Evaluation of updated nitric acid chemistry on ozone precursors and radiative effects

. This study shows that revising the reaction rate of NO 2 + HO · −−→ HNO 3 improves simulated nitrogen partitioning and [.. 1 ] changes the simulated radiative effects of several [.. 2 ] short-lived climate forcers (SLCF) . Both laboratory and ﬁeld study analysis have found that the reaction rate should be reduced by 13-30% from current recommendations. We evaluate the GEOS-Chem model over North America with and without the recommended update using observations from the INTEX-NA Phase A campaign . Revising the 5 NO 2 + HO · −−→ HNO 3 rate coefﬁcient improves model performance of oxidized nitrogen partitioning by in-creasing NO x concentrations in the upper troposphere and decreasing HNO 3 throughout the troposphere. The [.. 3 ][.. 4 ] increase in NO x [.. 5 ] concentrations has a corresponding global increase in O 3 concentrations and [.. 6 ] local increases in sulfate aerosols, causing a perturbation in the simulated radiative effects of tropospheric ozone. These ﬁndings demonstrate the [.. 7 ] positive inﬂuence the mechanism update has on the [.. 8 10 ] partitioning of oxidized nitrogen species , the beneﬁts it provides when compared to aircraft observations and the simulated radiative effects that the reduction induces. in nitric acid the of [.. [.. in . To assess the radiative effects of changing the nitric acid reaction rate, the Parallel Ofﬂine Radiative Transfer (PORT) model was [.. 134 ] used (Conley et al., 2013). This standalone model was developed at the National Cen-ter for Atmospheric Research (NCAR) and isolates the radiation code from the Community Atmosphere Model 195 (CAM). [.. 135 ] The model calculated the direct instantaneous radiative forcing due to the [.. 136 ] nitric acid kinetic update, strictly as it relates to changes in atmospheric composition simulated by GEOS-Chem.

1 2010). The benefit of [.. 9 ]GCTM's to their regional counterparts is the scale that decreases sensitivity to boundary conditions [.. 10 ] (Jacobson, 2005). When new information on a process emerges in the literature, the GCTM must be evaluated in the context of that [.. 11 ]update. In addition, an understanding into how this update 20 would have influenced conclusions from previous studies must be considered.
GCTMs are often used to predict [.. 12 ]ozone and aerosol concentrations that are products of photochemical oxidation. In the context of oxidation, the chemical component of GCTMs (a.k.a. chemical mechanism) indirectly influences all the other processes. Chemical transformation directly changes the chemical availability of compounds and the physical properties of compound families. For instance, [.. 13 ]Reaction 1 decreases the 25 photochemical availability of a hydroxyl radical (HO · ) and nitrogen oxides (NO x − −NO + NO 2 ). Reaction 1 also increases the solubility of oxidized nitrogen [.. 14 ]since the Henry's Law coefficient for HNO 3 (2.1 × 10 5 M /atm at 298 K) is seven orders of magnitude greater than that of NO 2 (10 −2 M /atm at 298 K). Uncertainty in [.. 15 ]Reaction 1 would, therefore, affect the lifetime of NO x emissions and the lifetime of NO y as a NO x reservoir. This is [.. 16 ]important for other molecules such as ozone since ozone production is limited, on average, by NO x 30 availability (Sillman et al., 1990;McKeen et al., 1991;Chameides et al., 1992;Jacob et al., 1993;Jaegl et al., 1998a).
Reaction 1 is widely recognized as a key reaction in atmospheric oxidation (e.g., Seinfeld, 1989;Donahue, 2011), but has not been well constrained. Despite its known influence [.. 17 ]and importance, Reaction 1 has [.. 18 ]proven difficult to measure at temperatures and pressures in the troposphere (Donahue, 2011). In a recent 35 study, Mollner et al. (2010) employed state-of-the-science techniques to accurately measure the reaction rate at standard temperature and pressure (T = 298K and P = 1atm). In a subsequent study, Henderson et al. (2012) constrain the rate of [.. 19 ]Reaction 1 using aircraft measurements from the upper troposphere (T = 240K and  et al., 2006), biogenic (Guenther et al., 2006), lightning (Ott et al., 2010), and anthropogenic emissions (described below).
Anthropogenic emissions of NO x , CO, and SO 2 are included at both a global and regional scale. At the regional scale, anthropogenic emissions of NO x , CO, and SO 2 are specifically provided for the United States of  (Parrish, 2006;Dallmann and Harley, 2010). The European emissions are provided by the Co-operative Programme for Monitoring and Evaluation of the Long-range Transmission of Air Pollutants in Europe (EMEP) 75 inventory for Europe in 2000 by Vestreng and Klein (2002). The Mexico emissions are derived from the 1999 Big Bend Regional Aerosol and Visibility Observational (BRAVO) emissions inventory for Mexico (Kuhns et al., 2003). [.. 47 ]Asia emissions are derived from Streets et al. (2003Streets et al. ( , 2006. For the rest of the world, emissions are [.. 48 ]derived from the EDGAR fossil fuel inventory and scaled from the year 2000 (Olivier et al., 2002).

80
In this study, we compare simulations with standard chemistry (base case) and revised chemistry (HNO3 case).
The reaction rate of NO 2 + HO · is decreased to account for emerging literature recommending a downward re-45 removed: timestep 46 removed: 2000 47 removed: The 48 removed: included 4 vision (Mollner et al., 2010;Henderson et al., 2012). [.. 49 ] Mollner et al. (2010) recommend a 13% [.. 50 ]decrease to the rate recommended by Sander et al. (2011), which is lower than that recommended by Atkinson et al. (2004). Donahue (2011) commended the recent work by Mollner et al. (2010), but asserted that there is 85 remaining uncertainty. Henderson et al. (2012) also re-evaluated the rate constant using Bayesian inference and measurements from the upper troposphere. The evaluation in the upper troposphere complements the Mollner et al. (2010) study with information at temperatures from 230-250 K. Henderson et al. (2012) conclude that the temperature sensitivity is currently overestimated and should be revised according to Equations 2 and 3. As such, updates to GEOS-Chem in the HNO3 case are as follows: characterize the middle free troposphere (e.g., Bertram et al., 2007;Singh et al., 2007). In order to account for measurement uncertainty, we use a method referred to as the two one-sided t-tests 160 (TOST) (Schuirmann, 1987). Using TOST, we can test whether the model predictions are within measurement uncertainty by rejecting one of two null hypotheses. The first null hypothesis is that the simulated mean is greater than the observations adjusted to their lower bound. The second null hypothesis is that the simulated mean is less than the observations adjusted to their upper bound. If we reject either hypothesis, we have rejected that the model mean is equivalent to the [.. 106 ]observation mean. This approach is equivalent to assuming a systematic bias 165 equal to the uncertainty in the measurement.
Using relative uncertainty, we formulate the null hypotheses (H 0,1 and H 0,2 , shown below) using products.
For each measurement, the observed accuracy is based on an estimate, which can be found in the header of the We account for different variances and observational uncertainty using a variant of 97 removed: . The 98 removed: assumes that the variances of the two populations are identical. The variances are not expected to be identical and, therefore, the standard Student 99 removed: is not appropriate for this evaluation 100 removed: Although the t-test can now compare the measurements and predictions, it cannot yet account for measurement accuracy. 101 removed: Having accounted for the variances, we must now address the reported accuracy and precision tolerances of the observations. 102 removed: For example, we now know that the 103 removed: measurement has an interference from peroxy nitrates. The methyl peroxy nitrate interference ranges from 2.5% at 265 104 removed: to 60% at 225 105 removed: . Therefore, we need to estimate measurement accuracy and account for it in our evaluation technique. 106 removed: observations 107 removed: An alternative formulation is to produce a confidence interval for the difference and compare that to the uncertainty of the mean. We did not use this approach because it does not account for adjustments to observational variance when uncertainty is provided as a factor. 111 removed: The null hypotheses are formulated to give the benefit of doubt to the model. The joint null hypothesis is that the model is within uncertainty, which must be rejected to conclude that the model is different (greater or less than) from observations. A higher bar would be equivalence testing where we reverse the null and alternative hypotheses. As defined, the analysis is conservative with respect to model evaluation.
112 removed: divisions to capture the influence of 113 removed: . Temperature and pressure affect 114 removed: chemical reactions including the reactions that produce 115 removed: . The affect of temperature/pressure sensitivity can, therefore, only be seen by evaluating the model with respect to altitude. 116 removed: When using statistical tests like the t-test, we must be careful to maintain the independent and identically distributed assumption 117 removed: These pairs help to preserve identical distribution because observations and predictions will represent the same geographic regions.
118 removed: this case, the set of model predictions will contain duplicates that must be removed to maintain independence. After removing duplicates, we have 119 removed: that are each a representative 120 removed: For each altitude, we compare the observed and simulated values of chemical concentrations. To reduce the influence of spatial averaging on variance, variables that demonstrate log-normal distributions will be log-transformed. By log transforming, the distribution becomes symmetric and reduces the skews influence on the mean. By converting all variables to normal distributions, we also allow for the use of statistical tests like the t-test. 121 removed: When equivalence of observations and simulations is rejected, we examine the bias further. For bias calculations, the duplicate model results are not removed. By retaining duplications, each observation can be paired with a prediction. This allows us to calculate the mean normalized bias (B N ) as defined in Equation ??. In Equation ??, o i is an observation, y i is a prediction, and n is the number of pairs. The number of pairs varies by compound because some observations are more available than others. 123   130 removed: short-lived climate forcers in the atmosphere. These changes result in variances in the radiative budget of the atmosphere and will change the predicted forcing at the surface and top of the model domain. For the updates to the nitric acid mechanism, these changes 131 removed: the 132 removed: , which is a large contributor to the radiation balance of our atmosphere. To a lesser extent, changes in radiative effects due to the updated nitric acid mechanism include concentration differences of certain aerosols, such as sulfuric acid. Ultimately, as previous mentioned, a decrease in the reaction rate of nitric acid formation will increase tropospheric photochemical ozone production, which is largely limited by 133 removed: availability. This would have a positive increase in radiative effects in the atmosphere and the intensity of such radiative changes will largely be spatially and temporally heterogenous. In addition, the nitric acid mechanism update can change the oxidation potential of the atmosphere. This change can affect the formation of aerosols and has a potential to vary the concentration and distribution of aerosols, such as sulfuric acid. This process has the potential of creating negative radiative effects. 134 removed: utilized 135 removed: By using this model , the direct radative 136 removed: mechanism updatecan be quantified. 137 removed: an 138 removed: Radiative effects due to ozone, sulfate, organic and black hydrophilic and hydrophobic carbon, sea salt and dust were quantified using PORT. While the main drivers of the radiative effects due to the mechanism update will be driven by 139 removed: sulfate aerosols, all of these variables were included due to availability   Using the updated chemistry also exacerbates an existing high bias of ozone (not shown). The base 173 removed: bin 174 removed: improves model performance and is consistent with observations at all levels. For 175 removed: , the HNO3 case improves predictions at all levels above 3 176 removed: , though there are many bins of statistically significant low bias between 4 and 10 177 removed: . However, this is seen in both simulated scenarios and is improved with the HNO3 case. 178 removed: Unlike 179 removed: , 180 removed: , and 181 removed: , using the udpated chemistry exacerbates an existing high-bias. The base case ozone predictions are high-biased throughout most of the troposphere (excluding 1 to 3 182 removed: ). The high-bias for ozone is likely the result of over-predictions 183 removed: and 184 removed: . Figure 2a shows 185 removed: over-prediction from 0 to 8 186 removed: . The high-biased 187 removed: is well correlated with a high-bias seen for 188 removed: that extends throughout the same vertical structure. 189 removed: The high-biased 190 removed: may be the result of lightning emissions that are highly uncertain. GEOS-Chem emits 191 removed: , produced from lightning flashes, according to a vertical profile published by Ott et al. (2010) shown in Fig. ??. The lightning profie shows a distinct similarity between normalized 192 removed: biases, as previously discussed. A high bias exists in the altitudes of 5 to 8 193 removed: , which corresponds to an area of high lightning flashes. The ratio of freshly emitted 194 removed: to 195 removed: shows a distinct similarity with the bi-modal lighting profiles observed by Ott et al. (2010)  ) and 212 removed: is biased low as a fraction of 213 removed: . 214 removed: is 215 removed: concentration is biased high , but the 216 removed: is only biased high 217 removed: and 218 removed: The PORT simulations had a spin-up period of 4-months to allow for radiative equilibrium due to the atmospheric perturbation. Following the spin-up period, the simulation was carried on for a full year to allow a calculation of a 219 removed: Evaluation 220 removed: As previously mentioned, an offline radative transfer model (PORT) was run, utilizing the output generated from the GEOS-Chem GCM. The input to this offline model included ozone , sulfate, organic and black hydrophilic and hydrophobic carbon, sea salt and dust. While many of these variables were not expected to be changed as a result of this mechanism update, each were included due to their availability. Each of these climate forcing variables were analyzed individually to determine the radiative effects associated with each climate variable. The complete difference associated with the mechanism update was also analyzed. As hypothesized, the results showed that ozone was the strongest contributor to surface and top of atmosphere direct radiative effects , with smaller and localize effects also observed for simulated differences in sulfate aerosols. These variables changes are due to the 221 removed: cycling that produces photochemical ozone and the changing atmospheric oxidation potential that the mechanism enables. The spatial and vertical changes, which further substantiate this assessment will be discussed further in the following section. 222 removed: The PORT simulations had a spin-up period of 4-months to allow for radiative equilibrium due to the atmospheric perturbation. Following the spin-up period, the simulation was carried on for a full year to allow a calculation of a 223 removed: change in radiative forcing . In total, this method enabled a global annual average radiative effects determination that included all seasons; and the simulation time step allowed an even analysis of day and night forcings. As previously mentioned, the time step for this analysis was every 2,190 minutes; which allowed a balance of computational strain and even season/daylight sampling routines. The global annual averaged change in radiative flux, including both solar and terrestrial radiation, at the surface from 224 removed: reaction rate was 6.8 225 removed: . The global annual averaged change in radiation flux at the effective top of the atmosphere was 27.9 226 removed: . The effective top of the atmosphere, in reality, is 227 removed: , which 228 removed: As simulated, these values were driven strictly by the ozone and sulfate aerosol climate variables. Due to the increases in tropospheric ozone , the resulting change in radiative effects from ozone were a net positive gain. These increases were 31.  atmosphere radiative forcing minus the surface radiative forcing and has strong influences on regional precipitation (Shindell et al., 2012). 231 removed: at the surface 232 removed: tropospheric 233 removed: aersols 234 removed: . This will be further discussed in the following section. However, in comparison to tropospheric ozone, this 235 removed: radiative effects. These decreases were -3.0 236 removed: at the 237 removed: and -3.3 238 removed: at the surface 239 removed: To put these global annual average values into perspective, the Intergovernmental Panel on Climate Change (IPCC) Assessment Report 5 (AR5) estimated that the total radiative forcing since pre-industrial times for ozone to be 240 removed: . The values from these results cannot be directly applied to these IPCC values since the IPCC values are estimated to occur at the troposphere, as is the definition of radiative forcing. However, it can be assumed that the values from this study would result in a net flux change at the troposphere to be somewhere between the simulated 241 removed: values that were obtained. While the concentrations of tropospheric ozone have many determinants beyond the kinetic rate of nitric acid formation, the comparison of model predictions to published values of historical ozone forcing enables a comparative base line to analyze results against. 242 removed: averaged spatial distribution of radiative effects 243 removed: . As seen, there is largely a net 244 removed: radiative forcing, which was hypothesized, 245 removed: global increases of tropospheric ozone resulting from the mechanism update 246 removed: it is observed that the 247 removed: mid-latitude regions, with a slight decrease along the equator between the two mid-latitude regions. Also, a larger magnitude of forcing occurs 248 removed: atmosphere 249 removed: The net 250 removed: the spatial plot at the bottom of the figure, is defined as the top of the atmosphere minus the surface forcing and has an influence on regional precipitation (Shindell et al. (2012)). As seen in this portion of Figure 3, the atmospheric forcing effects were entirely positive, with a maximum value situated near the equator . It is hypothesized that this result would cause precipitation increases in this portion of the world, which has the potential to further perturb the global radiation balance through indirect effects, which were not included in this simulation.  ]sulfate from the updated mechanism at the surface, top of model, and the net atmospheric forcing. 285 In contrast to Figure 3 and 269 ], due to the revised mechanism, will be shown. Fig. 5 and Fig. 6 show that the increases in ozone occurred globally, with maximum increases  Fig. 6 shows that the changes to sulfate concentrations were limited to areas near the surface and in the upper mid-latitudes. 300 Figure 5 shows that the localized concentration changes to HNO 3 and NO x in the surfacial layer had an inverse relationship with one another, and occurred in the same localized regions as the concentration changes to sulfate. When reviewing Reaction 1, this inverse relationship is expected. However, the decrease in the formation of nitric acid due to this mechanism update would [.. 273 ]lead to an expected increase in NO x , which is not shown in Figure 5. When viewing Figure 6, it is seen that this phenomenon 305 is limited to the surface and quickly [.. 274 ]changes throughout the rest of the troposphere. This is likely due to an increase in heterogeneous nitrogen chemistry on the surface of the locally increased sulfate 254 removed: near global radiative effects 255 removed: radiative changes 256 removed: aerosols were only localized; and the local areas were only above landmasses 257 removed: radiative effects assoicated 258 removed: the sulfate aerosols resulted in net decreases in 259 removed: The longitudinally averaged portion of the plot shows near zero values at all latitudes due to the strongly localized nature of these changes. It is hypothesized that these localized, traditionally polluted areas, are limited in their capacity to oxidize 260 removed: and the increase in hydroxyl radicals resulting from this mechanism update allowed the increase in production of sulfate aerosols. 262 removed: Assessment 263 removed: due to the revised chemistry kinetics. The analysisincludes simulated changes to 264 removed: , 265 removed: and 266 removed: , on both 267 removed: . As seen in Figure ??, which displays the difference in 268 removed: and 269 removed: concentrations at the surface between the HNO3 and Base Case simulations, 270 removed: localized variations in both species had an obvious inverse relationship, and 271 removed: localized over landmasses 272 removed: the localized directional changes are counter intuitive to the assumed directional change that the 273 removed: create. These directional changes are strictly 274 removed: change throughout the troposphere, as shown in Figure ??. aerosols (Bell et al., 2005;Liao et al., 2004). Figure 6 shows Figure ??, the changes in the spatial distribution of ozone at the surface resulted in near global increases. A majority of the ozone changes occurred in the upper mid-latitudes and spanned the entire vertical atmosphere, as shown in Figure ??. Vertically, most of the ozone changes occurred in the free troposphere 283 removed: above the planetary boundary layer. The previous hypothesis that the changes in sulfate radiative effects were a result of changing oxidation potential were further review by looking at the spatial distribution of sulfate aerosol changes, as well. When reviewing Figure ??, it is seen that the horizontal changes in sulfate at the surface occurred in the same localized regions as the surficial changes to   (Jaegl et al., 1998b;. In this study, the 320 284 removed: . When viewed on a vertical basis, Figure ?? shows the the vertical changes to sulfate aerosol concentrations strictly occurred near the surface and did not follow the same trends as 285 removed: and 286 removed: , which had large differences throughout the troposphere. 287 removed: Difference in mean 288 removed: and Sulfate mixing ratios between the HNO3 and Base Case simulations for the surface layer.  293 removed: model performance and its sensitivity to the resulting chemistry. In 294 removed: we have implemented updates to the GEOS-Chem chemistry and evaluated those updates during the INTEX-A observational campaign. Following an adjustment to this chemical mechanism, an evaluation of 295 removed: , its components and the resulting effects on atmospheric direct radiative effects were analyzed. We find that the base model has a high bias for 296 removed: , so 297 removed: components ( 298 removed: , 299 removed: , and 300 removed: ) were evaluated as fractional components to determine how the mechanism effects speciation. Overall, the updated chemistry improves total oxidized nitrogen partitioning and decreases the termination of 301 removed: through the formation of nitric acid. In addition, since the oxidation of 302 removed: was decreases, a near global increase in ozone concentrations were seen. This increase resulted in changes to the oxidation potential of localized regions, which changes the concentration of resulting aerosol formation. All of these results have a relationship with the simulated radiation budget of the atmosphere. 303 removed: The updated 304 removed: chemistry improves simulated partitioning of 305 removed: , 306 removed: , and 307 removed: throughout most of the atmosphere. In 308 removed: where this analysis was mainly targeting improved simulation results, the updated chemical mechanism improves modeled results for all 309 removed: components above 8 km. In the middle troposphere, 310 removed: and 311 removed: also experience improvements in predictions; however, the updated chemistry exacerbates a base model bias for 312 removed: that may be caused by the lightning emission profile.  generally the limiting species in tropospheric ozone production. Simulations using the updated chemical mechanism saw global increases of ozone throughout the troposphere, which increases the model bias. This further suggests that constraints on NO x emissions are needed to improve modeled ozone concentrations.

360
Updates to the NO 2 + HO · reaction rate, as suggested by Mollner et al. (2010) and Henderson et al. (2012), were implemented in GEOS-Chem and the resulting model performance was evaluated using observations from the INTEX-NA, Phase-A campaign. This evaluation considered total NO y concentrations, NO y partitioning, and the resulting direct instantaneous radiative forcing effects from this mechanism update. An initial comparison found that the base model had a high bias for NO y . As such, NO y 365 components (NO x , HNO 3 , and PAN) were evaluated as fractional components to determine how the mechanism effects speciation. Overall, the updated chemistry improves oxidized nitrogen partitioning and decreased the termination of NO x in the atmosphere through the formation of nitric acid. [.. 349 ]To put these global annual average values into perspective, the Intergovernmental Panel on Climate Change (IPCC) Assessment Report 5 (AR5) estimated that the total radiative forcing since preindustrial times due to ozone is 350 mW/m 2 . While the concentrations of tropospheric ozone have many 395 determinants beyond the kinetic rate of nitric acid formation, the comparison of model predictions to published values of historical ozone forcing enables a comparative base line to analyze these results against. As well, additional radiative effects can be expected due to this mechanism update. In the tropics, where a net positive increase in atmospheric forcing is simulated, additional atmospheric responses and feedbacks are likely to occur. These feedbacks include changes in atmospheric moisture and cloud 400 cover. Since the radiative transfer model used in this evaluation was offline, these calculations were not included and should be considered in future work.
Overall, this study demonstrates that updates to the [.. 350 ]nitric acid chemical mechanism generally improves oxidized nitrogen partitioning performance in GEOS-Chem throughout the troposphere. It should be noted, however, that this model evaluation is based on a model that is already high-biased for NO y con-405 centrations [.. 351 ]throughout a majority of the troposphere. As such, improvements to the global emission 335 removed: The larger differences in the upper troposphere are most likely due to long-range transport. As a result to these changes in tropospheric ozone, simulated climate forcing due to this climate variable were evaluated. 336 removed: changing atmospheric chemistry, mainly relating to 337 removed: aerosols, experienced changes due to this mechanism update. By utilizing 338 removed: , the radiative effects resulting from this kinetic update were quantified. Raditiave effects were seen in both the solar and terrestrial forms of the radiation spectrum, and were mainly caused by differences in ozone 339 removed: a positive net flux of 6.8 340 removed: a positive net flux of 27.9 341 removed: quantified 342 removed: the atmosphere 343 removed: Ozone contributed radiative effects in both the solar and terrestrial forms of radiative energy while sulfate only contributed effects in the solar form through scattering processes.

Appendix A Total Oxidized Nitrogen Concentrations
The main text shows total oxidized nitrogen partitioning (see Figure 2), but not concentrations of component species [.. 359 ]NO x , HNO 3 , or PAN. Figure A8 provides concentration data to complement Figure  Altitude (

Appendix B Referee Comments and Responses
Each referee comment will be listed and bold. The author comment will follow each referee comment and will 540 be italicized.
Comment: 3220, 21: Instead of speaking of "trade off", simply state "In comparison to regional models, GCTMs have decreased sensitivity to boundary conditions and increased sensitivity to emissions, transport, and chemistry. A reference to some paper showing this comparison would be useful. Response: Agreed and a reference discussing this comparison has been added. For each simulation, we evaluate the model in 1 km vertical bins. This method of evaluation was chosen since temperature, pressure and transport have large variability throughout the vertical troposphere, and these variables play a strong role in the rate of Reaction R1.
Comment: Section 2.5: Surface radiative forcing is confounded discussion. The authors need to clarify if the forcing is "instantaneous radiative forcing" or "radiative forcing". "Radiative forcing" was defined 570 as the change in flux (at the top of atmosphere or tropopause) including a stratospheric temperature adjustment under the assumption of fixed dynamical heating. If the authors have computed "instantaneous" forcing, then the surface forcing makes sense, otherwise they need to address the extent of atmospheric and surface process adjustments. Response: The values used in this analysis were based on instantaneous radiative forcing.

575
Comment: 3230, 15: While previous papers by Henderson, et. al., have focused on the 8-10 km region, readers of this paper will be caught off guard by this sentence. Perhaps a note in the introduction, or something clarifying the reason for this focus at this point in the paper would be useful. Response: An added sentence to provide additional clarification was added to the introduction.
Comment: In addition, the authors need to clarify whether the radiative forcing is computed as an in-580 stantaneous effect, or with the stratospherically adjusted temperature due to fixed dynamical heating. If Strat. Adjust. was not used, then the 4 month equilibrium is a red herring. Response: The radiative effects were calculated as an instantaneous radiative forcing and all values relating to the radiative effects analysis were reported as annual average instantaneous radiative forcing in mW/m 2 . Updates to the manuscript have been made to make this clearer. As well, the 4-month equilibrium was eliminated and the calculations were re-run. The 585 annual average instantaneous radiative forcing values changed by ¡ 0.1 mW/m 2 . The values have been updated.
Comment: One more clarification would be to state that the "change in flux" is a net increase in net downward solar and terrestrial flux due to the change in mechanism. (both "net"s are necessary as well as the "downward") You could, instead, simply refer to net trapped energy. Response: The radiative flux, which is how results in this analysis are presented, is defined as the net increase in net downward solar and 590 terrestrial (combined) radiation. This sentence has been added in the radiative effects methods section to provide clarification.
Comment: 3235, 3: "Due to the increase... " "The increased ozone leads to a net increase in trapped energy beneath the top of the atmosphere of ... and beneath the atmosphere of... Please also clarify that for the sulfate aerosol, the increase the albedo of the earth system, reflecting additional solar radiation to respectively. Similar to ozone, there was a net increase in sulfate aerosols, which occurred mainly in the lower troposphere and over landmasses. These increases resulted in a net decrease in instantaneous radiative forcing, driven by the reflectance of incoming solar radiation. The decreases were -3.3 mW/m 2 and -3.0 mW/m 2 at the 600 surface and top of the model, respectively.
Comment: 3235, 10-20: I am uncertain what the authors are trying to say. This paragraph needs to be rewritten. Perhaps they are trying to say that while the methods and altitude at which radiative forcing are computed are different from those used in the IPCC, the relative magnitude of the correction indicated that the change to the kinetics could be important to understanding processes relevant to policy? If so, this 605 paragraph may belong in the conclusion rather than results. Response: This paragraph was re-written to provide clarification and moved to the discussion portion of the paper.
Perhaps the authors mean "The net absorption of energy by the atmosphere as seen in the third panel of figure 4 will affect convective and transport processes." While the reference to Shindells analysis is nice, 610 does the total of the ozone effect and the aerosol effect lead to a clear effect on precipitation that is explicitly confirmed by the results in this paper, or should this also be in the discussion? Response: Those sentences were poorly constructed. The updated sentence in the results section defines net atmospheric forcing, as defined by Shindell. The portions related to the significance of atmospheric forcing have been relocated to the discussion. discussion. That particular sentence was simplified and rewritten.
Comment: 3238, 18: The authors need to clarify what they mean by "performance". Response: This is acknowledged and further clarification will be included in the re-write.

655
Comment: 3238, 28: "change in ozone sensitivity". Sensitivity to what? Response: Sensitivity was a poor choice of words. Rather Similar to the changes in oxidized nitrogen concentrations, the change in simulated ozone concentrations is modest.
Comment: Paragraph starting at 3238, 28: This paragraph needs help. I dont know what is meant by "modest", or how a "model uses NOx". Do the authors mean "sensitivity of predicted O3" or "change in O3 660 concentrations"? I am having a hard time understand the specific meaning of these sentences. Response: This paragraph will be revised in the re-write.
Comment: Paragraph starting at 3239, 9: The first two lines of this paragraph could be rewritten to say, "The radiative effects of the change in ozone and sulfate distributions was evaluated with an offline radiative transfer code". Please refer to previous discussion of how to be precise about forcing numbers.

665
(Yes, I know this is a bother. Thanks for being precise.) Do you mean variance or change? Response: The forcing numbers were the results of annual average instantaneous radiative forcing. The sentence was changed to the following: The radiative effects due to the changes in ozone and sulfate concentrations were evaluated using an offline radiative transfer model. Also, this is not a bother. Being precise is important.
Comment: 3239, 25: To which policy implications do the authors refer? I do not understand the second 670 sentence of this paragraph. Do the authors mean "robust" or "very similar"? Why do the updates need lab confirmation? What additional evidence, in particular, would be helpful? Response: References to policy implications relate to surficial pollutant concentrations and emissions. While the mechanism largely improved oxidized nitrogen partitioning, the changes in trace gas concentrations that we analyzed were not significant enough to alter either of these policy drivers. Overall, this final paragraph was mostly rewritten.

675
Comment: For all figures: Are these annual averages, or only average during the INTEX period. Response: They actually vary and updates to specify which is used have been added to each necessary figure description.
Comment: 3232, 29: In addition, is there a reference for the fact that the baseline model has a high bias?
Response: The high bias in the model that is being referenced is based on the results from the baseline model and 680 the INTEX-A observations.
Comment: The chemical reaction examined here critically influences NOy and HOx chemistry and compounds oxidized by HO. Thus, it should be noted that changing this reaction rate may affect other aspects of model performance not examined here, and the potential shortcomings of adjusting one reaction rate in isolation. Response: This was considered in Henderson et al. (2012;doi: 10.5194/acp-12-653-2012), where 685 the magnitude of the updated mechanism used in this study was developed. A detailed explanation of these other considerations was detailed in that publication. However, this is a very important point and should certainly be reiterated in this paper. I will make sure a discussion regarding this topic is included at some point in the introduction.
Comment: It would be helpful to compare the new reaction rate with the rate assumed in the base case 690