the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 21 Jun 2023
Global impacts of aviation on air quality evaluated at high resolution
Abstract. Aviation emissions cause global changes in air quality which have been estimated to result in ~58,000 premature mortalities per year, but this number varies by an order of magnitude between studies. The causes of this uncertainty include differences in the assessment of ozone exposure impacts and in how air quality changes are simulated, and the possibility that low-resolution (~400 km) global models may overestimate impacts compared to finer-resolution (~50 km) regional models. We use the GEOS-Chem High Performance chemistry-transport model at a 50 km global resolution, an order of magnitude finer than recent assessments of the same scope, to quantify the air quality impacts of aviation with a single internally consistent, global approach. We find that aviation emissions in 2015 resulted in 21,200 premature mortalities due to particulate matter exposure and 53,100 due to ozone exposure. Compared to a prior estimate of 6,800 ozone-related premature mortalities for 2006 our estimate is increased by 5.6 times due to the use of updated epidemiological data which includes the effects of ozone exposure during winter, and by 1.3 times due to increased aviation fuel burn. The use of fine (50 km) resolution increases the estimated impacts on both ozone and particulate matter-related mortality by a further 20 % compared to coarse-resolution (400 km) global simulation, but an intermediate resolution (100 km) is sufficient to capture 98 % of impacts. This is in part due to the role of aviation-attributable ozone, which is long-lived enough to mix through the Northern Hemisphere and exposure to which causes 2.5 times as much health impact as aviation-attributable PM2.5. This work shows that the air quality impacts of civil aviation emissions are dominated by the hemisphere-scale response of tropospheric ozone to aviation NOx rather than local changes, and that simulations at ~100 km resolution provide similar results to those at two times finer spatial scale.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(1281 KB)
-
Supplement
(407 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1281 KB) - Metadata XML
-
Supplement
(407 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-695', Anonymous Referee #4, 07 Jul 2023
General comments:
The authos use the GEOSChem model to run 1-year simulation with and without aviation emissions to attibure mortalities to the changes in ozone and PM2.5. They use different resolutions and calculate the differences aviation related mortalities. In general the paper is well written, however, a deeper analysis that tries to explain differences in the modelling results due to resolution is lacking. Moreover, I have severe doubts concerning the method of an incremental approach that is applied in this case. Is the sum of all incremental approaches for all individual sources, such as industries, households, transport, aviation, etc. giving the total mortalities due to either ozone and PM2.5? Thunis et al. (2019) and the refernces therein clearly show that this is not the case. They have a very strong argument that incremental approaches can not be used for source (and here mortality) attribution. I think addressing these points is crucial before the paper may become accepted.Major comments
- I have some doubts whether the approach in calculating aviation related mortalities is applicable. Around line 160: "This is the quantity for which Turner et al (2015) determined a 12% increase (95% confidence interval: 6.0-18%) in respiratory mortality per 10 ppb increase in ozone exposure." To my understanding this analysis is based on situations where an enhancement of ozone can be attributed to an increase in an emission source. However, here, we have a constant emission of NOx by aviation that competes with other sources of NOx, such as industry, lightning etc. In that case Thunis et al. 2019 (and the literature cited therein) clearly pointed out that an increamental approach, with and without aviation emissions in not applicable. If the mortalities due to ozone are estimated with your approach for all individual emissions separately and added up would that give the same number as for all emissions? It is a highly non-linear system, so they answer is not (see also Tunis et a.). I think that the uncertainties due to the used method are much larger than due to different resolutions
- The analysis of the impact of different resolutions lacks quite some analysis.
a) How are regional pattern e.g. around major airports changed? and why?
b) How is the transport changed due to resolution changes? Convective transport, lighning? Scavenging?
c) Where is actually ozone production and loss rates changed? - I think the intercomparison with other studies mostly relies on the same model, right? What about other aviation simulations for NOx and aerosols?
Specific comments
Eq. 1 and 2 (in addition to the above comments)
I am wonderning, although knowing that this approach has been used frequently in the past, if this eq. is actually well representing the aviation induced mortality.
We have many sources for PM2.5 and ozone such as households, industry etc.
My understanding is that the mortality from all sources is M_Base.
Hence the sum of all mortalities dM_i from individual sources i (aviation, households, fires, ...) should give M_Base.
M_Base=Sum_i dM_i=Sum_i M_Base * (RR_i-RR_Base)/RR_Base = M_Base/RR_Base* Sum_i (RR_i-RR_Base)
However, in general, the sum of all changes of the surface concentration and exposure is not base concentaiton or exposure, respectively.
(see e.g. Thunis et al. 2019 and the literature therein). Thunis et al cearly indicated the inaccuracy of incremental approaches.
Emmons et al. (2012) showed a differences of a factor of 2-4 for a surface source in using contribution and incremental apporaches.
Local conditions the differences might be even larger. And the nonlinear function in eq. (2) is adding to the discrepancy.
Hence, to my understanding, here the mortality changes by switching off aviation is investigated and not the impacts from aviation.
Please make this clear in the text and title.Eq. 1/2 Are you evaluating the exposure and mortality on the basis of, e.e., daily values and then averaging or yre you using the mean values in the first place?
How large are the differences between the two approaches? So day-by-day variations in the transport pattern and e.g. chemical responss might be large enough to considerably change the exposure as a mean.213 "This provides additional evidence" I wouldn't call a missing correlation as an evidence. I suggest to rewrite by using a statement such as
"not contradicting to ..."228 "The greater resolution has two effects: physical phenomena are more finely resolved .." While I agree in general with the statement,
I think there is more analysis required to understand the impact of resolution on physical processes.
Are you sure that the simlaiton is more realistic? HOw is the vertical transport changed? Do yo have an anaylsis of convective
up- and downward transports that are changed. Is the dynamical lifetime affected, and if, by which process. Are there measurements to allow a
judgement on the quality of the higher resolved proesses? E.g. 222Rn? Are natural processes and sources changed? e.g. lIghtning NOx?
I think a much deeper analysis os required to allow a more detailed judgement.246 "The relative contribution of NA aviation emissions" plese rephrase and see above the comment based on Thunis et al.
Sect. 3 in general: The simulation is based on one year. The year 2015 is known for intense heat waves. Is there any information available on the
robustness of the results, e.g. annual variability, etc.Intercomarison: As far as I see most citations refer to the use of GEOSchem. right? What about other models?
In general amore detailed explanation is missing.Citation: https://doi.org/10.5194/egusphere-2023-695-RC1 -
RC2: 'Comment on egusphere-2023-695', Anonymous Referee #1, 09 Jul 2023
Review of "Global impacts of aviation on air quality evaluated at high resolution" by Sebastian D. Eastham et al.
After reviewing this manuscript, I do not think in its present form it is appropriate for ACP. The manuscript runs a chemical transport model at different resolutions to determine the change in near surface ozone and 2.5 micron Particulate Matter (PM2.5), and then presents 'health effects' as premature mortality. I think too much of the uncertainty and analysis is due to epidemiological uncertainty which really cannot be assessed in ACP. If the paper were to focus on the physical modeling it would be fine: but there is zero uncertainty estimated there. I think the manuscript needs to be a much better analysis of the physical modeling for ACP. If the focus is to be on the health effects and uncertainties, it should go in a health focused journal.I do not like the use of headline numbers of thousands of deaths, when there is no uncertainty in the abstract, values have changed by a factor of 5 from previous estimates, and some of the work is based on a single health study. As a physical scientist, I am not able to assess whether these are valid or not, and hence I do not think ACP is appropriate.
So ideally, this would be put in a health oriented journal. If the authors wish it to be in ACP, I think it should focus on the changes to Ozone and PM2.5. Specifically there needs to be a better assessment of uncertainty, which probably means using more than one meteorological year.
I also do not think the 'different resolutions' is particularly interesting since the meteorology is just interpolated to drive the model, and because the scale of resolutions is still within the range of using exactly the same parameterization methods. This might also not be a wise focus for the manuscript as it is not particularly strong.
Specific comments:
Page 1, L27: What is the uncertainty on these numbers? It seems that the uncertainty is entirely due to epidemiological factors? How much is due to ozone changes? Any of it? I am not convinced that this is useful at all given that air quality is highly non-linear and the extremes are not well reproduced by any scale of models 50-400km. Are the ozone numbers different? Suggest that things be focused around ozone and NOx quantification rather than the epidemiology.
Page 2, L56: Why so much lower? Can you summarize?
Page 2, L62: Global total from aviation?
Page 3, L72: can you explain how you can equate climate impact with air quality impact? This seems not to be a scientific question.
Page 3, L79: All of these resolutions are using basically the same type of model set up. What would be the impact of using a more detailed treatment that would explicitly resolve the non-linear nature of exposure? Are there parameterizations for this from regional air quality models? That would be more interesting.
Page 3, L81: Please explain what the coded resolutions refer to the first time it is used.
Page 3, L94: Are the in and out emissions impacts linear? E.g., if you take Global - NoNA + NoNA - Off, do you get the same answer as Global - Off?
Page 4, L98: why do you say CN for resolution and then use C?
Page 4, L104: What happens is you try to run at much higher resolution (say 5-10km). Can you do this, even for a limited time to gauge the physical impact on O3 and NOx? Using the same chemical mechanism. Versions of the MERRA-2 system have been run at these resolutions.
Page 4, L119: Are the same MERRA2 data used and just averaged or binned to lower resolution? How high resolution does the MERRA-2 data go? What if you used one of the GEOS ‘nature runs’ at 3 or 7km?
Page 6, L144: How large are the biases? If they are larger than the signal, then could the non-linearities impact the results?
Page 6, L153: Given that the model is uncertain and you are dealing with small perturbations, can you give an uncertainty estimate for (a) the difference in concentration due to aviation and (b) the uncertainty in your mortality estimates?
Page 6, L165: You base the ozone values on one study? That doesn’t seem appropriate. There must be others you can refer to?
Page 7, L172: Here you describe the uncertainty in mortality rates: why is this not propagated through, especially to the abstract.
Page 7, L182: Again, these uncertainty estimates need to be in the abstract.
Page 7, L185: The uncertainty interval for PM2.5 is ±8%? How different is that than a previous estimate? That seems ridiculously low. Also: what is the uncertainty due to the physical model difference in PM? This uncertainty is only due to epidemiology right?
Page 7, L186: The ozone mortality estimated uncertainty is ±30%. Yet the difference from previous work is 560%! Is that entirely due to differences in the epidemiological assumptions?
Page 7, L191: Figure 2 is hard to interpret due to use of the default python color scale. Maybe get rid of anything which is not significant (make those points white, not a dominant blue): you need a definition of what is a significant difference and what is just noise. I suggest the variability either within a year, or between multiple years could give you a standard deviation here.
Page 7, L191: Why would the O3 changes from aviation AT THE SURFACE be largest in the W. US and Tibetan Plateau? Does that really make sense? I suggest there is a transport problem with the model due to a terrain following coordinate and anomalous cross-isentropic transport to elevated topography. This is a common problem with transport schemes in global models.
Page 7, L193: So these particulates are secondary aerosols produced where there already is air pollution?
Page 8, L205: This seems anomalous in the model. Does observed near surface air in the Western US have higher background O3 than elsewhere? Representative of transport of higher altitude air to the surface?
Page 8, L218: This begs the question: are the epidemiological effects linear? Does a 10 ng/m3 change in PM2.5 have the same impact if the background is 10ng/m3 (100% increase) or 100ng/m3 (10% increase)?
Page 9, L220: Given the importance of secondary particulate matter, how uncertain is this production in the model? It seems as if this is one of the most uncertain elements.
Page 9, L238: are these absolute or relative values? If relative they are small, but if absolute they are large. And it’s confusing given the the previous percentages if they are relative.
Page 10, L252: see earlier comment: are these linear with the total effect over the US?
Page 11, L260: Figure 5: similar color scale problem to figure 2: you should blank regions without significant changes.
Page 12, L281: So you get exactly the same answer? That’s suspicious. Are you using the same exposure and same modeling tools to do this? What is the difference in their ozone and PM2.5?
Page 12, L287: I think this needs to be in the main text.
Page 12, L299: how much is attributable to any physical differences in the change in PM2.5 and Ozone? Initial background state and model biases?
Page 13, L322: PM2.5 emissions are not due to NOx are they? Maybe you need to remind the reader that NOx emissions are what change ozone since NOx is not mentioned in the results section.
Page 13, L328: I think this is disingenuous since you have really just limited the resolution differences to interpolation of input data. A proper assessment of resolution would alter the balance between parameterized and resolved quantities, but you do none of that and test only a limited range of resolutions and then apply them to a much higher resolution (1km) population data set. Which seems a big scale mis-match.
Page 13, L332: Why would inconsistent models yield a smaller difference? Is that chance? I would expect a larger difference perhaps?
Page 14, L335: This is difficult to follow. Are you saying that Venam et al had non US aviation emissions in both? Wouldn’t this tend to reduce changes ‘neglecting the larger response’? So why is the response larger? I don’t follow the logic.
Page 14, L351: You lost your uncertainty range again. Please put it here and in the abstract.
Page 14, L352: How much uncertainty results from the different meteorology in 2005 v. 2015?
Citation: https://doi.org/10.5194/egusphere-2023-695-RC2 -
RC3: 'Comment on egusphere-2023-695', Anonymous Referee #5, 31 Jul 2023
This manuscript presented a modeling analysis of global health impact associated with ozone exposure due to aviation emission. Impacts of anthropogenic emissions on air pollution have been thoroughly investigated over inland areas, yet the contribution from aviation remains poorly documented. The study applied a solid modelling tool and focused on an interesting topic. But there are two critical main issues need to be addressed before the acceptance could be considered.
First, the manuscript proposed a very interesting question at the introduction section but didn’t mention it in the discussion or conclusion. Line#57 mentioned: “whether the air quality impacts of aviation are dominated by local sources or are the result of larger atmospheric changes.”, and line#63 mentioned “there questions urgently need to be resolved”. So the introduction indicated this is one of the question that would be at least discussed in this study but unfortunately no such discussion was mentioned later. Section3.2 provided a detailed comparison of the estimated mortalities between this study and previous studies, but the most significant difference was due to using a more epidemiological data, while the proposed question remain unsolved.
Second, as a modeling study the manuscript lacks a necessary evaluation section. Almost all discussions were made based on simulation results, so it is very important to demonstrate the reliability or remaining uncertainty of the simulation results. Without solid evaluation the rest of discussion regarding contributions of aviation to air pollution and associated mortalities would be difficult to interpret.
There are a few other minor issues with the organization of the manuscript. For example, line#80 mentioned simulations were conducted for 400, 100, and 50km and section3 demonstrated the differences between. Apparently finer resolution can better reproduce atmospheric chemistry especially for ozone which is sensitive to mixing of NOx and VOCs. The manuscript indeed shows a large difference between fine and coarse simulations, but this seems more like a technical improvement other than a key innovative science finding.
Citation: https://doi.org/10.5194/egusphere-2023-695-RC3 -
EC1: 'Comment on egusphere-2023-695', Xavier Querol, 04 Aug 2023
Dear authors,
As you see the comments of the referees rise very critical points on the scientific quality of the manuscript.
Please be sure that you reply their questions and address changes enough to convince them because they will review again the revised version.
With very kind regards
Xavier
Citation: https://doi.org/10.5194/egusphere-2023-695-EC1 -
AC1: 'Response to reviews (RC1-3, EC1)', Sebastian Eastham, 14 Nov 2023
Dear Dr. Querol and reviewers,
Please find attached a single document which includes responses to all reviewer comments, along with a marked-up version of the manuscript which shows where additions have been made to address the comments. We are grateful to the reviewers for their valuable input, and believe that the edits we have made (including new information gained from a battery of additional simulations performed in service of these revisions) have significantly improved the manuscripts. We appreciate your patience as we have worked to address the concerns of the reviewers.
Regards,
Sebastian Eastham
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2023-695', Anonymous Referee #4, 07 Jul 2023
General comments:
The authos use the GEOSChem model to run 1-year simulation with and without aviation emissions to attibure mortalities to the changes in ozone and PM2.5. They use different resolutions and calculate the differences aviation related mortalities. In general the paper is well written, however, a deeper analysis that tries to explain differences in the modelling results due to resolution is lacking. Moreover, I have severe doubts concerning the method of an incremental approach that is applied in this case. Is the sum of all incremental approaches for all individual sources, such as industries, households, transport, aviation, etc. giving the total mortalities due to either ozone and PM2.5? Thunis et al. (2019) and the refernces therein clearly show that this is not the case. They have a very strong argument that incremental approaches can not be used for source (and here mortality) attribution. I think addressing these points is crucial before the paper may become accepted.Major comments
- I have some doubts whether the approach in calculating aviation related mortalities is applicable. Around line 160: "This is the quantity for which Turner et al (2015) determined a 12% increase (95% confidence interval: 6.0-18%) in respiratory mortality per 10 ppb increase in ozone exposure." To my understanding this analysis is based on situations where an enhancement of ozone can be attributed to an increase in an emission source. However, here, we have a constant emission of NOx by aviation that competes with other sources of NOx, such as industry, lightning etc. In that case Thunis et al. 2019 (and the literature cited therein) clearly pointed out that an increamental approach, with and without aviation emissions in not applicable. If the mortalities due to ozone are estimated with your approach for all individual emissions separately and added up would that give the same number as for all emissions? It is a highly non-linear system, so they answer is not (see also Tunis et a.). I think that the uncertainties due to the used method are much larger than due to different resolutions
- The analysis of the impact of different resolutions lacks quite some analysis.
a) How are regional pattern e.g. around major airports changed? and why?
b) How is the transport changed due to resolution changes? Convective transport, lighning? Scavenging?
c) Where is actually ozone production and loss rates changed? - I think the intercomparison with other studies mostly relies on the same model, right? What about other aviation simulations for NOx and aerosols?
Specific comments
Eq. 1 and 2 (in addition to the above comments)
I am wonderning, although knowing that this approach has been used frequently in the past, if this eq. is actually well representing the aviation induced mortality.
We have many sources for PM2.5 and ozone such as households, industry etc.
My understanding is that the mortality from all sources is M_Base.
Hence the sum of all mortalities dM_i from individual sources i (aviation, households, fires, ...) should give M_Base.
M_Base=Sum_i dM_i=Sum_i M_Base * (RR_i-RR_Base)/RR_Base = M_Base/RR_Base* Sum_i (RR_i-RR_Base)
However, in general, the sum of all changes of the surface concentration and exposure is not base concentaiton or exposure, respectively.
(see e.g. Thunis et al. 2019 and the literature therein). Thunis et al cearly indicated the inaccuracy of incremental approaches.
Emmons et al. (2012) showed a differences of a factor of 2-4 for a surface source in using contribution and incremental apporaches.
Local conditions the differences might be even larger. And the nonlinear function in eq. (2) is adding to the discrepancy.
Hence, to my understanding, here the mortality changes by switching off aviation is investigated and not the impacts from aviation.
Please make this clear in the text and title.Eq. 1/2 Are you evaluating the exposure and mortality on the basis of, e.e., daily values and then averaging or yre you using the mean values in the first place?
How large are the differences between the two approaches? So day-by-day variations in the transport pattern and e.g. chemical responss might be large enough to considerably change the exposure as a mean.213 "This provides additional evidence" I wouldn't call a missing correlation as an evidence. I suggest to rewrite by using a statement such as
"not contradicting to ..."228 "The greater resolution has two effects: physical phenomena are more finely resolved .." While I agree in general with the statement,
I think there is more analysis required to understand the impact of resolution on physical processes.
Are you sure that the simlaiton is more realistic? HOw is the vertical transport changed? Do yo have an anaylsis of convective
up- and downward transports that are changed. Is the dynamical lifetime affected, and if, by which process. Are there measurements to allow a
judgement on the quality of the higher resolved proesses? E.g. 222Rn? Are natural processes and sources changed? e.g. lIghtning NOx?
I think a much deeper analysis os required to allow a more detailed judgement.246 "The relative contribution of NA aviation emissions" plese rephrase and see above the comment based on Thunis et al.
Sect. 3 in general: The simulation is based on one year. The year 2015 is known for intense heat waves. Is there any information available on the
robustness of the results, e.g. annual variability, etc.Intercomarison: As far as I see most citations refer to the use of GEOSchem. right? What about other models?
In general amore detailed explanation is missing.Citation: https://doi.org/10.5194/egusphere-2023-695-RC1 -
RC2: 'Comment on egusphere-2023-695', Anonymous Referee #1, 09 Jul 2023
Review of "Global impacts of aviation on air quality evaluated at high resolution" by Sebastian D. Eastham et al.
After reviewing this manuscript, I do not think in its present form it is appropriate for ACP. The manuscript runs a chemical transport model at different resolutions to determine the change in near surface ozone and 2.5 micron Particulate Matter (PM2.5), and then presents 'health effects' as premature mortality. I think too much of the uncertainty and analysis is due to epidemiological uncertainty which really cannot be assessed in ACP. If the paper were to focus on the physical modeling it would be fine: but there is zero uncertainty estimated there. I think the manuscript needs to be a much better analysis of the physical modeling for ACP. If the focus is to be on the health effects and uncertainties, it should go in a health focused journal.I do not like the use of headline numbers of thousands of deaths, when there is no uncertainty in the abstract, values have changed by a factor of 5 from previous estimates, and some of the work is based on a single health study. As a physical scientist, I am not able to assess whether these are valid or not, and hence I do not think ACP is appropriate.
So ideally, this would be put in a health oriented journal. If the authors wish it to be in ACP, I think it should focus on the changes to Ozone and PM2.5. Specifically there needs to be a better assessment of uncertainty, which probably means using more than one meteorological year.
I also do not think the 'different resolutions' is particularly interesting since the meteorology is just interpolated to drive the model, and because the scale of resolutions is still within the range of using exactly the same parameterization methods. This might also not be a wise focus for the manuscript as it is not particularly strong.
Specific comments:
Page 1, L27: What is the uncertainty on these numbers? It seems that the uncertainty is entirely due to epidemiological factors? How much is due to ozone changes? Any of it? I am not convinced that this is useful at all given that air quality is highly non-linear and the extremes are not well reproduced by any scale of models 50-400km. Are the ozone numbers different? Suggest that things be focused around ozone and NOx quantification rather than the epidemiology.
Page 2, L56: Why so much lower? Can you summarize?
Page 2, L62: Global total from aviation?
Page 3, L72: can you explain how you can equate climate impact with air quality impact? This seems not to be a scientific question.
Page 3, L79: All of these resolutions are using basically the same type of model set up. What would be the impact of using a more detailed treatment that would explicitly resolve the non-linear nature of exposure? Are there parameterizations for this from regional air quality models? That would be more interesting.
Page 3, L81: Please explain what the coded resolutions refer to the first time it is used.
Page 3, L94: Are the in and out emissions impacts linear? E.g., if you take Global - NoNA + NoNA - Off, do you get the same answer as Global - Off?
Page 4, L98: why do you say CN for resolution and then use C?
Page 4, L104: What happens is you try to run at much higher resolution (say 5-10km). Can you do this, even for a limited time to gauge the physical impact on O3 and NOx? Using the same chemical mechanism. Versions of the MERRA-2 system have been run at these resolutions.
Page 4, L119: Are the same MERRA2 data used and just averaged or binned to lower resolution? How high resolution does the MERRA-2 data go? What if you used one of the GEOS ‘nature runs’ at 3 or 7km?
Page 6, L144: How large are the biases? If they are larger than the signal, then could the non-linearities impact the results?
Page 6, L153: Given that the model is uncertain and you are dealing with small perturbations, can you give an uncertainty estimate for (a) the difference in concentration due to aviation and (b) the uncertainty in your mortality estimates?
Page 6, L165: You base the ozone values on one study? That doesn’t seem appropriate. There must be others you can refer to?
Page 7, L172: Here you describe the uncertainty in mortality rates: why is this not propagated through, especially to the abstract.
Page 7, L182: Again, these uncertainty estimates need to be in the abstract.
Page 7, L185: The uncertainty interval for PM2.5 is ±8%? How different is that than a previous estimate? That seems ridiculously low. Also: what is the uncertainty due to the physical model difference in PM? This uncertainty is only due to epidemiology right?
Page 7, L186: The ozone mortality estimated uncertainty is ±30%. Yet the difference from previous work is 560%! Is that entirely due to differences in the epidemiological assumptions?
Page 7, L191: Figure 2 is hard to interpret due to use of the default python color scale. Maybe get rid of anything which is not significant (make those points white, not a dominant blue): you need a definition of what is a significant difference and what is just noise. I suggest the variability either within a year, or between multiple years could give you a standard deviation here.
Page 7, L191: Why would the O3 changes from aviation AT THE SURFACE be largest in the W. US and Tibetan Plateau? Does that really make sense? I suggest there is a transport problem with the model due to a terrain following coordinate and anomalous cross-isentropic transport to elevated topography. This is a common problem with transport schemes in global models.
Page 7, L193: So these particulates are secondary aerosols produced where there already is air pollution?
Page 8, L205: This seems anomalous in the model. Does observed near surface air in the Western US have higher background O3 than elsewhere? Representative of transport of higher altitude air to the surface?
Page 8, L218: This begs the question: are the epidemiological effects linear? Does a 10 ng/m3 change in PM2.5 have the same impact if the background is 10ng/m3 (100% increase) or 100ng/m3 (10% increase)?
Page 9, L220: Given the importance of secondary particulate matter, how uncertain is this production in the model? It seems as if this is one of the most uncertain elements.
Page 9, L238: are these absolute or relative values? If relative they are small, but if absolute they are large. And it’s confusing given the the previous percentages if they are relative.
Page 10, L252: see earlier comment: are these linear with the total effect over the US?
Page 11, L260: Figure 5: similar color scale problem to figure 2: you should blank regions without significant changes.
Page 12, L281: So you get exactly the same answer? That’s suspicious. Are you using the same exposure and same modeling tools to do this? What is the difference in their ozone and PM2.5?
Page 12, L287: I think this needs to be in the main text.
Page 12, L299: how much is attributable to any physical differences in the change in PM2.5 and Ozone? Initial background state and model biases?
Page 13, L322: PM2.5 emissions are not due to NOx are they? Maybe you need to remind the reader that NOx emissions are what change ozone since NOx is not mentioned in the results section.
Page 13, L328: I think this is disingenuous since you have really just limited the resolution differences to interpolation of input data. A proper assessment of resolution would alter the balance between parameterized and resolved quantities, but you do none of that and test only a limited range of resolutions and then apply them to a much higher resolution (1km) population data set. Which seems a big scale mis-match.
Page 13, L332: Why would inconsistent models yield a smaller difference? Is that chance? I would expect a larger difference perhaps?
Page 14, L335: This is difficult to follow. Are you saying that Venam et al had non US aviation emissions in both? Wouldn’t this tend to reduce changes ‘neglecting the larger response’? So why is the response larger? I don’t follow the logic.
Page 14, L351: You lost your uncertainty range again. Please put it here and in the abstract.
Page 14, L352: How much uncertainty results from the different meteorology in 2005 v. 2015?
Citation: https://doi.org/10.5194/egusphere-2023-695-RC2 -
RC3: 'Comment on egusphere-2023-695', Anonymous Referee #5, 31 Jul 2023
This manuscript presented a modeling analysis of global health impact associated with ozone exposure due to aviation emission. Impacts of anthropogenic emissions on air pollution have been thoroughly investigated over inland areas, yet the contribution from aviation remains poorly documented. The study applied a solid modelling tool and focused on an interesting topic. But there are two critical main issues need to be addressed before the acceptance could be considered.
First, the manuscript proposed a very interesting question at the introduction section but didn’t mention it in the discussion or conclusion. Line#57 mentioned: “whether the air quality impacts of aviation are dominated by local sources or are the result of larger atmospheric changes.”, and line#63 mentioned “there questions urgently need to be resolved”. So the introduction indicated this is one of the question that would be at least discussed in this study but unfortunately no such discussion was mentioned later. Section3.2 provided a detailed comparison of the estimated mortalities between this study and previous studies, but the most significant difference was due to using a more epidemiological data, while the proposed question remain unsolved.
Second, as a modeling study the manuscript lacks a necessary evaluation section. Almost all discussions were made based on simulation results, so it is very important to demonstrate the reliability or remaining uncertainty of the simulation results. Without solid evaluation the rest of discussion regarding contributions of aviation to air pollution and associated mortalities would be difficult to interpret.
There are a few other minor issues with the organization of the manuscript. For example, line#80 mentioned simulations were conducted for 400, 100, and 50km and section3 demonstrated the differences between. Apparently finer resolution can better reproduce atmospheric chemistry especially for ozone which is sensitive to mixing of NOx and VOCs. The manuscript indeed shows a large difference between fine and coarse simulations, but this seems more like a technical improvement other than a key innovative science finding.
Citation: https://doi.org/10.5194/egusphere-2023-695-RC3 -
EC1: 'Comment on egusphere-2023-695', Xavier Querol, 04 Aug 2023
Dear authors,
As you see the comments of the referees rise very critical points on the scientific quality of the manuscript.
Please be sure that you reply their questions and address changes enough to convince them because they will review again the revised version.
With very kind regards
Xavier
Citation: https://doi.org/10.5194/egusphere-2023-695-EC1 -
AC1: 'Response to reviews (RC1-3, EC1)', Sebastian Eastham, 14 Nov 2023
Dear Dr. Querol and reviewers,
Please find attached a single document which includes responses to all reviewer comments, along with a marked-up version of the manuscript which shows where additions have been made to address the comments. We are grateful to the reviewers for their valuable input, and believe that the edits we have made (including new information gained from a battery of additional simulations performed in service of these revisions) have significantly improved the manuscripts. We appreciate your patience as we have worked to address the concerns of the reviewers.
Regards,
Sebastian Eastham
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 578 | 243 | 28 | 849 | 45 | 15 | 12 |
- HTML: 578
- PDF: 243
- XML: 28
- Total: 849
- Supplement: 45
- BibTeX: 15
- EndNote: 12
Viewed (geographical distribution)
| Country | # | Views | % |
|---|---|---|---|
| United States of America | 1 | 415 | 50 |
| China | 2 | 63 | 7 |
| Germany | 3 | 50 | 6 |
| Netherlands | 4 | 41 | 4 |
| United Kingdom | 5 | 39 | 4 |
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
- 415
Guillaume P. Chossière
Raymond L. Speth
Daniel J. Jacob
Steven R. H. Barrett
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1281 KB) - Metadata XML
-
Supplement
(407 KB) - BibTeX
- EndNote
- Final revised paper