Ozone Sensitivity Analysis and Ozone Formation Regimes Division in the Beijing–Tianjin–Hebei Region Based on Satellite Remote Sensing Data

Round 1

Reviewer 1 Report

This paper tried to analyze the ozone formation sensitivity based on satellite observations of ozone precursors. It is crucial to make ozone control strategies. Overall the structure is properly designed and I can understand the authors' points from the descriptions. However, there are some errors/points that need to be corrected/clarified.

1. Line 55, "and some regions it may have ......" should be "and in some regions it may have......";

2. Line 70, "VOCs are limited to......" should be "VOCs are the limiting factor";

3. The paper did not tell us which "satellite observation data" (line 125) are used as input to derive the ozone profile in the text. How is the column concentration of ozone or other gases from satellite observations used to derive the vertical profile of ozone in the troposphere and the surface concentration of ozone? Figure 2 shows the "ozone profile data" are from satellite observation, however I did not find the relevant descriptions from the text.

4. Line 127, "the HCHO and NO₂ column concentrations provided by TROPOMI with those of ozone precursors". The sentence is confusing that HCHO and NO₂ are regarded as the ozone precursors, and the latter part of the sentence "with those of ozone precursors" is puzzling.

5. In section 2.2.1, multiple places used wrong spelling of TROPOMI as TROOMI.

6. Line 207, "MOODIS" or "MODIS"?

7. Figure 5, the left panel should be colored, right?

8. Line 319, the description "the statistical relationship between NO₂" is not complete.

9. Proper subscripts for NO₂ and O₃ should used  in the text. 

Minor corrections should be made to improve the readability. For example, "April September" should be changed to "April to September" or "April - September". More accurate spelling are required for abbreviations, for example, "TROPOMI" and "MODIS".

Author Response

Thank you for your valuable comments.

Author Response File: Author Response.pdf

Reviewer 2 Report

In this manuscript, Song et al. investigated the formation regime of surface ozone in Beijing Tianjin Hebei (BTH) region based on the ratio of formaldehyde to nitrogen dioxide derived from TROPOMI data. The threshold range for the transitional regime of ozone formation in the BTH region was determined by polynomial fitting, and the spatial variations of ozone formation regimes were analyzed for the high ozone season. The results of this study could provide scientific basis for effective control and mitigation of surface ozone pollution on the regional scale of BTH. I recommend it to be accepted after some revisions.

The specific comments can be found in the "Coments.docx" file attached.

Comments for author File: Comments.pdf

The quality of English language need to be greatly improved by native English scholars.

Author Response

Thank you for your valuable suggestions. These suggestions are not only very helpful to my current paper, but also lay a foundation for the next step.

Author Response File: Author Response.pdf

Reviewer 3 Report

This article entitled “Ozone sensitivity analysis and FNR threshold division in Beijing Tianjin Hebei region based on TROPOMI data” try to identify the formation regime of ozone by using the FNR method. This study provides theoretical accordance to determine O₃ mitigation strategy in the North China Plain. I have major questions that should be addressed before it can be accepted:

Major comments:

In the introduction, you have clarified that the threshold classification criteria for FNR has regional difference due to the influence of meteorology and other factors. However, you didn’t discuss this issue in your results. The threshold classification criteria for FNR may differ between different areas within the BTH region.

You use machine learning to acquire near-ground O₃ concentrations covering high-resolution grids. Did you verify the established model? I suggest to text the performance of the model by using test dataset.

Sec 2.3.3: the data point used for the analyzed should be further explained. Does one data point represent one grid? How about the spatial resolution? You have stated that there are 2138 equal parts in total. In Fig. 6, only several hundred data are presented. How do you filter out other data points?

In the abstract, you spend a lot of space discussing the NO2 and HCHO concentrations. I suggest to mainly focus on O₃ sensitivity regime in the abstract.

Other comments:

It is not good to use abbreviation in the title.

Line 19: “BTH” should be defined here.

Lines 35-39: Related references should be provided.

Line 55: “and some regions it may have the opposite effect” has grammatic error.

Line 63: HCHO is not defined.

Line 70-71: “VOCs are limited to” this expression sounds weird.

Lines 80: “have” should be “has”.

Lines 78-80: The sentence is unclear. Do you mean the differences in these factors affect the threshold of FNR? In addition, the related references should be provided here.  

Line 127: “with those of ozone precursors” is redundant.

Lines 168-169: It seems that your aim is to acquire the near-surface O₃ concentration that can cover all grids, rather than to acquire O3 concentrations that can accurately reflect the overall ozone concentration in the troposphere. If so, this sentence is not appropriate here.

Line 221: Do you mean the urban expansion areas include the urban aggregation areas.

Line 223-225: This sentence is unclear to me.

Line 230: “within a certain HCHO/NO2 range”. Why does the ozone excess probability relate to HCHO/NO2 range?

Line 236: “attachment” is redundant.

Line 244: “near surface” should be “near-surface”.

Line 259: The statement that HCHO is lowest in April 2021 contradicts with above statement that the lowest concentration occurs in winter.

Figure 4: The changing rates and statistical significance should be given.

Line 265: I would rather to say “the temporal variation trend of ….”

Line 280: Why do you use “However” here?

Lines 296-298: It seems that the reduction of FNR in urban expansion is larger than in non-urban areas, which contradicts with above statement that agglomeration<urban expansion<non-urban areas.

Line 313: “the ozone exceeding the standard rate” this sentence sounds weird.

Line 314: You use several different names: “ozone excess probability”, “ozone exceeding the standard rate” and “ozone exceedance probability” for one definition. This is very confusing.

Figure 6. “probabilitu” should be “probability”.

Figure 6: The determination of the range of FNR for transitional regime should be clearly clarified. In addition, the meaning of the blue dots and lines are not explained and discussed.

Line 349-350: This sentence is grammatically incorrect.

Lines 355-356: Primary emissions are also important sources of HCHO.

Line 362-363: “controlled by NOx or VOCs” is not a good expression. It better to say “is VOC-limited or NOx-limited”

Line 388: “the ozone generation sensitivity threshold” used here is not approximate. From my understanding, the ozone generation sensitivity threshold refers to the value of FNR that distinguish VOC-limited and NO_X-limited regime. However, in this paragraph you mainly discuss the change in O3 sensitivity regime rather than the threshold value.

There are a lot of grammatic errors that should be carefully checked.

There are lots of grammatic problems that should be double checked.

Author Response

Thank you for your valuable suggestions. These suggestions are not only very helpful to my current paper, but also lay a foundation for the next step.

Author Response File: Author Response.pdf

Reviewer 4 Report

This paper uses satellite data and ground-level ozone measurements to analyze sensitivity of ozone levels to VOC and NO_x as a function of location and year in an important region of China using data from 2019-2022. Satellite data is used to obtain formaldehyde, NO₂ and O₃ concentrations throughout the region, where formaldehyde and NO_x are used as surrogates for VOC and NO_x emissions, and where ground-level O₃ data are used in conjunction with the satellite data to obtain ground level O₃ estimates throughout the region. The ration of formaldehyde to NO₂ (FNR) is used as a measure of O₃ sensitivity to VOC and NO_x and the distribution FNR throughout the region and in various land-use types for each of these years is discussed. The results are generally consistent with other work indicating that the urban areas of China tend to be VOC limited while rural areas are NO_x limited, as is generally well known for other countries as well as China. The observations and trends were discussed, though they did not really provide any new insights for the reasons for the trends or an adequate discussion of the differences observed from year to year.

The methodology employed appears to be useful for obtaining data for spatial and temporal distributions of O₃ sensitivity to emissions, and the data obtained are interesting and should be useful for developing O₃ control strategies in this part of China. This is well within the scope of this journal and is therefore appropriate for publication on this basis. However, this paper has a number of problems that need to be addressed prior to publication. These are discussed below, in approximate order of importance.

The greatest problem I have with this work is lack of adequate discussion and analysis of uncertainty and variability. For example, the aggregated data shown on Figure 5 are averages of many data points, and without error bars or some other measurement of variability of the data, the extent to which the annual and land-type differences shown on that figure are statistically significant is uncertain, especially for formaldehyde, where the differences are relatively small.

There is also inadequate background discussion of uncertainties involved with using FNR based on satellite data as an indicator of regional ozone sensitivity. The FNR range corresponding to the transition region is different in this study compared to that observed in other studies, and the potential roles of VOC reactivity, significant differences in atmospheric lifetimes of formaldehyde (which is regenerated by reactions of secondary VOC pollutants) and NO_x (which is not), and role of temperature (which affects VOC reactivity), transport, mixing heights, and other aspects meteorology are not discussed. There needs to be a citation of how FNR is derived in the U.S. on Lines 76-78. More discussion of other studies of previous studies using FNR would have been useful in the introduction.

The presentations of the data are inadequate in some respects, and resulting in opportunities to fully exploit the data obtained being missed. Both Figures 4 and 5 should include trends in ozone or OEP data, and it would be instructive to see separate plots like Figure 4 and Figure 6 for the three land use types.

Figure 6 is important but it is difficult to interpret because it lacks a caption indicating exactly what the curves and points refer to. I think I might have figured it out, but if the black balls refer to FNR data then there does not appear to be indications of areas where increasing FNR means lower O₃ in the VOC-limited region, contrary to expectations, and the extrapolations of their third order fit that was derived primarily from highly scattered data, primarily from NO_x-limited regions. This extrapolation is not meaningful for the high FNR, VOC-limited region. Had there been a separate plot like Figure 6 for the urban regions, the cases of decreasing O₃ as FNR increases in the VOC-limited regions may be more evident, and the data may be less scattered, making polynomial fits through the data possibly somewhat more meaningful.

Figure 7 is also important, and it is unfortunate that the plots are so small that they are difficult to read. At least the captions on the right size indicating what data are shown should be more conspicuous. Do the colors on the FNR plot on this figure correspond to the three regions of O₃ sensitivity, as they do on Figure 8? If so, this should be indicated on the caption and more clearly in the discussion in the text.

It is unclear because of lack of adequate discussion of uncertain and variability that the annual differences for the 4-year period are statistically significant. If they were, one would think that 2019 should be the outlier because it was prior to China's draconian zero-Covid policies, which were in effect for the other three years. One would think that these would affect emissions, but available information or ideas regarding this are not discussed. However, 2021 seems to be the outlier based on their data, with Figures 5 and 8 suggesting only minor differences between 2019, 2020, and 2022. Could meteorology, which could affect VOC reactivity through the role of temperature or mixing heights and transport, explain why 2021 is so different? The role of meteorology is not discussed.

It would have been useful if they discussed whether the trends shown on Figure 5 are consistent with trends in emissions inventories for this region. If the authors believe that emissions inventories in China are inadequate to determine trends then this should at least be stated.

The discussion of their "machine learning" methodology to process ground level O₃ data is unclear. They should state what exactly they mean by an "inversion model" and its purpose. Is it intended to calibrate or correct the satellite O₃ data, improve resolution or sensitivity, or interpolate between ground-level measurements to areas between the measurement stations? What are they fitting in the fitting models?

The discussion of the study area in Section 2.1 is important background and should be in the introduction, not the "Materials and Methods" section.

Some of the writing is awkward and could improved. For example, the sentence on line 111-115 needs to be rephrased.

The acronym BTH used in the abstract should be defined. What is the meaning of the "[2.0,3.1]" on line 19? It looks like the "[1.0,2.0]" should be replaced by some word or phrase, but I don't know what. Shouldn't the reference to Figure 7 on line 362 actually be a reference to Figure 8? Shouldn't the phrase "VOCs are limited" in line 70 be changed to "Ozone sensitive to VOCs"?

 My recommendation is only accept after major revision and re-review.

Mostly understandable but awkward in places. See Comments to Authors.

Author Response

Thank you for your valuable suggestions. These suggestions are not only very helpful to my current paper, but also lay a foundation for the next step.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

In general, the author didn’t improve the paper according to my comments during the last round. I suggest major modifications again. In addition, I hope the author can provide a point-to-point response, and show where you revise in the manuscript and provide the sentences or Figures that you updated.

Major comments:

In the introduction, you have stated that the threshold changes among different regions and you aim to accurately understand the spatiotemporal variability of the thresholds. However, your results and discussion didn’t solve this problem. You use a constant threshold to distinguish O3 sensitivity regimes over BTH. I suggest to change your major aiming stated in the introduction and tune the tone.

I have suggested to text the performance of the model by using test dataset. You have replied to me in the response. However, I didn’t see any relevant modifications in the manuscript. I suggest to discuss it in the text and present the figure.

For Fig. 6, you should clarify in the manuscript why the number of points is far fewer than total number of grids. Otherwise, this will make readers confusing.

For the abstract, I insist that the discussion about the concentrations of NO₂ and HCHO should be deleted or only be mentioned briefly, because they are not your emphasis of this study. You should give the major conclusion about the spatial distribution and changing trends of FNR and O₃ sensitivity regimes.

 “Beijing Tianjin Hebei” should be “Beijing-Tianjin-Hebei”.

Line 74: Iimiting should be limited.

Line 135: “Surface ozone” should be “surface ozone”.

For other comments that I proposed in the last round, I found the authors didn’t revise accordingly. Please provide a point-to-point response, and show where you revise in the manuscript and provide the sentences or Figures that you updated.

 It is not good to use abbreviation in the title.

Line 19: “BTH” should be defined here.

Lines 35-39: Related references should be provided.

Line 55: “and some regions it may have the opposite effect” has grammatic error.

Line 63: HCHO is not defined.

Line 70-71: “VOCs are limited to” this expression sounds weird.

Lines 80: “have” should be “has”.

Lines 78-80: The sentence is unclear. Do you mean the differences in these factors affect the threshold of FNR? If true, the related references should be provided here. Otherwise, your statement is baseless

Line 127: “with those of ozone precursors” is redundant.

Lines 168-169: It seems that your aim is to acquire the near-surface O3 concentration that can cover all grids, rather than to acquire O3 concentrations that can accurately reflect the overall ozone concentration in the troposphere. If so, this sentence is not appropriate here.

Line 221: Do you mean the urban expansion areas include the urban aggregation areas.

Line 223-225: This sentence is unclear to me.

  Line 230: “within a certain HCHO/NO2 range”. Why do you stress this for a definition of ozone exceedance probability? I didn’t see the relevance.

Line 236: “attachment” is redundant.

Line 244: “near surface” should be “near-surface”.

Line 259: The statement that HCHO is lowest in April 2021 contradicts with above statement that the lowest concentration occurs in winter.

Figure 4: The changing rates and statistical significance should be given.

Line 265: I would rather to say “the temporal variation trend of ….”

Line 280: Why do you use “However” here?

Lines 296-298: It seems that the reduction of FNR in urban expansion is larger than in non-urban areas, which contradicts with above statement that agglomeration<urban expansion<non-urban areas.

Line 313: “the ozone exceeding the standard rate” this sentence sounds weird.

Line 314: You use several different names: “ozone excess probability”, “ozone exceeding the standard rate” and “ozone exceedance probability” for one definition. This is very confusing.

Figure 6. “probabilitu” should be “probability”.

Figure 6: The determination of the range of FNR for transitional regime should be clearly clarified. In addition, the meaning of the blue dots and lines are not explained and discussed.

Line 349-350: This sentence is grammatically incorrect.

Lines 355-356: Primary emissions are also important sources of HCHO.

Line 362-363: “controlled by NOx or VOCs” is not a good expression. It better to say “is VOC-limited or NOx-limited”

Line 388: “the ozone generation sensitivity threshold” used here is not approximate. From my understanding, the ozone generation sensitivity threshold refers to the value of FNR that distinguish VOC-limited and NOX-limited regime. However, in this paragraph you mainly discuss the change in O3 sensitivity regime rather than the threshold value.

There is space to improve the language

Author Response

"Please see the attachment." 

Author Response File: Author Response.pdf

Reviewer 4 Report

The revised manuscript did not address my most important comments and the some of the modifications regarding uncertainty made me change the "Scientific Soundness" rating from "Average" to "Low". My overall recommendation to reconsider after major revisions is unchanged.

My major concern was inadequate discussion of uncertainty and variability, and the lack of "error bars" indicating uncertainty and variability in Figures 4 and 5. The "confidence bands" on the revised Figure 4 look unrealistically small. The data show a number of points far outside the bands. They don't say what the "confidence bands" refer to, but it can't be spatial variability or variability over shorter time periods -- that is not credible. There must be much greater variability in the column data throughout the entire country of China, and the days for the days within the averaging periods. The same can be said about the error bars added to Figure 5. They are extremely small compared to what might reasonably expect as variability throughout the years and throughout the regions. There has to be some uncertainty on how the three regions are defined, which should also cause variability for the cases where the regions are very different. They stated in the conclusions that there are possible errors in the satellite retrieval process, but this certainly isn't reflected in their very small "confidence" bands.

There were no changes to Figure 6 following my suggestions, nor was it addressed in the response. I wasn't suggesting future research -- I was suggesting making the figure clearer and removing the unsupportable extrapolation in the low FNR region.

They gave no real response to my suggestion that they include ozone trends in Figure 4. What they said is that it would result in too many figures and that they consider them "relatively important" -- presumably meaning they consider them "relatively UNimportant". Since the subject of the paper is sensitivities of factors affecting ozone and how they changed over the 4 year period, the trends and variability in ozone over that period is hardly unimportant. Furthermore, addressing this would only add an additional curve to Figure 4, having minimal impact on journal space.

There was no discussion of the possible role of meteorology in the year-to-year variability. They did add some additional discussion of what the meteorology of the area is like, but not its variability. Their response was that this is a useful area of future research. However, all that is needed to address this in the current paper is to state in the discussions and conclusions that meteorological variability MIGHT explain some of their results, and (if this is their intention) will be a subject of future work.

Similarly, there is still inadequate discussion of their results in terms of possible changes in emissions during this period. Their response states that the satellite data are consistent with inventories in terms of where NO2 concentrations are high, but year-to-year variability is a different issue. There was also no mention in the response concerning the possible role of Covid in emissions and therefore ozone formation. If the authors believe that this is not an important factor, they should at least say so and why they believe this to be the case.

The figure they gave in response to my suggestion to show ozone trends in Figures 4 and 5 had no relevance to my suggestion, but the figure showing the map of locations of the three major land type regions was very interesting and relevant, and should be included in the paper. Perhaps it could be done as an overlay or next to the map on Figure 3, with very little impact on the journal space required. Comparing this map with the maps of FNR on Figure 7 is also highly relevant, and they should discuss the differences and similarities. This is highly relevant to their conclusion that urban areas are NOx limited -- this doesn't always seem to be the case based on comparing these maps.

I suggested that Section 2.1 is more appropriate for the introduction than the materials and methods section. The response said they modified this, but in fact this was not changed.

I did not make a major effort to review this -- you should rely on the comments of the other reviewers in this regard. Overall, the English was understandable to me.

Author Response

"Please see the attachment."

Author Response File: Author Response.pdf

Round 3

Reviewer 3 Report

The authors have addressed most of concerns that I proposed. This article can be accepted after addressing the following minor suggestions.

Line 122: “so solve25“, what is this?

Line 196-201: Figure 3 does not show the validation result. Do you mean Figure 2?  In addition, the explanation of Horizontal and vertical axis in the text is inconsistent with the description in the Figure. The Figure 2 looks very casual, such as the definition of the axis, and RMSE and MAE are not defined.

Line 342: “rad” should be red.

There are lots of other grammatic errors that should be double checked.

There are lots of grammatic problems that should be double checked.

Author Response

 "Please see the attachment."

Author Response File: Author Response.pdf

Reviewer 4 Report

This revised manuscript addresses my concerns sufficiently that I change the recommendation to "Accept after minor revision". The areas that need to be addressed include the following.

Are the points shown on Figure 4 for NO₂, HCHO, and FNR (but not O₃) the monthly averages of the columns throughout the BTH reason. If so, this should be stated explicitly -- if not then this needs to be clarified. Assuming that the fitting curves are supposed to fit the points shown, the curve for HCHO does not fit the data very well for the early years. They seem to be forcing the data to a function that assumes that the magnitude of the seasonable changes (min vs. max) are the same each year, which clearly is not the case for HCHO for at least 2 of the years. The fitting curve for HCHO needs to be improved to reflect the data, not force artificially to fit an assumed relationship that doesn't work all that well.

There needs to be a "y" axis label showing O₃ concentrations on Figure 4. If they are the same as the NO₂ and HCHO concentration, the label should include O₃ as well as NO₂ and HCHO.

The paper needs to state what the error bars on Figure 5 mean. The response to my comments, they said they "adjusted the error bars for Figure 6 to 15% of the data". By "Figure 6" I assume they mean Figure 5 because Figure 6 doesn't have error bars. Where did the 15% come from? I hope it is not just something they arbitrarily came up with to make it look more credible to me. What I was thinking of is having the error bars show something like the standard deviations (e.g., one sigma) derived when the annual averages were calculated. In that case, the percentage should be different each year. 15% may look more credible than what was shown before, but it needs to be based on something, and explained.

At least Figure 6 now has a better fitting curve for ozone exceedence probability than they had before. I wonder why they removed the cumulative frequency curve, which has some interest? I would think it would fit the data at low FNR better using their better exceedence probability fit curve.

They moved what was Section 2.1 to the introduction as I suggested, but they didn't relabel what was Section 2.2 and 2.3 to 2.1 and 2.2.

Line 122 has "so solve25" -- should this be "to solve". The phrase "changes in meteorological" is awkward.

Overall the English is understandable, though there are a few awkward places.

Author Response

 "Please see the attachment."

Author Response File: Author Response.pdf