as a major driver of the inter-model differences in arctic CO. The multimodel mean of the Atmospheric Chemistry and Climate Model Intercomparison Project (ACCMIP) simulations also shows a negative bias compared to both MOPITT and surface observations in the northern extratropics (Naik et al., 2013). Naik et al. (2013) found that ACCMIP models underestimate the methane and methyl chloroform lifetimes compared to the observation-based estimates of Prinn et al. (2005) and Prather et al. (2012) and produce a high bias in the NH to southern hemisphere (SH) ratio of OH, consistent with the underestimate of NH CO. A recent comparison of simulated and observed methyl chloroform levels also indicates that the NH/SH OH ratio is 0.97+0.12 rather than the value of 1.28 calculated by the ACCMIP models (Patra et al., 2014).
However, OH and CO are major losses for each other, complicating the determination of how much CO bias drives OH bias versus OH bias driving CO bias.
Previous studies used models to examine the consistency of CO emission estimates with surface and satellite observations of CO concentration. A modeling study by Duncan et al. (2007a) showed that their model compared well to observations in the NH extratropics, but they point out that a low bias in their model emissions may have been compensated by an assumption made in their simplified chemical scheme that nonmethane hydrocarbons (NMHCs) oxidize to CO immediately upon release; this assumption is valid during summer months for short-lived NMHCs (e.g., alkenes, isoprene), but is not valid in winter and spring for longer-lived NMHCS (e.g., alkanes). Stein et al. (2014) found that a combination of higher winter traffic emissions from North America and Europe and reduced dry deposition during boreal winter improved the agreement between simulated and observed CO. These findings are consistent with inversions of MOPITT CO data that show that including greater winter emissions from the NH reduces the negative bias in springtime CO at northern latitudes (Petron et al., 2004). Kopacz et al. (2010) inverted CO observations from multiple satellites and concluded that northern mid-latitude CO sources were underestimated in winter, and that implementing large seasonal variations in emissions improved model agreement with observations. The inversion study of Fortems-Cheiney et al. (2011) also found that the posterior CO emissions had large seasonality in the NH, with the maximum occurring in spring. However, the source strengths estimated by inversions are influenced by factors such as model transport (Arellano and Hess, 2006) and the concentrations of other species that interact with CO through OH chemistry (Pison et al., 2009;Muller and Stavrakou, 2005;Jones et al., 2009). Given the complexities of the nonlinear CO-OH-CH 4 system, we conduct a series of sensitivity studies in which we adjust individual inputs to a chemistry climate model (CCM) one at a time.
Uncertainty in the tropospheric burden and distribution of OH leads to further uncertainty in the CO budget. Hooghiemstra et al. (2011) found that higher NH OH concentrations led to higher anthropogenic CO emissions in their inversion study, while lower OH concentrations over tropical land masses and the SH led to lower biomass burning CO emissions and less CO from NMHCs. Duncan et al. (2007a) found that reducing OH by 20% globally improved the comparison of their simulated CO with surface observations in some locations, but degraded the comparison at other locations. Patra et al. (2014) suggested that top-down emission estimates from models with much higher OH in the NH than SH likely overestimate NH countries' emissions of CO and other reactive species. Mao et al. (2013) found that including conversion of HO 2 to H 2 O on aerosols reduced OH concentrations in a global model, correcting much of the negative model bias in the extratropical NH with the largest CO increase occurring in spring.
Accurate simulation of tropospheric OH requires models to represent multiple drivers of atmospheric OH concentrations. Holmes et al. (2013) found that temperature, water vapor, stratospheric ozone, and emissions from biomass burning and lightning could together explain most of the interannual variability in methane lifetime against OH. Duncan and Logan (2008) found that changes in the ozone column were a major driver of OH variability over 1988-1997. Murray et al. (2013 found that lightning was more important for OH variability over 1998-2006, when there was less variability in the overhead ozone column. Murray et al. (2014) found that on glacial-interglacial timescales, OH concentrations were proportional to the tropospheric ozone photolysis rate, specific humidity, and reactive nitrogen emissions, and inversely proportional to CO emissions. The present study investigates the sensitivity of simulated CO and methane lifetime to biases in some of these processes.
Understanding the causes and implications of CO bias in CCMs is important for climate prediction as it may indicate or contribute to biases in methane and ozone and their respective radiative forcing contributions. The goal of this study is to quantify the relationship of the extratropical NH CO bias seen in CCMs with bias in oxidant concentrations and methane lifetime. Our focus is primarily on NH spring and summer, when the NH CO bias is large. We use the GEOS-5 Chemistry Climate Model (GEOSCCM) to investigate how attributing a CCM's CO bias to CO emissions versus OH chemistry impacts ozone and methane lifetime. We also quantify the contribution of model biases in other constituents such as ozone and water vapor to the simulated OH and CO distributions and methane lifetime.

Constituent Observations and Assimilated Fields
Our primary constraint on the model CO distribution comes from surface observations from the NOAA Global Modeling Division (GMD) network (Novelli and Masarie, 2014). We use the monthly mean data. The MOPITT instrument on the TERRA satellite provides additional constraints on the CO distribution. MOPITT provides almost global coverage every three days from March 2000 to present (Edwards et al., 2004). We use the level 3 CO column data from the MOPITT version 5 thermal infrared (TIR) product (Deeter et al., 2011;Deeter et al., 2013).
Observations of tropospheric ozone are important for constraining the source of OH. Ziemke et al. (2011) created a climatology of tropospheric column ozone (TCO) based on the difference between the stratospheric column ozone (SCO) from the Microwave Limb Sounder (MLS) and total ozone column data from the Ozone Monitoring Instrument (OMI). The observations are cloud-filtered, so there is sensitivity throughout the troposphere, although there is some reduction in retrieval efficiency in the lower troposphere. We use the TCO product for 2004-2010 to constrain the tropospheric ozone column. Stratospheric ozone is also important for constraining the OH source due to its effect on photolysis (Rohrer and Berresheim, 2006). We use the Global Modeling and Assimilation Office (GMAO) ozone assimilation product for 2005-2010 to constrain stratospheric ozone concentrations. The GMAO assimilated ozone product, described in (Ziemke et al., 2014;Wargan et al., 2015), is a gridded product that was created by ingesting MLS ozone profiles and OMI total column ozone into the GEOS-5 assimilation system.
Water vapor is another important influence on OH concentrations. We use specific humidity from the Modern-Era Retrospective Analysis for Research and Applications (MERRA) (Rienecker et al., 2011).

Model and Methodology
Our analysis uses the GEOSCCM to assess possible causes and impacts of CO and OH bias. After spin-up, we conduct a series of time slice simulations of 1999-2009 with fixed emissions using observed sea surface temperatures (SSTs) to drive the CCM meteorology, and then average our results over all years of the time slice. All simulations use the Fortuna version of GEOS-5 (Molod et al., 2012) and have 2 degree latitude by 2.5 degree longitude horizontal resolution and 72 vertical levels.
We use a series of sensitivity studies to analyze the role of CO emissions, OH concentrations, and transport. Two methods are used to examine the sensitivity of CO concentrations to CO emissions from different sources: scaling up the CO emissions, and scaling up CO tracers tagged by source. We quantify the sensitivity of CO to OH concentrations by applying scaling factors to the OH field. We analyze the sensitivity to transport by comparing a free running CCM simulation with a simulation that has prescribed meteorology. Several different chemistry options within the GEOSCCM framework are used to isolate specific processes. We describe each chemistry option and associated experiments below and in Table 1.

GMI Chemistry option
GEOSCCM integrates the chemistry mechanism of the Global Modeling Initiative (GMI) CTM (Duncan et al., 2007b;Strahan et al., 2007) into the GEOS-5 Atmospheric GCM (AGCM). The GMI chemistry includes a comprehensive mechanism of tropospheric and stratospheric chemistry, including 117 species and over 400 reactions.
The reference simulation for this study, hereafter called RefGMI, is the year 2000 timeslice simulation conducted for the ACCMIP intercomparison. The configuration of the ACCMIP simulations is described in Lamarque et al. (2013). The biases in CO and methane lifetime seen in the GEOSCCM simulation are similar to those seen in the ACCMIP multi-model mean (Naik et al., 2013). Consequently, our analyses of bias in the GEOSCCM are likely applicable to other CCMs as well. We use the GMI chemistry option to quantify the impact of changes in CO emissions on methane, OH, and ozone.

CO-only Option
GEOSCCM includes a CO-only option to "tag" CO according to source type or location (Bian et al., 2010). This simplified chemistry option allows us to separate the contributions of different CO sources and to quantify the impact of a specific change in OH. In this chemistry option, the loss of CO is calculated based on prescribed OH fields, so changes in CO do not feed back onto OH. Our reference tagged-CO simulation, called RefCOonly, uses monthly OH fields archived from the RefGMI simulation, and the CO sources from methane and isoprene oxidation are calculated using monthly methane and isoprene fields archived from RefGMI as well. Emissions and other forcings are chosen to parallel the RefGMI simulation; however, the COonly option includes an amplification factor for anthropogenic and biomass burning CO emissions to account for the absence of co-emitted NMHCs (Duncan et al., 2007a). We use the CO-only option to calculate the influence of specific sources on CO concentrations, and to isolate the impact of specific changes in OH on CO.
The GEOSCCM includes an option to constrain the meteorology with MERRA or any GMAO assimilation product. The simulation is pulled towards the MERRA analysis through application of an incremental analysis update (Bloom et al., 1996), calculated every six hours from comparison of the simulation with the analysis. We conduct a COonly simulation with specified dynamics from MERRA, which we refer to as COonlySD, where "SD" stands for "specified dynamics". The COonlySD simulation has the same emissions and OH field as RefCOonly, but the tracer transport will differ between the two simulations.

CO-OH Option
A third chemistry option within the GEOSCCM is the CO-OH option (Duncan et al., 2000;Duncan et al., 2007a). This chemistry option is of intermediate complexity

Results
We use the three chemistry options of the GEOSCCM to separate the contributions of emissions, chemistry, and transport to model bias in CO and methane lifetime. In Section 3.1, we compare the CO distribution simulated by the three options to observations and discuss the consistencies and differences between simulations. We analyze the impacts of increasing emissions, decreasing OH, and changing model transport on the CO distribution in Section 3.2, and examine how changing CO emissions affects ozone and OH in Section 3.3. In Section 3.4, we investigate the contribution of known model biases to simulated OH and methane lifetimes. Table 1 summarizes the key results of each of the sensitivity studies.

Observations
This section presents a comparison of CO and OH distributions from the RefGMI, RefCO-only, and RefCO-OH simulations to each other and to observations. The choice of chemistry option affects the CO and OH distributions produced by the GEOSCCM.
Annually averaged OH is higher in the RefGMI simulation in both hemispheres (Table 2; Fig. 1). High OH values extend further down in the troposphere in RefGMI than in RefCO-OH. Global annual mean mass-weighted tropospheric OH is approximately 7% lower in RefCO-OH than in RefGMI, but the NH/SH OH ratio is marginally higher: 1.22 instead of 1. 19. Prather et al. [2012] report an observation-based estimate of methane lifetime of 9.1+0.9 years, and a methane lifetime against tropospheric OH of 11.2+1.3 years. The lifetimes of methyl chloroform and methane against tropospheric OH in the RefGMI simulation are 5.9 and 9.6 years, respectively ( Table 2). The lifetimes of methyl chloroform and methane against tropospheric OH in the RefCO-OH simulation are 6.4 years and 10 years, respectively, within the uncertainty of the observation-based estimates. The seasonal cycles are similar in both simulations, but the difference in NH OH is larger in the first half of the year than the second half ( Fig. 1d). Consequently, when using the CO-OH option, we present the changes due to a given factor rather than the absolute values of CO, OH, and methane lifetime for easier comparison with the other simulations. The RefCO-OH simulation (orange) shows less springtime bias compared to observations than the RefGMI and RefCOonly simulations. However, its annual cycle is shifted later, resulting in more negative biases in September-December at the higher northern latitudes. The difference between the RefCO-OH and the other two reference simulations is explained by the differences in the OH field calculated by the parameterization compared to that calculated by the GMI chemistry mechanism. In April, when the greater NH bias of the RefGMI simulation is most evident (Fig. 2), the RefGMI OH is markedly higher than the RefCO-OH OH (Fig. 1).

Sensitivity of simulated CO to sources, OH, and transport
We conduct a series of sensitivity studies to examine possible causes of the bias in NH CO seen in the Ref-GMI and Ref-COonly simulations during spring and summer.
We focus on spring since it is the season with the largest bias in NH OH (Fig. 2), and contrast the spring results with those from summer, since the bias persists into summer despite seasonal differences in transport and chemistry.

Sensitivity of CO-only simulations to sources and OH
Adjusting the strength of mid-latitude CO sources can reduce the bias in the interhemispheric CO gradient. We use CO-only simulations to examine the impact of increasing specific sources of CO. Following the method of Strode and Pawson (2013), we estimate the impact of increasing a particular source by increasing the tagged-CO tracer for that source and then re-computing total CO. We impose increases of 10%, 20%, 50%, 100%, and 150% for each tagged tracer. Figure  Increasing COtrbb improves the model agreement with observations at low latitudes, but creates overestimates at some tropical sites while providing only a modest reduction in the high latitude bias (Fig. 3b).
We examine the impact of reducing OH concentrations globally or only in the NH in our COonly simulation and find that a large decrease in NH OH is effective in reducing the high latitude CO bias. Naik et al. (2013) found that the ACCMIP multi-

Impact of sources and OH on global mean and inter-hemispheric gradient bias in CO
We find that achieving zero bias in both the inter-hemispheric gradient (IHG) and the global mean would require changes in multiple emission sources. Figure 4  where as increasing the other sources yields a slight improvement in correlation (Fig. 4b,   5b). Increases in both COtrbb and CObio reduce the correlation with observations in summer, but the effect is also present for CObio in spring. We therefore exclude increases in COnaa and CObio in the rest of our study.

Figures 4 and 5 also
show the impact of changing OH. The sensitivity of CO to changes in OH is location dependent, with higher sensitivity in regions without strong local CO sources (Holloway et al., 2000). Reducing OH globally reduces both the IHG and global mean bias. However, reducing OH by 20% in the NH only yields a greater reduction in both biases as well as the greatest improvement in correlation. We refer to the simulation with the 20% decrease in NH OH as COonlyLowNHOH.

Impact of adjusting sources versus NH OH on comparison to observations
We next compare the impact of increasing emissions with that of reducing NH OH by 20%. We compare the COonlyLowNHOH scenario with a simulation called GMI-HiEmis that includes increased CO emissions, further described in Section 3.3. The RefCOonly simulation shows a similar surface CO distribution to RefGMI (Fig. 6a,c).
The COonlyLowNHOH scenario (Fig. 6d) improves the agreement with the remote high latitude sites compared to the RefCOonly simulation, but like the HiEmis case (Fig. 6b), it leads to an overestimate of CO concentrations over Europe and eastern United States.
Consequently, a combination such as reduced NH OH and reduced emissions over the eastern U.S. is likely needed to reconcile the simulated CO with observations.
Sparse surface data makes it difficult to determine from comparison with GMD observations whether the higher CO seen in GMI-HiEmis and COonlyLowNHOH is realistic ( Fig. 6), so we also compare the four simulations shown in Figure 6 to the CO column from MOPITT. Both the RefGMI and RefCOonly simulations show a large negative bias in NH CO compared to MOPITT (Fig. 7a,c). The GMI-HiEmis and COonlyLowNHOH simulations both reduce this negative bias (Fig. 7b,d), but the increased emissions of the GMI-HiEmis simulation lead to a greater overestimate of CO over East Asia and Indonesia. Furthermore, eliminating the increase in COea to reduce the high bias compared to surface observations over Europe and compensating with a larger adjustment in Asian CO would lead to an even greater bias over Asia compared to MOPITT.

CO sensitivity to transport
Transport, in addition to chemistry and emissions, plays a role in the IHG of CO.
The simulations discussed so far are free-running CCM simulations driven by SSTs, since our goal is to understand the biases seen in CCM studies such as ACCMIP.
Consequently, the simulated tracer transport is affected by any differences between the simulated and actual meteorology. We examine the sensitivity of the CO bias to model transport by comparing RefCOonly, which is a free-running CCM simulation, with the COonlySD simulation, which has year-specific meteorology from MERRA. The largest difference between the simulations occurs poleward of 30°N during July through October (Fig. 8a). The use of specified meteorology makes only a small difference in the global mean and IHG CO biases compared to surface observations in March-May (Fig. 4), but leads to a large reduction in bias as well as improved spatial correlation with the GMD observations in June-August (Fig. 5). The biases in global mean CO and the CO IHG are -3.5% and -6.6%, respectively, in COonlySD, compared to -11% and -26%, respectively, in RefCOonly. More CO from northern hemisphere anthropogenic and boreal biomass burning sources remains in the lower troposphere and reaches the high latitudes in COonlySD, where as more is transported to the upper troposphere in RefCOonly (Fig. 8b).

Impact of Increased CO Emissions on Ozone, OH, and CH 4 Lifetime
The tagged tracer results presented in Section 3.2.2 suggest that increasing CO emissions can improve the agreement with CO surface observations. However, increasing CO emissions will lead to feedbacks on OH, CH 4 , and ozone that are not captured by the COonly chemistry option. We therefore conduct a sensitivity simulation with the GMI chemistry option called GMI-HiEmis that is identical to the RefGMI simulation except for an increase in CO emissions. Since trace gases in the GMI option of the GEOSCCM are radiatively coupled to the underlying GCM, altering emissions within this option produces feedbacks between CO, ozone, CH 4 , and radiation and transport.
The results shown in figures 4 and 5 suggest that increasing CO from tropical biomass burning along with Asian anthropogenic, European anthropogenic, and Russian biomass burning CO, can eliminate both the IHG and global mean bias. We therefore adjust the emissions for winter, spring, and summer in the GMI-HiEmis simulation based on the biases and tagged tracer sensitivities for the season. We do not alter the September through December emissions since the RefGMI simulation shows little NH bias in those months (Fig. 2) and our focus is on spring and summer. We choose the adjustment factors by solving for the linear combination of COaa+COea+COrubb and COtrbb that minimizes the error in both the IHG and the global mean bias. Table 3 shows the emissions adjustments for each season. We apply the same adjustment to COaa, COea, and COrubb since these three sources show nearly the same slope (IHG bias / model bias) in Figures 4 and 5. The purpose of this experiment is not to calculate the optimum CO emissions to reproduce the CO observations, but rather to determine how a reasonable set of CO emission adjustments impacts the simulated concentrations of ozone and OH as well as CO.
Including larger CO emissions increases the loss of OH, reducing OH concentrations.
The mass-weighted global mean tropospheric OH in the GMI-HiEmis simulation is 1.11x10 6 molec/cm 3 , a 3% decrease compared to the RefGMI simulation ( Table 2). The methane lifetime against tropospheric OH also increases slightly, from 9.6 years in the RefGMI simulation to 9.9 years in the HiEmis simulation. This slightly improves the agreement with the observation-based estimate of 11.2+/-1.3 years of Prather et al.
(2012). Since the decrease in OH occurs primarily in the northern hemisphere, the NH/SH OH ratio shows a small reduction from 1.19 in the RefGMI simulation to 1.16 in the GMI-HiEmis simulation, but the large hemispheric asymmetry in OH remains.
Both the RefGMI and GMI-HiEmis simulation show a large high bias in NH TCO compared to the OMI/MLS observations, as well as a low bias in the equatorial Pacific and the extratropical SH ( Figure 9). Similar biases were present in the ACCMIP multimodel mean (Young et al., 2013), indicating that these biases are a common feature in CCMs. The increased CO emissions in the GMI-HiEmis simulation slightly increase the high bias in TCO in the northern midlatitudes, but the increase is small compared to the bias in the RefGMI simulation.

Sensitivity of OH to model biases
Given the sensitivity of the CO biases to OH (Figs. 3,4,and 5) and the relatively small changes in OH resulting from the increased CO emissions in the GMI-HiEmis simulation, we next examine how other biases in the RefGMI simulation may impact OH concentrations and consequently the CO distribution and methane lifetime. We conducted a series of sensitivity studies using the CO-OH parameterization option to isolate the impact of several known model biases on OH and CO distributions. We analyze the results for the entire year in order to compare our results to observation-based estimates of methane lifetime.
The primary source of OH in the troposphere is ozone photolysis followed by reaction of O 1 D with water vapor, and secondary production of OH occurs through reaction of HO 2 with NO or ozone (Spivakovsky et al., 2000). Consequently, simulated OH concentrations are sensitive to errors in NO x and ozone concentrations, water vapor, and factors that influence photolysis such as overhead ozone column and clouds. Here, we examine the sensitivity of OH to model biases in some of these factors.

Sensitivity to Tropospheric Ozone
Comparison of output from the CCMs that participated in the ACCMIP study to OMI/MLS TCO reveals positive biases in simulated tropospheric ozone over the NH midlatitudes, and negative biases over the tropical Pacific and SH (Young et al., 2013), which lead to an overestimate in NH OH production and an underestimate in SH OH production (Naik et al., 2013). The multi-model mean tropospheric ozone also shows a high bias in the NH and low bias in the SH compared to the Tropospheric Emission

Sensitivity to Stratospheric Ozone
We next examine how biases in stratospheric ozone affect OH through their role in photolysis. Voulgarakis et al. (2013) found that changes in stratospheric ozone and tropospheric OH in the ACCMIP models both correlated strongly with J(O 1 D). We conduct a sensitivity study for 2005 through 2009, CO-OHSensStO3, which parallels the RefCO-OH simulation but replaces the simulated ozone with the GMAO ozone assimilation in the stratosphere for input into the parameterization. The change in global mean OH in CO-OHSensStO3 is nearly identical to that of CO-OHSensTCO, but CO-OHSensStO3 places more of the change in the SH, causing the NH/SH OH ratio to increase to 1.23 ( Table 4). The change in OH due to stratospheric ozone bias is small in part because GEOSCCM has relatively small biases in tropical stratospheric ozone.

Sensitivity to Water Vapor
We investigate the impact of model biases in water vapor on OH concentrations through its role in OH production. Inter-model differences in OH in the POLMIP study Simulated OH is 6% lower in the CO-OHSensQ simulation than the RefCO-OH simulation, and the methane and methyl chloroform lifetimes against OH are thus 6% longer (Table 4). Consequently, water vapor bias has a larger impact on OH concentrations than the stratospheric or tropospheric ozone biases in our simulations.
The OH reduction is similar in both hemispheres in the annual mean, so the NH/SH OH gradient is unchanged compare to RefCO-OH.

Sensitivity to NO x
Anthropogenic NO x emissions, which contribute to secondary production of OH, are located primarily in the northern hemisphere. Consequently, an overestimate of these

Summary of OH Sensitivities
In the previous sections, we examined the sensitivity of OH and its IHG to model biases in tropospheric and stratospheric ozone, water vapor, and NH NO x emissions. We find that water vapor has the largest impact on global mean OH, while NH NO x emissions have the largest impact on the NH/SH ratio. However, none of these biases individually explains the 20% reduction in NH OH that would remove the interhemispheric asymmetry and which Figures 4 and 5 suggest are necessary to remove most of the CO bias. We note, however, that the CO-OH option simulations do not account for all the chemical feedbacks between ozone, methane, OH, and other species, and may underestimate the sensitivity of the full chemistry simulation to some of these biases.  Figure 11 shows the latitudinal distribution of the change in CO for each sensitivity simulation versus RefCO-OH. The sum of the changes for CO-OHSensTCO, CO-OHSensStO3, CO-OHSensQ, and CO-OHSensNOx, shown in the dotted line, is similar but slightly smaller than the change for CO-OHSensAll, indicating small non-linearities in the system. Correcting water vapor (CO-OHSensQ) makes a large contribution to the CO enhancement in both hemispheres, while adjusting NO x emissions (CO-OHSensNOx) and to a lesser extent tropospheric ozone (CO-OHSensTCO) contributes to the larger increase in the northern hemisphere.

Conclusions
We examined possible causes of CO model bias, such as underestimated emissions or overestimated OH, in a global CCM. An underestimate of CO emissions can impact the chemistry-climate simulation through the interaction of CO with methane via OH and with ozone, but we find the effects to be small. In contrast, a CO bias due to excess OH would imply biases in methane lifetime, further influencing the simulated climate.
Either increasing emissions or decreasing northern hemisphere OH can remove the bias in the latitudinal gradient of CO compared to surface observations. However, we find that the large increases in Asian anthropogenic emissions needed to remove the negative CO bias at remote surface sites leads to overestimates of CO over Asia compared to MOPITT. In contrast, reducing OH in the northern hemisphere improves the agreement between simulated and observed CO concentrations. This is consistent with the finding of Patra et al. (2014) that the ratio of northern hemisphere to southern hemisphere OH is close to one. We note that biases in OH, CO emissions, and transport are not mutually exclusive, and the model bias in CO is likely influenced by a combination of these factors.
The availability of satellite-based constraints on CO, ozone, and water vapor enables us to assess model biases that affect the major sources and sinks of OH and hence CO concentrations and methane lifetime. We used a CO-OH parameterization to explore the effect of model biases in ozone and water vapor on simulated OH and CO. Removing the high bias in northern hemisphere tropospheric ozone, a common feature of CCMs, had only a small effect on simulated OH and methane lifetime. A high bias in water vapor had a larger impact on global mean OH, but removing this bias did not remove the northern versus southern hemisphere asymmetry in OH. Removing both ozone and water vapor biases, as well as decreasing northern hemisphere NO x , provided the desired increase in methane lifetime, but was insufficient to remove the hemispheric asymmetry in OH. Thus, while a NH/SH OH ratio near one improves the simulation of CO, we cannot generate this ratio in our simulations by removing known model biases in ozone and water vapor.
Our study suggests that the springtime low bias in CO at northern latitudes often seen in CCM simulations likely indicates a bias in methane lifetime. Improving the model representation of water vapor and ozone, as well as reducing uncertainty in NO x emissions, could reduce these biases. However, additional research is needed to understand the causes of northern versus southern asymmetry in simulated OH. Future field missions that provide data on the latitudinal distribution of CO and oxidant sources and losses will be valuable for understanding biases in simulated CO and OH.