Retrieval of daytime mesospheric ozone using OSIRIS observations of O2 (a1Δg) emission

Li, Anqi; Roth, Chris Z.; Pérot, Kristell; Christensen, Ole Martin; Bourassa, Adam; Degenstein, Doug A.; Murtagh, Donal P.

doi:https://doi.org/10.5194/amt-13-6215-2020

Articles | Volume 13, issue 11

https://doi.org/10.5194/amt-13-6215-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/amt-13-6215-2020

© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 13, issue 11

Research article

|

20 Nov 2020

Research article |

| 20 Nov 2020

Retrieval of daytime mesospheric ozone using OSIRIS observations of O₂ (a¹Δ_g) emission

Anqi Li, Chris Z. Roth, Kristell Pérot, Ole Martin Christensen, Adam Bourassa, Doug A. Degenstein, and Donal P. Murtagh

Download

Final revised paper (published on 20 Nov 2020)
Preprint (discussion started on 29 Apr 2020)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC1: 'Review', Anonymous Referee #1, 20 May 2020
- AC1: 'Reply to ref #1', Anqi Li, 24 Aug 2020
RC2: 'Review', Anonymous Referee #2, 28 May 2020
- AC2: 'reply to ref#2 (the formatted text is in the 'supplement file')', Anqi Li, 24 Aug 2020

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

AR by Anqi Li on behalf of the Authors (24 Aug 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (31 Aug 2020) by Bernd Funke

RR by Christian von Savigny (09 Sep 2020)

Suggestions for revision or reasons for rejection

General comments:

Overall the authors did a good job responding to the points raised in my earlier review and I also find the new parts of the manuscript interesting and relevant. I have some more specific comments that I ask the authors to consider, most of them very minor. One more general point is the treatment of the measurements made close to local sunset. The equilibrium calculations are only done for the morning sector, i.e. measurements after local sunrise. But the effects will also become important before sunset, right? IRI also takes measurements close to sunset, but it’s not clear, whether these were processed or not or treated in a special way? I apologize if I missed something here.

Specific comments:

Line 13: ´However, the IRI ozone data set is consistent with the compared data set about the overall atmospheric distribution of ozone.´

This sentence is somewhat odd. I suggest replacing, ´about´ by ´regarding´ or ´in terms of´, e.g..

Line 14: ´We attribute these differences TO´

Line 15: ´This implies that´

It’s unclear what ´This´ refers to and the logical connection between the sentences is unclear. I guess you don’t really intend to argue that the retrieval should be applied to the entire data set, because of the calibration issues and potential problems with the photochemical model?

Line 43: ´especially OF the species´ ?

Same line: ´whose lifetime are´ -> ´whose lifetime is´ or ´whose lifetimes are´ ?

Line 81: ´And agree, despite their intrinsically´

This is not a complete sentence.

Line 106: ´this small number .. have´ -> ´this small number .. has´

Line 156: The usual per nm wavelength dependence of the radiance, common in remote sensing observations of a spectrum, it is not used in this formulation as the measurement is an integral over the infrared band filter´

The quantity with the units you report is generally called ´radiance´. If a ´1/nm´ is added to the units, the corresponding quantity is generally called ´spectral radiance´. In this respect, calling your quantity ´radiance´ is absolutely fine.

Line 166: ´the error in the measured digital number of read-out electronics´

This is not well phrased and may be misunderstood. Please correct.

Fig. 1: Is there a simple reason, why the relative calibration errors changed with altitude?

Line 194: ´((´ and ´))´

Caption Fig. 2: I suggest deleting ´size´ in ´error size´

Line 226: ´This, because´

Something is missing here.

Line 238: ´returning back´ -> ´returns back´

Caption Fig. 4: I suggest deleting ´size´ in ´error size´

Caption Fig. 4: ´, with ’nan’ labels the sunrise is absence at the summer pole´

Please improve grammar.

Same line: ´the cross section´

It’s simply ´vertical profiles´, not cross sections, right?

Line 260: ´will lead to under-estimate ozone´ -> ´will lead to an underestimation of ozone´

In addition: I’m wondering, whether such an approach will always underestimate ozone. You are arguing for the morning sector, i.e. after local sunrise, but what about the evening sector (before sunset). If I understand correctly, your argument for the near-sunrise measurements is, that the O2(a1Delta) state has a relatively long lifetime (74 min.) and when it is formed (mainly by O3 photolysis) after sunrise, the emission at a certain time does not reflect the concentration of O2(a1Delta) at this time. A simplified approach would then underestimate ozone, I agree. But what about the evening sector? O3 will increase around sunset, because O3-photolysis gets smaller and smaller. The emission, however, will reflect the O2(a1Delta) production at an earlier point in time. It appears that ozone will also be underestimated, if these effects are not taken into account, right?

I suggest explaining the difference between measurements close to local sunrise and sunset in a few sentences. And I think your approach with the equilibrium level does only work for the measurements after sunrise, right? Perhaps this should be mentioned as well.

OSIRIS will also make (close to) sunset measurements of course. I think it’s not mentioned in the paper that these measurements are not used (perhaps I missed it). This seems an important point, I think.

Line 313: ´while the other photochemical sources are only sensitive below 90km´

The 'g_a-process' is also strongly dependent on SZA above 90 km, right, i.e. the statement is not fully correct.

Line 314: ´The Barth-type mechanism contributes very little and mainly between 100-110km´

Well, it's rather 90 - 100 km, right?

Line 329: ´x´ should be bold like in the equation below.

Line 353: ´such an assumption will lead to an under-estimation of the derived ozone.´

See my earlier comments on this aspect.

Caption Fig. 7: ´A naive estimation´

I suggest mentioning briefly in what sense this estimation is naive.

Same caption: Please mention, where he ´equilibrium index´ is described, because it hasn't been discussed, when the main text refers to this Fig. the first time – or refer to the Fig. later.

Line 366 – 369: I suggest mentioning explicitly, that [O2(a1)] is zero before sunrise, will then rise and assume the equilibrium state at some point. In principle this is obvious, but this statement would have helped me to understand the approach more easily.

Line 373: ´As the ratio t/$\tau$ to be 1.6, 2.3, 3, 4´

Please improve grammar.

Equation (20): is the power 8 intended? If yes, why was this power used?

Line 395: ´No data is available at the tropics´

Is it really 'no data'? This is difficult to see because of the color scale. Looking at the Fig. it seems some data is available.

Line 399: ´is as large as 50% of positive bias´

Please improve grammar.

Line 400: Suggest replacing ´The detail investigation may´ by ´A detailed investigation of the reasons for these differences may ..´

Line 401: ´detail modelling´ -> ´detailed modelling´

Same line: ´focuses´ -> ´foci´

Line 403: ´source´ -> ´sources´

Table 1: ´ <20% below 90 km, >100% above 90 km´

This is a step from <20% to > 100% at 90 km! This doesn't really seem plausible?

Line 449: ´the thermal emission line´

It’s a band, not a line, right?

Line 472: ´the agreement .. worsenS´

Line 477: ´in an altitude-time series´

It’s not really time series, right?

Line 481: ´in both hemisphereS´

Line 482: ´that THE IRI ozone ..´

Line 484: ´while BEING relatively stable´

Line 490: ´values exist in the regions between the secondary and primary ozone layer IN THE SMR DATA.´

Caption Fig. 13: ´colour scales of 2D-histograms´ -> ´greyscales of 2D-histograms´

Same line: ´The upper colour bar´ -> ´The upper greyscale´

Line 506: ´of daytime measurementS´

Line 511: ´with still more data NEAR the summer pole´

Line 512: ´IRI loses most of the data above 70 km in the tropics, since they were mostly measured very close to the local sunrise.´

What about the measurements carried out close to sunset?

Line 527: ´may be THE main reasonS´

Line 528: tidal effects are also non-negligible at altitudes lower than 90 km.

Fig. 14, top left panel: I suggest reducing the size of the legend – or making it more transparent.

Line 544: ´a clear a´ -> ´a clear´

Line 562: ´do not measure over the same´ -> ´do not cover the same´

Line 570: ´The inter-comparisons .. showS´

Hide

RR by Anonymous Referee #2 (15 Sep 2020)

Suggestions for revision or reasons for rejection

I would like to thank the authors for considering and addressing all my comments and for the large effort they have undertaken for improving the manuscript. Many of them have been taken, which I acknowledge. Very valuable are, for example, the equilibrium index and the absorption correction.

I noted, however, that my major recommendations on the content of the manuscript have not been taken, e.g., to focus the manuscript only on the description of the dataset and present a proper validation of the whole dataset and do not present partial studies on a limited comparison and on the O3 behaviour but leave them for further studies.

I made those comments from the perspective of an eventual community user and for the potential of the paper for being useful and citable. With that option the message would be "here there is a new O3 middle atmosphere dataset". As it is written, however, the message is just that a dataset will be available in the future.

As for the comparison with other datasets and the discussion on the O3 behaviour, they are partial and incomplete studies (e.g., only some months, or the latitudinal dependence shown only for one month) and therefore the conclusions reached might not be fully valid. That is, as a user, I would not fully trust them. As an example, the authors show, as a typical example of the good agreement between IRI with OS and SMR, one orbit and one profile, from the many millions(?) available. Again, as a user, I would like to see a statistical study. That single profile in the bottom panel of Fig. 11 would tell me little.

It seems the authors would like to publish two papers on this dataset. One, this, with the message that there will be available a dataset and, another, where the proper dataset will be presented with its validation. I think that is also a fair view, although I do share. The same applies to the comparison and O3 behaviour, I would leave them out but if the authors prefer to include them, I would not object. Therefore, I am ready to accept the publication of this work in ATM with its current major contents.

I am giving below a couple of comments (the most important) on the revised manuscript and minor suggestions (mainly editing).

1) Error budget. Overall, I am satisfied with the analysis. However, it seems there is a mixed up between random (also usually called instrumental, noise or precision) and systematic errors. Something which I understood nearly the end of the manuscript but needs to be clarified. For example, in legend of Fig. 1, we read "total error". This clearly does not include the systematic error discussed later on (Table 1). Is this total "calibration" error? or instrument (noise) error? Or a contribution to the total systematic error?
See also comment below on:
- the comment on l. 245.
- Legend of Fig. 4
- in l. 397 it is clearly written: "... has a "precision" of around 5-20\% based on the retrieval noise estimate". This should be stated from the beginning.

2) Lowermost altitude coverage. In some places it is written that the data start at 40 km but in others, e.g. lines 392-393, (and mainly in the figures) the lowermost heigh is 50 km. I would recommend to state as lower altitude 50 km, particularly in Table 2. The reason is that according to the figure shown in the response (absorption factor tables), there is already a considerable absorption at 40-45 km, between 40 and 60\% absorption. What I am surprised is that even with such absorption the AKs still show a high sensitivity, MR near 0.8.

Other comments:

Abstract. "completely new" -> "new"?

Abstract, l. 9. I would use "feasibility" instead of "performance". The latter implies a quantification (e.g. the performance of an instrument), which has been done only partially.

Abstract, l. 13. Sentence: "We find that IRI appears to have a positive bias of up to 25\% below 75 km, and up to 50\% in some regions above." This is an example of my general comment above (to justify my view). This sentence is a typical example of characterisation of an available database. It may well be that you try to understand and correct this bias in the future so the actual database will not have such a bias. In that case this sentence is meaningless. Or would you process the whole dataset with that known bias?

l. 13 " ... data set about the overall atmospheric distribution of ozone." Do you mean the whole atmosphere? Between 50 and 100 km?

l. 42, equilibrium -> equilibrium,

l. 66, change "novel technique" to "novel treatment" or "novel approach", as written later on. "Technique" embraces the whole inversion.

l. 83 "venerable". This is a question of taste, very personal. I would avoid religious terms in scientific papers. You can use other adjetives as "very productive", useful, fruitful, etc. Also later on you sometimes use "believe" which I would change to "think".

p. 7 Legend of Fig 1. (already mentioned above). "total error". This clearly does not include the systematic error discussed later on (Table 1). Is this total "calibration" error? or total instrument (noise) error? Or a contribution to the total systematic error? It should be clarified.

l. 186-7. "... and it changes with the atmospheric temperature at the tangent point.". In this case Phi should be inside the integral in Eq. 5.

l. 190. suggested writing: " ... the atmospheric layer at the line-of-sight is optically thick."

l. 194. "... is used for temperature and pressure." and I believe also for "O2 vmr".

P. 8 legend of Fig. 2. I suggest to change "error size" to just "error" or "error values" or "error ratio or percentage". One possibility is to multiply it in this figure by 100 and give it in %.

p. 10. Fig. 3. (Already mentioned above). If the absorption is so strong at tangent paths above 50 km (see fig. in the reply to the referees), how can the sensitivity be so large at 40 km?
Minor: label box partially overlap/hide the data.

l. 245. Here it also needs to clarify which is this error. I believe it is the random (instrumental or noise error). Note that these values contrast with the upper limits listed in Table 1. So they need to be clarified.

l. 249. zic-zac -> zig-zag? or zigzag?

Fig. 4. Again clarify which is this error?

Fig. 7 (p. 16). Legend. I would change "naive" to "simple" or similar. "naive" is relative to persons, ideas, not to quantifications. Change it also in the text.

ll. 392-393. " the lowest 10km grids in the retrieval are filtered out to avoid biases due to the possible edge effect." So, at the end, the lowermost retrieved altitude is 50 km? (Change it in Table 2).

l. 397. "... has a precision of around 5-20\% based on the retrieval noise estimate" OK. Now it is clear. This should be stated much earlier in the manuscript.

l. 403. OK. Now it is clear the difference between the precision (random error or noise) and the systematic error sources. Make this clear from the beginning, including the abstract.

p. 19, Table 1. Which the the precision or bias of the pointing?
Change "estimated error sizes" by "estimated errors"
Footnote. a), "... sensitivity of each of the parameter" -> " ... sensitivity of each parameter.

p. 20, Table 2.
"Retrieval uncertainty" -> random error, precision, noise error (use one of these terms).
40-100 km -> 50-100 km

ll. 413-414. I do not understand this sentence. The retrieved quantity is always number density. However, if you measured pressure and temperature simultaneously you can translate it, without lack of accuracy, to VMR, which is a better unit for understanding the chemical and physical processes behind, and do not show the large variation of several orders of magnitude with altitude.

l. 450 KIT-IMK -> KIT-IMK and IAA-CSIC

l. 454. I suggest to add this reference for the MIPAS p-T data used:
Garcia-Comas, M., Funke, B., Gardini, A., López-Puertas, M., Jurado-Navarro, A., von Clarmann, T., Stiller, G. P., Kiefer, M., Boone, C. D., Leblanc, T., Marshall, B. T., Schwartz, M. J. and Sheese, P. E.: MIPAS temperature from the stratosphere to the lower thermosphere: Comparison of vM21 with ACE-FTS, MLS, OSIRIS, SABER, SOFIE and lidar measurements, Atmos. Meas. Tech., 7(11), 3633–3651, 2014.

Fig. 11. Legend, line 2. "(above 40 km)". It seems it is above 50 km (consistent with the whole manuscript).

l. 463. "IRI ozone covers the altitude range from 50 to 100km as ..." This seems to be correct and more likely. Harmonise the rest of the manuscript with this value.

l. 468. "... the background density included in the SMR product". Is the density measured by SMR? Clarify and mention the ultimate source used.

ll. 471-472. As mentioned in the general comment above, this coincidence (lower panel) is fine. However, if one would like to have a solid consistency between the three datasets it should not rely on one single profile from ~millions. That is, as a user I would not trust it unless it is done on a statistically meaningful sample.

l. 474. believe > think

l. 481. hemisphereS

l. 493. "We are going to look at the ..." -> We discuss the ...

l. 495. I think the study would be more solid if based on several months (not just one) but then this is probably "beyond the scope of the paper". This if the reason I argue in my general comment to focus the paper on one aspect and leave the other for other studies.

Fig. 12. Legend. Monthly mean DAYTIME ozone number density ...

Fig. 13, legend, line 2. Again mention 50 km as the uppermost altitude. OK. consistent with everything else. How was performed the merging between the 2 datasets at 50-60 km?

l. 521. "...where the differences are bigger, up to -70\%." This is correct but maybe you should warn that O3 vmr at 80 km is very small, which exploits the relative difference. What is the absolute diff in VMR?

l. 527. ... maybe be THE main reasonS ...

Fig. 14. Left upper panel: The legend box hide some of the data
Legend, line 3. Specify which type of error.

ll. 554-555 "...according to how much the measurement conditions diverge from ..." -> ... according to the divergence from ...

l. 563. "The comparison also demonstrates the advantage of the high
sampling rate of IRI." Probably the major advantage would be to reduce the precision of the mean values when averaging large samples?

l. 567. believe -> think

Hide

ED: Publish subject to minor revisions (review by editor) (16 Sep 2020) by Bernd Funke

AR by Anna Wenzel on behalf of the Authors (02 Oct 2020) Author's response

ED: Publish as is (06 Oct 2020) by Bernd Funke

Short summary

The OSIRIS IR imager, one of the instruments on the Odin satellite, routinely measures the oxygen airglow at 1.27 μm. In this study, we primarily focus on the steps done for retrieving the calibrated IRA band limb radiance, the volume emission rate of O₂(a¹∆_g) and finally the ozone number density. Specifically, we use a novel approach to address the issue of the measurements that were made close to the local sunrise, where the O₂(a¹∆_g) diverges from the equilibrium state.