Reply to RC1 by M. Cazorla, M. Trujillo, R. Seguel, and L. Gallardo

Please, find point-by-point responses (bold) below each reviewer’s comments (Italic). In a separate document, we will submit supporting figures.

The authors ran box model simulations in two South American cities, Santiago and Quito, to investigate the differences in ozone production rates and their sensitivity to variations in VOC and NOx levels. They discovered that, although the ozone production rates (PO3) were similar in both cities, ozone concentrations tend to rise to unhealthy levels in Santiago due to weaker mixing of air. The authors also demonstrated that simultaneous control of both NOx and VOC is the most effective strategy for reducing ozone production in both locations. One of the most intriguing insights from the study was the strong diurnal variations in chemical conditions that differed from one city to another. However, there are some facile assumptions made in the manuscript that render it unconvincing in several respects, particularly regarding the disregard for clouds and the use of pseudo-observations for Quito that are based on data from a different environment. Therefore, I do not recommend the publication of this manuscript in its current form unless the authors can demonstrate that their significant errors in estimating VOC levels in Quito do not adversely affect their main conclusions. Additionally, the impact of clouds must be properly addressed.

We used in situ measurements of CO, NO, NO2, O3 and meteorology taken in Quito and we applied a linear model anchored to CO measurements in Quito to estimate VOCs. The rationale for using CO to estimate VOCs is based on the knowledge that CO is an anthropogenic tracer of combustion, that ambient measurements of traffic-associated VOCs correlate linearly with CO, and that VOC/CO ratios (slope of linear regression) have been found to be consistent across many different cities (Baker et. al., 2007; Borbon et. al., 2013; Brito et. al., 2015) (Please, find links to publications at the end of this document). Based on these useful notions, we used VOC vs. CO linear regressions found experimentally in Santiago (for 11 VOCs R2>0.83) to scale CO measurements in Quito (98% of CO in Quito originates from traffic (Vega et al., 2015; Parra, 2017; Hernandez and Mendez, 2020)). Estimations were used because VOC measurements are out of reach in countries with financial disadvantages such as Ecuador, yet the need to understand atmospheric chemistry processes is urgent for public health and climate reasons. However, we agree that there is uncertainty in our estimation model and that errors need to be better quantified. In the revised version of the paper, we will address this issue by implementing an array of possible model inputs for VOCs. To this end, we will perturb VOCs using a Monte Carlo approach, as suggested in this review. Thus, we will discuss the sensitivity of ozone production to these changes, and we will quantify the uncertainty.

 A couple of the major comments in this review mention that we used data from Mexico City. However, we did not use such data. Therefore, those comments are not connected to the Methods in our paper.

 Regarding clouds, we did not study photochemistry under cloudy conditions. We filtered out overcast days from the entire time series in Quito and Santiago (March 2021 and year 2022). This important methodological step is included in the Methods, sections 2.3.1, 2.3.2, and Figure S1. Some of the comments in this review state that the tropics are cloudy for which photochemistry is reduced in these regions. In the revised version of the paper, we will expand the discussion about solar radiation over the Quito region as we understand that conditions in the equatorial Andes (a very different place from other tropical regions such as the cloud forests) are not necessarily widely known by all readers from different parts of the world. As a brief example, the typical UV index in Quito is 11-16 all year, which is an indicative of the intensity of solar radiation at this location (Parra et. al. 2019).

Major comments

Pseudo observations aren’t real observations – The entire VOC dataset used for the F0AM simulations in the city of Quito is fictional.

The VOC dataset for Quito comes from a model that uses measurements of CO in Quito as a predictor. With 98% of ambient CO in Quito originating from traffic emissions (Vega et al, 2015; Parra, 2017; Hernandez and Mendez, 2020), it is reasonable to think that VOCs from the same source correlate with CO as it is the case in many other cities (Baker et. al., 2007; Borbon et. al. 2013; Brito et. al., 2015) including Santiago (Table 1 in the paper). However, we acknowledge that applying VOC vs. CO regressions from Santiago to scale CO in Quito does not necessarily represent the entire reality of VOCs in Quito. Rather, this strategy provides one possible input for the model. In the revised version of the paper, we will incorporate a set of VOC inputs and we will discuss the sensitivity of the results.

The authors assumed a strong correlation between measured CO concentrations and speciated VOCs based on data collected from a different city, Mexico City (Jaimes-Palomera et al., 2016). This framework relies on two key assumptions: i) that the chemical environment of Mexico City is interchangeable with that of Quito, and

We did not use CO and VOC measurements collected in Mexico City. Therefore, this comment does not have a connection with the Methods stated in our paper.

ii) that both primary and secondary pathways of CO production can be adequately represented by applying a linear regression model to the VOCs-CO concentrations. However, both assumptions can be flawed. For example, there may be significant primary CO sources in Quito that do not provide useful information about the speciated VOCs.

As indicated earlier, 98% of ambient CO in Quito is a primary emission that comes from on-road traffic (Vega et al., 2015, Parra, 2017; Hernandez and Mendez, 2020). Therefore, we have a strong reason to assume that traffic-associated VOCs should be correlated with CO. In the revised version of the paper, we will include a discussion about CO sources in Quito.

The problem is further complicated because there are several unknowns—such as the chemical evolution of VOCs, different primary and secondary pathways, and various deposition rates—against a backdrop of limited data. This oversimplified framework has led to poor representations of many VOCs; for instance, formaldehyde (HCHO) has an R² value of approximately 0.3, which can affect PO3 and HOx concentrations. While the authors can make any ad-hoc assumptions they wish, they must demonstrate the robustness of their conclusions given the substantial errors in their estimated VOCs. I strongly recommend that the authors perturb the VOC concentrations, considering the large uncertainties in their estimates, and conduct all simulations in a Monte Carlo fashion to determine whether their conclusions change.

In our study, 11 VOCs have R2>0.83 and 3 compounds have R2 below 0.3. In the paper we acknowledge that the low R2 found for formaldehyde indicates the secondary nature of this compound. In the revised version of the paper, we will reassess including this compound. However, model runs done with compounds that are directly associated with traffic (R2>0.83) are valid as possible inputs to the model. We agree that perturbing VOCs is a good way to evaluate the results and to better quantify the error. We appreciate the idea provided by the reviewer. In the revised version of the paper, we will implement VOC perturbations as a method to strengthen this section.

The errors should be calculated based on the inaccuracies derived from the linear fit and the comparison of the environmental conditions between Mexico City and Quito.

We did not use information from Mexico City. Therefore, this comment is disconnected from our work.

If the results differ significantly—which I anticipate due to the considerable errors in the VOCs—then it would be wise for the authors to withdraw the paper and wait for real measurements in the future. I am baffled by how the authors can justify contrasting two locations: one based on actual observations and the other on hypothetical data.

Atmospheric chemistry studies in cities such as Quito are urgent under the current global regime of rapidly deteriorating environmental conditions, which impact public health and climate. We believe that finding estimates of VOCs for modeling work is a valid method to study these pressing issues as opposed to waiting, especially at times when environmental action based on science is much needed. Research in atmospheric science often uses approximations and models for many quantities across the globe and averaged over entire regions due to well-known gaps in observations. In the revised version of the paper, we are implementing significant improvements to discuss to the best of our knowledge ozone production chemistry in our cities.

On the definition of PO3 and LROX/LNOx –  The authors assumed it was safe to omit several contributing factors to net ozone production rates because they are considered minor. They noted that water vapor, the interaction of ozone with VOCs, and the interaction of ozone with HOx, as well as the production of RONO2, are several orders of magnitude smaller than the terms included in their calculations. Consequently, they decided not to account for them. While I agree that some of these factors are small, the cumulative effect of these terms, especially water vapor, could become significant (see Table 2 in https://acp.copernicus.org/articles/21/18227/2021/acp-21-18227-2021.pdf). Are these neglected terms truly minor in specific case studies, or did the authors presume they would be small?

We did calculate these losses and found that these contributions were one to two orders of magnitude lower than the loss to nitric acid. In the revised version of the manuscript, we will include these results, as requested.

Additionally, the L/Q thresholds used in the work by Kleinman et al. have been questioned by Schroeder et al. (2017) (https://agupubs.onlinelibrary.wiley.com/doi/10.1002/2017JD026781), who showed that Kleinman et al. incorrectly assumed the net effects of NOz, such as PAN and alkyl nitrate, on PO3 to be insignificant. Therefore, I am uncertain whether the thresholds applied in this study are up-to-date. The authors may want to create isopleths to identify the appropriate thresholds by locating the ridgelines based on their mechanism.

Thank you for pointing out this reference. In this article, authors cite that Kleinman in his 2005 paper acknowledged that when emissions are fresh, PAN is not in steady state, which would lower the LROx/LNOx ratio to values below 1. Under this premise, the authors included losses to alkyl nitrates as an additional fate for HOx + NOx losses and so they obtained a new threshold of LROx/LNOx = 0.35. Authors recommend that future work evaluates the shift in ozone production regimes by applying this new threshold. This amendment brings the ratio LNOx/(LNOx+LROX) to 0.25 instead of 0.5 in our paper. In the revised version of the paper, we will incorporate this discussion.

 One of the most important messages of our paper for our region is that isolated controls on NOx should be avoided and simultaneous controls on VOCs and NOx should be considered both in Santiago and Quito. We are confident in this conclusion, which is also supported by what we observed during the pandemic (Cazorla et. al. 2021; Seguel et. al., 2022). However, proposing specific thresholds on specific levels of NOx and VOCs is an endeavor that we think needs to be approached from a different perspective and preferably using a chemical transport model in a future paper.

Clouds might be the missing piece of the puzzle – The authors repeatedly mentioned throughout the paper that Quito is located closer to the tropics, resulting in more radiation and photochemistry; however, it is known this city experiences more overcast compared to Santiago. All the experiments done in this paper presumed that the sky was clear. The presence of clouds can substantially affect photolysis rates and the ROX-HOX cycle. Why is there no discussion about this key component? The authors could have used satellite-based cloud optical thickness and fraction (such as TROPOMI) to calculate below-cloud photolysis rates using RADM formulation (https://agupubs.onlinelibrary.wiley.com/doi/10.1029/JD092iD12p14681). The reason behind the lower ozone concentration in Quito over Santiago could be less PO3 caused by the clouds. This can potentially debunk the main conclusion from the paper (similar PO3, thus different meteorology). Furthermore, vertical mixing does not necessarily reduce surface ozone concentration as the ozone profile tends to increase by altitude so you can potentially bring down higher ozone aloft close to the surface.

We did not study photochemistry under cloudy conditions. We extracted out of the time series overcast days in Quito and Santiago. This is not a presumption but an actual methodological step described in the Methods, sections 2.3.1 and 2.3.2. Therefore, we do not see how clouds could debunk a conclusion.

It is important to mention that Quito is located right on the equator and at high altitude in the Andes. This is a very different environment from other tropical regions, such as the tropical cloud forests, where it is overcast all the time. As an example, the typical UV index in Quito ranges between 11 and 16 all year (Parra et. al. 2019), which is considered extremely high and shows the intensity of solar radiation at this location. In a separate document, we will show figures of solar radiation taken at our ground station to better illustrate this aspect.

As stated earlier, in our work, we found that under sunny conditions in both cities, ozone production rates are similar in March 2021, but we also double checked these results using sunny conditions in year 2022. Please refer to Figure 13. Only during one season (summer) ozone production rates are high in Santiago. Meanwhile in Quito, the peak P(O3) stays high all year. However, ozone in Quito usually meets the WHO guideline, which coincides with the national air quality standard in Ecuador (51 ppbv, 8-h average), while Santiago has struggled over two decades with ozone pollution and seasonal exceedances of their national standard (61 ppbv, 8-h average). This is a fundamental difference. From previous studies in both cities, which we cite in the paper, we explored mixing conditions as one potential reason for differences. Santiago is permanently affected by a high-pressure system that sits over the city inducing stable conditions (Gallardo et al., 2002; Garreaud et al., 2002; Huneeus et al., 2006; Muñoz and Undurraga, 2010). This explains ozone accumulation in the ambient air. In regard to the shape of ozone profiles over Quito, they lack the usual “S” shape, with higher ozone aloft, commonly observed at other tropical stations. Rather, the Quito soundings show deep vertical mixing. A complete discussion of these observations can be found in previous papers, cited in the text (Cazorla, 2017; Cazorla et al., 2021a; Cazorla and Herrera, 2022b). Thus, we propose that mixing conditions could partially explain ozone accumulation (Santiago) or dilution (Quito) in the ambient air under similar ozone production rates.

NOx-VOC experiments and the problem of ozone – Why did the authors not perturb VOC emissions only? The perturbations in NOx and VOC at the same have two combined effects. One potential outcome of reducing VOC emissions only could be that anthropogenic aromatic VOCs could be responsible for elevated PO3 and can be regulated over NOx reduction. Although it is essential to recognize that ozone has a long lifetime, even though reducing NOx will cause PO3 enhancement at the city's core, it will reduce PO3 in the outflow of the city (suburbs), causing the regional background ozone to reduce significantly. So, in general, I don’t think a 0-D experiment is sufficient to discuss surface ozone regulation. A box model is ill-suited for understanding ozone concentration.

In the revised version of the paper, we will discuss changes in VOCs and the effect in PO3.

Our paper focuses exclusively on the photochemical dimension of ozone formation and our aim is to calculate ozone production rates, not to model the ozone concentration in the ambient air (this requires a chemical transport model). The ozone production rate is the most important term in the ozone budget equation because it is the only source of ozone during daytime. The chemical dimension of ozone production is thoroughly studied across the literature with the use of chemical box models or 0-D models, which is an equivalent term. For example, the paper recommended by the reviewer in a previous comment, Schroeder et al. (2017), uses a box model. In similar manner, important papers in the field of ozone chemistry use box models (Ren et. al., 2003; Shirey et. al., 2006; Ren et. al., 2013; Sebol et. al., 2024). Therefore, the suitability of chemical box models to study the chemistry of ozone production is well demonstrated in the literature. On the other hand, the use of 3-D transport models in cities such as Quito and Santiago, characterized by uniquely complex and steep topography, complex emissions, etc., add many other degrees of freedom and uncertainties that are steps away from the current use of a 0-D model.

Minor comments

L35. This is a bit too cynical given the improvements we have observed in several regions worldwide, such as the US (e.g., Simon et al., 2015: https://pubs.acs.org/doi/full/10.1021/es504514z).

Our intention is to discuss that ozone control remains a challenge in many cities around the world. We will provide more context in the revised version of the paper.

The National Air Quality Standard for ozone in the US is 70 ppb (8-h average), 19 ppb higher than the guideline established by the World Health Organization (51 ppbv or 100 µg m^-3 at STP). In other countries, the national standards are also above the WHO guidelines. For example, in Chile the national standard is 61 ppbv. This difficulty in meeting WHO guidelines demonstrates that ozone levels in many places are still difficult to curb despite continuous efforts.

L36. This would be only appropriate if the lifetime of ozone was short. Having a species with several weeks of lifetime would mean that ozone pollution cannot be easily controlled at the city scale unless there are highly volatile VOCs.

This comment is difficult to interpret. Unless we are misunderstanding the meaning, the reviewer seems to be disputing that controls on ozone precursors must be customized according to the local composition of precursors in the ambient air. This is against what the wealth of ozone studies shows.

L37. Why is it a new challenge?

This is because climate change adds a new layer of complexity to ozone controls (https://doi.org/10.1017/9781009157896.008). We will add more context to this portion of the text.

L42. You need to add some context here, as it is not clear why March is chosen. Is it a month with high ozone levels?

L43 (down to the end of the paragraph). It seems to be relevant to the methodology part, not the introduction.

Requests for edits to lines 42 and 43 are contradictory.

L50. Why is it counterintuitive? The topical region gets more clouds, resulting in reduced photochemistry. A quick look at satellite data proves this point: (https://giovanni.gsfc.nasa.gov/giovanni/#service=QuCl&months=01,02,03,11,12&starttime=2000-01-01T00:00:00Z&endtime=2024-12-31T23:59:59Z&bbox=-107.4375,-58.4414,-32.9062,18.1992&data=MOD08_M3_6_1_Cloud_Fraction_Mean_Mean):

 As indicated earlier, solar radiation in equatorial Quito, a high-altitude city in the Andes, is intense to the point that the typical UV index is 11-16 year-round. During sunny conditions, the intensity of solar radiation combined with traffic emissions in Quito meet the known physics of urban photochemistry. Thus, the idea that this is a reduced photochemistry environment is not plausible. In a separate document we include solar radiation figures, so readers contrast ground-station observations with other sources of information.

L60-65. Could HOx uptake through reduced aerosols also contribute to it? Please also consider that the global background tropospheric ozone was reduced during COVID, so it is hard to disentangle local contributions from external sources. Have those cited studies looked at Ox (NO2+O3)? If Ox stayed the same before and after COVID, it would imply that the NO+O3 was reduced. Also, please do not confuse these chemical conditions with O3 titration through NO. This is why it’s better to talk about Ox sensitivity over O3 sensitivity or limit your conditions to radical-abundant regions (like afternoon sunny times). Because the NO+O3 is just a temporary effect (still important for health exposure) that goes away once photochemistry becomes active.

It is unclear what exactly the reviewer is proposing in this comment. This portion of the text describes how NOx and PM2.5 plummeted in Santiago, Quito, and other South American cities due the COVID-19 mobility restrictions, but the effect on ozone was the opposite. We provide references in South America that discuss this topic extensively and can be consulted for further information.

L69. “eye-opening” is way too strong for known non-linear chemistry. Additionally, this part of the introduction does not tell us anything about the effect of PM2.5 on ozone through modifying photolysis rates and heterogeneous chemistry.

The eye-opening part is not about general notions of ozone chemistry. This part refers to a dichotomy in the environmental outcome if changes in precursors take place. For example, if we reduce or replace diesel vehicles, we will achieve much needed PM2.5 reductions, but ozone levels could rise because of a decrease in NOx.

In the second part of the comment, the reviewer seems to be referring to the effect of optically thick boundary layers loaded with aerosols, such as the case of cities in Asia, which impact photochemistry and ozone formation. However, conditions in South American cities differ from those in Asia. For example, in New Delhi, the annual mean of PM2.5 is usually higher than 100 µm^-3 (104.7 µg m^-3 in 2024, https://www.msn.com/en-in/news/other/delhi-s-pm2-5-levels-in-2024-were-over-2x-the-national-limit-despite-stubble-burning-dip-cse-report/ar-AA1x37OW?ocid=BingNewsSerp). In contrast, from air quality network data, in Santiago the annual mean for PM2.5 in 2023 was 23 µg m^-3 and Quito 15 µg m^-3. In our work we did not consider these effects, only gas phase chemistry.

L75. It is unclear how the authors would answer the second part of the first question with a box model. It seems that they do not have the right tool.

 We will restate this question for clarity.

L95. You also have convection and cloud chemistry affecting ozone.

This comment is disconnected from the sentence in line 95. Again, we only analyzed data under sunny conditions.

L135. These are also known as LNOX and LROX.

Ok.

L190. It is unclear why you are dealing with these cities in two different ways (monthly-grouping vs. season-averaged). You mentioned it later but it should come first in this section.

The reason is stated at the beginning of this line. This was done to expand the set of observations to other conditions seen within a year and not to be limited only to the month of March. We will make sure that this aspect is clear throughout.

Section 2.4 How reliable are NOx measurements given NOz interference in chemiluminescence sensors?

As stated in the paper, we used NOx data measured with instrumentation customarily used by air quality networks, which is subject to EPA standards and to public scrutiny. In this paper we do not report precision NO2 techniques. Chemiluminescence sensors are EPA approved methods for NOx detection, for which they are widely used by monitoring networks. NOx measurements in Santiago were obtained from the air quality network and data are subject to quality control that meets the national legislation in Chile. NOx measurements in Quito were collected at the EMA USFQ station, and calibrations were done with certified standard mixtures as described in the Methods. Therefore, we don’t have reasons to doubt the quality of measurements. In regard to the principle of operation, chemiluminescence sensors detect NO in a reaction cell where ozone is injected and the light signal emitted from the NO+O3 reaction is measured.  NO2 is also detected as NO. To this end, these instruments use a molybdenum catalytic converter heated to 315 °C to convert NO2 into NO, but reactive nitrogen (NOy) could also undergo conversion. According to manufacturers, a substantial amount of NOy is lost along the internal sampling design before gas conversion to NO.

L240. You need to be very specific about the days with gaps. How many days are they?

Please, check 11 lines before line 240. Line 229 contains the specific dates when VOC were measured. We will reiterate this aspect in this portion of the text.

L260. I don’t understand the logic here. Why is the VOC estimation in Quito seen an improvement?

Because Santiago has a more complete and recent set of measurements and because we have direct access to the methods and to the data from our co-authors. In the revised version of the paper, we will further improve this section as discussed earlier.

L274. I don’t think it is necessary to talk about software and licenses for ACP papers.

OK.

Section 2.5.1. Incomplete descriptions of F0AM setup – The methodology doesn’t discuss several key factors. How was the dilution factor optimized? What assumption was made for surface albedo? Is it in the UV range? Do you account for wavelength-dependent albedo? How many days do you cycle to approach the steady state?

The F0AM set up and model options are explicitly detailed in Section 2.5 and Table S2. Albedo and total column ozone are discussed in lines 297-300. Table S2 provides all other model options.

F0AM simulations. I don’t like that both NO and NO2 are constrained individually. At least one should be left alone to cycle with the sun. F0AM has an option by which you can constrain total NOx but un-constrain individual NO and NO2.   

We are not making air quality simulations. The purpose of using the F0AM is to calculate OH, HO2 and RO2 from solving the set of equations associated with the chemical mechanism (MCM) at every time step of the run. Therefore, we constrain all other variables (NO, NO2, O3, VOCs, meteorology). Once we have OH, HO2, and RO2, we calculate P(O3) using NO.

Section 2.5.2. It would have been nicer to add some context about why we do these perturbations from a regulatory perspective. Do we expect future changes in land use land cover to result in more or fewer biogenic isoprene emissions? Are we doing this to disentangle the natural from anthropogenic VOC influences?

In the revised version we will clarify this point.

L329. Clouds?

Clouds were filtered out.

L333. Did any of these days get impacted by wildfires?

No.

L335. Is the CO pattern difference due to the traffic pattern or meteorological/topography effects?

The CO cycle is associated with the urban activity and work habits of citizens in Santiago. In this city, work hours extend into the evening and night, often past the work schedule, and citizens take a long time to return to their homes. In Quito, citizens usually leave work before dark and usually do not extend the eight-hour work schedule. We will add this information to the paper.

L356. Somewhere, it should be mentioned that these are real observations (and not simulations) and can be affected by clouds, an important component missing from the F0AM simulations.

The air quality data discussed in the manuscript correspond to observations during sunny conditions (overcast days were filtered out from the time series in both cities). The Methods state this important aspect, but we will make sure that this idea is clearly stated throughout.

L355. Albedo, aerosols, overhead ozone, and clouds can all modulate photolysis rates. There is too much emphasis on solar radiation.

We filtered out cloudy days. Albedo and the total column ozone were taken into account in the model. We will elaborate more on these aspects in the revised version.

L376. Because the results are based on a model, the authors should have a clear explanation of HOx budget. How does NO+RO2 contribute to these results? How about CO and HCHO? See Figure 2 in https://acp.copernicus.org/articles/20/13011/2020/acp-20-13011-2020.pdf. This section is incomplete. H2O can’t explain HOx budget alone in an urban setting.

We did not explain the HOx budget only with H2O. Formaldehyde photolysis is quantified and discussed in the paper. The discussion on HOx sources continues beyond line 376. We will double check that all relevant contributions are discussed.

L384. Is the current model setup appropriate for studying HONO? Do you consider all heterogenous chemistry involved with HONO chemistry, which depends on aerosol shape and light intensity?

The model was not set up to produce a specialized study about HONO chemistry. In this portion of the paper, we only show HONO photolysis as a source of OH for which we used the HONO output of the model and its frequency of photolysis. We used an explicit gas phase chemistry mechanism. The set of reactions was generated using the Master Chemical Mechanism from https://mcm.york.ac.uk/MCM/ and is accessible as model input in the link provided in the data availability statement.

L425. The discussion in Figure 8 came across as being hasty. It doesn’t convey how these many modeling scenarios explain the diurnal changes in L/Q.

In the revised version of the paper, we will improve the discussion about this figure.

L440. The whole paragraph is speculation. The authors do not study these and limit their focus to convection rather than other physiochemical processes (clouds and aerosols, dry deposition rates, background ozone, …).

Mixing conditions in Santiago are well known, these are not speculations. If we turn on a source of ozone in a stable mixing volume, the result will not be a surprise. 10 years of ozone soundings in Quito, whose shapes show deep mixing, are not speculations either. All these observations in Santiago and Quito are documented and discussed in previous peer-reviewed publications, please refer to the papers cited. Hence, from an observational perspective, we combine previous knowledge with our current results, and we find a partial explanation, important to our region, that is worth exploring in the future. Concisely, we found that similar ozone production rates, that cause ozone accumulation in one city but not in the other, point towards different mixing conditions as one potential explanation. We do not see a plausible reason not to make this logical connection. Aside from this physical aspect, we state in the paper that it is necessary to investigate other chemical paths for precursors, different from ozone production, such as nitrate particle and PAN formation.

Editorial comments:

L94. Responsible for instead of the culprit.

OK.

L191. Which model are you referring to? F0AM?

Yes.

L347. It’s hard to follow. More photochemistry in Santiago?

We will edit this part for clarity.

L385. What is the definition of NOx-rich environments?

We will replace by “VOC-limited also referred as NOx-saturated”.

Can you please change L1/(L1+L2) to L/Q? You can alternatively use LNOx/LROx (L1/L2).

OK.

-----------------------------------------------------

List of references

Baker et. al. (2007). Measurements of nonmethane hydrocarbons in 28 United States cities. Atmospheric Environment. https://doi.org/10.1016/j.atmosenv.2007.09.007

Borbon et. al. (2013). Emission ratios of anthropogenic volatile organic compounds in northern mid-latitude megacities: Observations versus emission inventories in Los Angeles and Paris. JGR Atmospheres. https://doi.org/10.1002/jgrd.50059

Brito et. al. (2015). Vehicular Emission Ratios of VOCs in a Megacity Impacted by Extensive Ethanol Use: Results of Ambient Measurements in São Paulo, Brazil. American Chemical Society. https://pubs.acs.org/doi/10.1021/acs.est.5b03281

Parra, R. (2017). Assessment of planetary boundary layer schemes of the WRF-Chem model in the simulation of carbon monoxide dispersion in the urban area of Quito, Ecuador. WIT Transactions on Ecology and the Environment, DOI: 10.2495/AIR170041. https://www.witpress.com/elibrary/wit-transactions-on-ecology-and-the-environment/211/36032

Vega, D., Ocaña, L., Parra, R. (2015). On-road traffic air pollutants emission inventory in the Distrito Metropolitano de Quito. Base year 2012. ACI Avances en Ciencias e Ingenierías, Vol.  7, No. 2, Pgs. C86-C94. 2015. https://revistas.usfq.edu.ec/index.php/avances/article/view/270/271

Hernandez, W., and Mendez, A. (2020). Robust Estimation of Carbon Monoxide Measurements. Sensors, 20(17), 4958. https://doi.org/10.3390/s20174958

Parra, R., Cadena, E., Flores, C. (2019). Maximum UV Index Records (2010–2014) in Quito (Ecuador) and Its Trend Inferred from Remote Sensing Data (1979–2018). Atmosphere. https://doi.org/10.3390/atmos10120787

Ren et. al. (2003). OH and HO2 Chemistry in the urban atmosphere of New York City. Atmospheric Environment. https://doi.org/10.1016/S1352-2310(03)00459-X

Shirley et. al. (2006). Atmospheric oxidation in the Mexico City Metropolitan Area (MCMA) during April 2003. ACP. https://acp.copernicus.org/articles/6/2753/2006/

Ren et. al. (2013). Atmospheric oxidation chemistry and ozone production: Results from SHARP 2009 in Houston, Texas. JGR Atmospheres. https://doi.org/10.1002/jgrd.50342

Sebol et. al. (2024). Exploring ozone production sensitivity to NOx and VOCs in the New York City airshed in the spring and summers of 2017–2019. Atmospheric Environment. https://doi.org/10.1016/j.atmosenv.2024.120417

Thank you for the input. 

In the revised version of the paper we present sensitivity experiments with a set of VOCs generated using Monte Carlo simulations and following suggestions by other reviewers.

OK.