Retrieval algorithm for aerosol effective height from the Geostationary Environment Monitoring Spectrometer (GEMS)

Park, Sang Seo; Kim, Jhoon; Cho, Yeseul; Lee, Hanlim; Park, Junsung; Lee, Dong-Won; Lee, Won-Jin; Kim, Deok-Rae

doi:https://doi.org/10.5194/amt-18-2241-2025

Articles | Volume 18, issue 10

https://doi.org/10.5194/amt-18-2241-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

Special issue:

GEMS: first year in operation (AMT/ACP inter-journal SI)

https://doi.org/10.5194/amt-18-2241-2025

© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 18, issue 10

Research article

|

26 May 2025

Research article |

| 26 May 2025

Retrieval algorithm for aerosol effective height from the Geostationary Environment Monitoring Spectrometer (GEMS)

Sang Seo Park, Jhoon Kim, Yeseul Cho, Hanlim Lee, Junsung Park, Dong-Won Lee, Won-Jin Lee, and Deok-Rae Kim

Download

Final revised paper (published on 26 May 2025)
Supplement to the final revised paper
Preprint (discussion started on 04 Jul 2023)

Interactive discussion

Status: closed

RC1:
'Comment on amt-2023-136', Anonymous Referee #1, 29 Jul 2023

Please see attachment for detailed comments.

Citation: https://doi.org/10.5194/amt-2023-136-RC1
- AC1: 'Reply on RC1', Sang Seo Park, 16 Sep 2023
  
  Thank you for the reviewer's comments. We attached our reply for the revised manuscript.
  
  Citation: https://doi.org/10.5194/amt-2023-136-AC1
RC2:
'Comment on amt-2023-136', Anonymous Referee #2, 01 Aug 2023

Review of "Retrieval Algorithm for Aerosol Effective Height from the Geostationary Environment Monitoring Spectrometer (GEMS)" by Sang Seo Park et al, 2023.
The paper introduces the newly developed GEMS aerosol effective height product. This is an interesting new product. Aerosol vertical profile and height products are scarse and satellite instruments that monitor aerosol height are needed to constrain models. Global products are few, notably the CALIOP instrument which resolves the aerosool vertical profile, albeit in a small swath, the TROPOMI instrument with global coverage which retrieves an effective height for global average aerosol properties, and the GOME-2 instruments, which retrieves an effective height for dense absorbing aerosol plumes. At present, no aerosol height product from a geostationary satellite is available, and the results from GEMS are very interesting scientifically.
General comments:
The paper presents some technical details of the retrieval algorithm and validation results from a limited but well-chosen intercomparison. However, at present the presentation quality of the paper and the scientific results are not sufficient to warrant publication. My main issue is that the level of detail is poor and seems rather arbitrary. From the manuscript, it cannot be deduced what methods were used, and even a definition and derivation of the AEH, the main product, is missing. On the other hand, some very detailed information of tests are presented, of which the results are not given and referred to in other papers.
The second major issue is that the presentation quality should be increased. At present, the manuscript is poorly structured, with results of tests in the algorithm section, paragraph describing different subjects than indicated, and numerous unjustified qualitative statements, like "good", and especially the overused word "significant". This should be improved by restructering the manuscript and quantifying statements. The use of English should be improved.
The scientific relevance is "significant", so I recommend that the manuscript be published eventually, but only after major revisions have been made. First, I recommend that the manuscript be completely restructured, following a normal setup including Introduction, Method and Data, Results, Conclusions and Discussion section, with all text strictly confined to their proper section, and with the specific comment detailed below included. Second, I recommend a full new review, because the manuscript in its present form cannot be reviewed for content.

Specific comments:
Introduction:

l. 53-58. It is suggested here that UVAI is retrieved for atmospheric correction of gaseous species retrievals. This is not true. The UVAI has been retrieved as a ozone absorption residual, which turned out to be mainly due to absorbing aerosols. As such it has been useful for the detection of absorbing aerosols, but only ever in a qualitative way. These sentences should be changed to eitehr properly describe atmospheric correction or properly describe the function of the UVAI (and scattering AI).
l. 55. cloud signals are not used to derive UVAI. In fact, aerosol signals are not used to derive UVAI as such, but aerosol (absorption) affects the reflectance in the UV measured by a satellite compared to a cloud-free and aerosol-free atmosphere, which can be computed. Clouds affect the reflectance in a different way, which can also be expressed in terms of the UVAI. Please, rephrase and be concise and be specific about the message here.
l 58-60. AOD and clouds are not specifically retrieved from the UV, more often in the visible (aerosols) and the near IR and IR for clouds. What is the purpose of this sentence about UV wavelengths here?
l. 78-80. and because of the generally much smaller signal from aerosols compared to clouds.
2. AEH retrieval algorithm
Main comments:

* Please, introduce the AEH here. Only a reference is given to the paper by Park et al (2016). The reader should have some knowledge of the subject of the current paper, without having to search for other papers.
* A GEMS instrumentation and data section is missing.
* A product description is missing. Please, provide some guidance with the presented flowchart.
l. 122: The AEH is apparently something with (1-1/e)* AOD. Please, give a definiition and/or derivation of the product.
l. 121: "(...) the aerosol vertical layer parameter, which represents the highest altitudes with existing effective aerosol extinction sensitivity." Please, rephrase. This sentence has no meaning. Which "highest altitudes"? What "existing sensitivity"?
l. 125: specify the GDF and the FWHM of the assumed aerosol layer profile.
l. 135.: "Because the spectral coverage is limited to 300-500 nm, .." Does this refer to GEMS? Why is it limited?
l. 136 -140: repetition of information, merge with introduction.
l. 142: ".. considering THE aerosol types .." What aerosol types? This is not clear at this point. Please, provide a methods section, in which the aerosol models are introduced which are used by the algorithm. Merge with the information provided later on aerosol properties/LUT calcalutions.
l 143-144: Repetition of information. Merge with an accuracy assessment section.
l. 151: "long-term GEMS (...) data." What do you mean with long-term? Be specific, if this is relevant at all.
l. 167: "From Nanda (...) retrieval." This is irrelevant information in this section. Please, restructure the manuscript to describe the AEH, ALH and the different tests and analyses seprately.
l. 174-176: Merge this with the actual description of the LUT on page 10.
l. 181: "Observed radiance fitting is affected by noise signals during radiance observation." I do not understand what this sentence means. Please, rephrase.
Table 2: It is unclear what is fitted. Please, describe the test and/or the polynomilas and their meaning

and the test results.
page 9: Somehow during the description of the AEH algorithm we have entered some fitting testing. It is unclear what the reason and the outcome is. Please, provide an overview of this test, if relevant, and quantatative results.
l184-185: "Although the fitting quality was good overall, the setting with the smallest error was used in this study." I surely hope so! Howveer, as mentioned, the test and results are unclear. Pleasew, remove this statement and provide more useful information.
l. 192-197. This information is not about the LUT. Please, put it in a more appropriate section.
l. 201 and 203: Please rephrase the following:

"Thick aerosols"

"the effective scattering layer penetrates more deeply ..."

"dominant aerosols"
l. 206: How significantly? This is useless information, please quantify.
Table 3: Please, provide the real part of the refractive index as well for the various aerosol models, and I suggest to add a separate section on aerosol models used.
l. 216: "...(AER_LH) retrieves vertically localized aerosol layers in free troposphere .." This is not true. Do you mean in cloud free scenes?
l. 218-219: "Spectral fit estimation of reflectance around the O2-A band is based on a neural network for the forward model calculation." This is not true. The spectral fit is the optimal estimation of the retrieval part, mentioned in l. 220. The forward model consists of a simulated spectrum provided by the neural network.
l. 221-223: "During the radiance fitting, AOD is also used as the main fitting parameter, but other aerosol parameters, such as SSA and scattering phase function, are assumed to be fixed values (Nanda et al., 2020)." ALH and AOD are (currently the only) fitted parameters, the aerosol model is assumed (SSA, g and aerosol layer thickness are fixed).

l. 241-242 "(..) layer height parameter is estimated by using the vertical profile of extinction coefficient at 532 nm. ". How is this done? Do you calculate an effective height? How? Or use the top and bottom aerosol layer products? Please, specify.
l. 248. "Based on the retrieval sensitivity of AEH (...)". Which sensitivity? Please, show this in this paper, or repeat the conclusions from another.
l. 255: "Although the AOD and SSA for these plumes are differed significantly, their AEH results were similar. " Improper English, please rephrase. Why would you assume a correlation?

l. 256. " (...) the retrieved AEH values exhibited insignificant diurnal variation (..)"

What is insignificant? Can you show this in e.g. a line plot?

Figure 2: These small plots are not readable. Please, device a better way to show the case, it's diurnal variation and the correlations you try to show. 21 plots just won't do it. The caption is insufficient
l. 266: "siginficant". What does this mean?
l. 267: Please rephrase the following:

"severe aerosol plumes"

"AEH was stably estimated"

and l. 279.

"larger aerosol heights"
l. 280-281: "In an ideal case, the AEH from GEMS was overestimated by around 0.5 km relative to the ALH from TROPOMI, (..). Why in an ideal case? What is overestimated? Do you mean the altitude was higher?

l. 278-284. Place in a discussion section and elaborate.
l. 283-284. "(...) the GEMS AEH retrievals for the dust transport case study were good." Why? What is good? Please, quantify.
l. 285-286: "Furthermore, the retrieval area covered by GEMS is larger than by TROPOMI, as demonstrated by a comparison of Figures 2 and 3.". The shown area in Figs 2 and 3 are equal, so this is not demonstrated from comparing the figures. Rephrase.
l. 289. What does "greater aerosol height information" mean?
l. 289-291: "Although high height values were retrieved for clear-sky regions, in particular low latitude ocean regions, the AEH from GEMS was successfully retrieved over the area of interest for the case study.". Even after three tries, I fail to understand this sentence. Please, rephrase.
l. 311-321. This paragraph discusses Fig 4b., but it starts with the sentence on Fig 4a. Elaborate on Fig 4a, because it is not further described. Then describe Fig 4b in another paragraph.
l. 317-319. If surface reflectance is retrieved, why use a LER? Why mention this?
l. 319-321. What is the impact? If you mention this, then you should either explain why a LER is used and quantify the error/uncertainty, or use the surface reflectance.
l. 329. "To correct the inconsistency of definition, (...)". What correction has been applied?
l. 330. "After correction, ". What correction?
Figure 5. Same as Fig2: decrease the number of panels and device a cleverer way of presenting the results, 21 panels are not readable.
l. 351. "(...) the GEMS (..) AEH can be used in several application studies.". What studies? How? Please, remove a vague commentary and be explicit in what you mean.
l. 378-397. should go into a discussion section and be removed from the results section.
l. 381-382. "Although L2AERAOD accurately retrieved the optical and physical properties of aerosols (...)". Improper English, please rephrase.
l. 385. "has significant root mean square errors ". What does that mean? Please, rephrase.
l. 388. Quantify "significantly".
Figure 9. Caption is not complete. Specify the boxes, line and bars.
l. 439. "TROPOMI is limited to retrieving the ALH over scattering-dominant aerosols". This is difficult to understand and I think it is untrue (although I'm not sure what it mneans). The ALH v1 that was used in Nanda et al (2020) was retrieved only in cloud-free scenes with UVAI>1, so for absorbing aerosol only.
l. 440-441. "Griffin et al. (2020) reported that the small absorbing AI pixels are identified with small QA values in the offline product of ALH.". Please rephrase. There are no small pixels. You probably mean something like: "pixels with low UVAI values are identified by a QA value lower than .. (O.5?)". But this is not true, because low UVAI values are identified by low UVAi values. Please, explain what you actually mean.
l. 452-454. If this is true, then the reason for passive sensors to have a smaller variability needs to be explained, because I don't see it.
l. 557. "driving large deviations in some pixels". What does that mean?

Minor comments:
l. 74. Oxygen-related absorption bands-> oxygen absoprtion bands
l. 78. add a reference to Xi et al, 2021, who combined O2-A and O2-B for the retrieval of ALH from TROPOMI.
l. 79-80. as the optical properties of aerosols in the atmosphere differ among aerosol types -> because of varying aerosol properties.
l. 87-90. ... and implemented the algorithm operationally.
l. 95-96. What is 'the standard product of aerosol'? Please, rephrase.
l. 153: "To consider the various retrieval products, the L2SFC retrieves .." Please, rephrase. L2FSC does not do something in consideration of products.
l. 158-159: "However, L2SFC accurately retrieved surface optical properties with high spatial resolution." Please, rephrase this vague statement. What is "high"? What is "accurate"?
l. 172-174. Remove this unnecessary sentence.
l. 213. The TROPOMI is spectrometer -> TROPOMI is a spectrometer
l. 215. crossing to equator -> crossing the equator
l. 230. lidar sensor -> lidar
l. 234-236. "This sensor is Sun synchronous orbit constellated to the A-train, and also cross to equator at 13:30 local time by ascending node." Improper English. Please, rephrase.
l. 240. "Because the aerosol profile product exists the vertical distribution of aerosol extinction coefficient, ... " Improper English. Please, rephrase.
Figure 3: What is the unit of the colorbar?
l. 276-277. Place this under the TROPOMI data section.
l. 360. ". most" -> ", most"
l. 362. "the AI value for UV" -> UVAI
l. 437. " scattering dominant aerosols". There is not such a thing as a dominant aerosol. The authors define "Scattering-dominant" as non-absorbing, but I suggest not to introduce these kind a strange wording. Just stick to absorbing or non-absorbing and better yet give the SSA as done here and leave it at that.

References
Xi Chen, et al., First retrieval of absorbing aerosol height over dark target using TROPOMI Oxygen B band: algorithm development and application for surface particulate matter estimates , Remote Sens. Env., doi: 10.1016/j.rse.2021.112674 2021.

Citation: https://doi.org/10.5194/amt-2023-136-RC2
- AC2: 'Reply on RC2', Sang Seo Park, 16 Sep 2023
  
  Thank you for the reviewer's comments. We attached our reply to the revised manuscript.
  
  Citation: https://doi.org/10.5194/amt-2023-136-AC2

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

AR by Sang Seo Park on behalf of the Authors (16 Sep 2023) Author's response Author's tracked changes

EF by Sarah Buchmann (18 Sep 2023) Manuscript

ED: Referee Nomination & Report Request started (02 Oct 2023) by Ben Veihelmann

RR by Anonymous Referee #2 (18 Oct 2023)

Suggestions for revision or reasons for rejection

The manuscript has been revised considerably, which is appreciated. The results from GEMS are scientifically very interesting, because they add valuable temporal information of aerosol vertical distrubution which is not currently available from other instruments, and therefore the results are important. However, the manuscript should be edited for English, because many paragraphs are still difficult to read, even after the suggestions of the two reviewers in the first round. Also, suggestions have not been followed throughout, e.g. it was made clear that an satellite passes THE equator, not TO the equator. Yet, this was corrected for S5P only, not for CALIPSO. Similarly, suggestions to remove and quantify vague statements like "significantly" have been made at some places but many remain. The figures with 21 panels are unchanged, etc. I feel some more effort is needed to improve the readability.

The main criticism is that it is still not clear what physical quantities are compared. AEH is defined for GEMS, but the manuscript states that "representative" AEH and ALH parameters from CALIOP were estimated, it is not mentioned how. ALH can be defined as SUM(ext * z) / SUM(z) or aerosol layer (bottom+top)/2, both available from CALIOP. The first has a direct relationship with AEH, but for the latter the relationship can be quite obscure for multiple layers and distinct non-Gaussian profiles. TROPOMI AER_LH is defined in similar terms, aerosol layer (top+bottom)/2, and the authors state that the comparison between AEH and ALH was "corrected" to minimum the bias. The correction is not given and there is not a bias but a comparison of different quantities. The ALH should be compared to ALH and AEH to AEh for all comparisons. It should be clear which definitions are used at all times.

More suggestions are given below:

L55 The manuscript still states that that cloud signals are detected using the UVAI. This is not true and certainly not in one of the mentioned references. In the response to reviewer 2 it is first stated that this would be changed, but it is mentioned still.

Figure 2. A description of the flowchart was requested. However, the manuscript was extended with the paraghraph 158-179 according to the response, but there is no mention of Figure 2 anywhere in the manuscript. The flowchart is still unreadable. Why?

Table 2 and the description of the fitting are still unclear for me. It may be common for DOAS retrievals, but I can't judge whether the description makes any sense.

281-283. "Although the AOD and SSA for these plumes are differed significantly, the retrieved AEH results from these different plumes show similar ranges."
(1)The improper English is in this part: "are differed significantly", please rephrase, and remove significantly also in this case
(2) I still don't understand why a correlation is assumed here.

l 90. "other satellite for passive sensors" -> Do you mean: "other passive satellite sensors"?

l93-94 "It means that the number of information for aerosol vertical distribution have a limitation." ->
"It means that the number of information for aerosol vertical distribution has a limitation for OMI". What about other instruments?

l 106 remove either "in addition" or "additionally"

L 126. "AEH is a layer height parameter that considers the penetration of photons into the aerosol layer."
I would think the AEH considers the penetration of photons into the atmosphere. But to be honest, the definition of AEH is strange to me and this statement may actually be right. However, it must be understood that the ALH considers the penetration height of photons in the atmosphere, so then the relationship between AEH and ALH becomes obscure for any vertical aerosol distribution that is non-Gaussian and not integrated from the surface to the top of the atmosphere.

L150 Remove "For this reason"

L. 166 "Main error sources for AEH retrieval can be obtained from the L2AERAOD results." This would be very useful, but I don't think you can get the error sources for the AEH from the AOD results. Do the authors mean that they can calculate AEH errors using the AOD product as input?

L 224 - 226. Aerosols are still called "dominant" here, please rephrase.As a minimum, change "absorbing dominant aerosol" to 'absorbing-dominant aerosol", but I would prefer to simply write "absorbing aerosols" since you also simple refer to scattering aerosols as "non-absorbing aerosols".
Please, also rephrase "is more efficiently absorbed the radiance."

L 253: Resolution of TROPOMI AER_LH is 3.5x5.5 km^2

L 261 It was suggested to change "crossing to equator" into "crossing the equator". This also holds for the second time this term was used improperly. Please, rephrase this wrong sentence: "crosses to the equator".

L 263-264: Please rephrase: "..the pixel data can retrieve with extremely high horizontal and vertical resolutions .."

Figures 3 and 6 still cannot be read. As mentioned before, a figure with 21 panels will just not do. Why not add them in the supplement or make a movie out of them in the supplement, and show here two times that align with the TROPOMI and CALIOP overpasses?

L 310 "the GEMS AEH were (...) mostly shown the the expected uncertainty of 1 km". How was this shown then?

L. 316-318. "Although high AEH values were retrieved for low AOD regions over low latitude ocean surface, the AEH from GEMS was successfully retrieved over the area of interest for the case study." Please, rephrase this sentence, it is completely unclear.

L. 361. " To correct the inconsistency of definition between ALH and AEH, the difference between two retrieval results decreased to 0.5 km bias. " What is the correction? I think apples and pears are compared here. The proper comparison is AEH and AEH or ALH and ALH. Please, compare the same physical quantities and change your analysis accordingly.

L. 411-413 "Go et al. (2020) noted that the UV aerosol retrieval algorithm, which is the basic method to the L2AERAOD algorithm, has significant discrepancies for both AOD and SSA compared to ground-based data."
This sentence has no meaning ("significant" again) to a reader who cannot cite Go et al (2020) by heart. Please, state what the UVAI algorithm has to say about discrepancies between AOD and SSA from GEMS and ground-based measurement and why this is relevant here.

L 421-428 is actaully interesting and useful information!

L. 484-485. ". In addition, the difference in the definition of ALH from TROPOMI and AEH from GEMS impacted the comparison. " So, why did you not convert AEH to ALH?

L. "For dust plumes over East Asia, AEH indicated significant aerosol vertical information and insignificant diurnal variation in regions with severe dust plumes. "
Significant and insignificant are complete meaningless. And you compare technical information ("good enough" height retrieval) with geophysical information (static dust plume). Please, rephrase.

L 556-558 "In addition, AEH from GEMS was overestimated compared to the TROPOMI ALH results due to different definitions of ALH from TROPOMI and AEH from GEMS." Again, simply compare the same physical parameters and remove this statement.

Hide

RR by Anonymous Referee #1 (24 Oct 2023)

RR by Anonymous Referee #3 (01 Nov 2023)

RR by Anonymous Referee #4 (14 Nov 2023)

ED: Reconsider after major revisions (15 Nov 2023) by Ben Veihelmann

AR by Sang Seo Park on behalf of the Authors (31 Jan 2024) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (28 Feb 2024) by Ben Veihelmann

RR by Anonymous Referee #3 (15 Mar 2024)

RR by Anonymous Referee #4 (18 Mar 2024)

Suggestions for revision or reasons for rejection

I applaud the authors for making many revisions to the manuscript. I still have many concerns but will only list a few here.
(1) Thank you for showing the look-up-table information. Though the three panels show the LUTs for different aerosol types, I do not see any sensitivity; if any, it is really a poor sensitivity.
(2) Abstract says "The AEH difference depends on aerosol optical properties and surface albedo. Compared to the aerosol layer height obtained from the tropospheric monitoring instrument (TROPOMI), differences of 1.50± 1.08 km, 1.59 ± 1.22 km, and 1.71 ± 1.24 km were obtained for pixels with single scattering albedo (SSA) < 0.90, 0.90 < SSA < 0.95, and SSA > 0.95, respectively, with significant dependence on aerosol type." Again, given the uncertainties, are 1.5 km, 1.59 km, and 1.71 km that different and statistically significant?
(3) Scatter plots comparing GEMS and CALIOP show no skill, whatsoever, for the retrievals but r values are still very high. Histograms of GEMS - CALIOP show a broad spread (~2 km to 4 km bias) but the shape is gaussian and hence close to zero mean bias. This does not give me any confidence in the retrievals. The precision is not good based on those figures.
(4) Word "significant" is used in many places with no context.
(5) Despite other reviewers asking for potential way of improving the algorithm, no mention of it except to say combine this technique with Oxygen A/B bands
(6) etc.
In conclusion, I find that the authors have missed an opportunity to write a good paper. GEMS has been flying for three years now. No excuse for not showing at least one full year of data analysis to get good sampling of AEH retrievals for different aerosol types, and most important of all show what sorts of diurnal variations can be picked up and what issues exist with solar/satellite geometries. I am tempted to say "reject" but the authors have put in a good faith effort to revise. So please have someone rewrite to fix all editorial mistakes prior to publishing if that is the decision the journal makes.

Hide

ED: Reconsider after major revisions (29 Apr 2024) by Ben Veihelmann

According to the reviewers, substantial issues remain unresolved after the last major revision. The issues raised concern both the AEH retrieval scheme, the content of the analyses presented, as well as the quality of the presentation. Nevertheless, the reviewers do highlight the relevance of the topic for the AMT community. Therefore, I would like to encourage the authors to make another major revision to the manuscript addressing the issues identified.

Please see below my comments on the manuscript:

1. The AEH retrieval strategy needs to be presented more clearly. The retrieval of AEH is challenging since the information content of the measurement is limited while many other parameters (AOD, aerosol characteristics, scene parameters) influence the measurement. It needs to be clarified, which prior information is used for the retrieval, and where this prior information on these other parameters is taken from.

2. The presentation of the AEH retrieval scheme needs to be supported by sensitivity analyses. The scheme exploits the O2-O2 Slant Column Density (SCD) retrieved form the O2-O2 absorption signature near 470 nm. The sensitivities of the AEH to the O2-O2 SCD and to the other parameters that influence the measurement, need to be presented.

3. The expected AEH retrieval uncertainty needs to be quantified, in particular in view of the uncertainty in the available a-priori information on these other parameters that influence the measurement. I recommend trying to identify regimes in which the retrieval is expected to provide useful information and regimes in which this is not expected. Such regimes could be defined in terms of AOD, surface albedo, even layer height itself, etc.

4. The poor correlation with reference data from CALIOP and TROPOMI (Figures 6 and 8) raises questions on the performance of the AEH algorithm. This needs to be explained. Sub-setting of the data should be considered such that the comparisons are restricted to regimes in which both the AEH and the correlative observation data set are expected to be meaningful.

Hide

AR by Sang Seo Park on behalf of the Authors (21 Jun 2024) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (07 Aug 2024) by Ben Veihelmann

RR by Anonymous Referee #5 (01 Nov 2024)

Suggestions for revision or reasons for rejection

General

Substantial issues remain unresolved even after the last major revision. These issues concern the clarity of the description of the AEH retrieval scheme, the content of the analyses presented, as well as the quality of the presentation of the results. A major revision is necessary.

1. The manuscript shows substantial shortcomings in language and clarity. A very large number of sentences are unclear or inaccurate (see detailed comments).

2. The poor correlation with reference data from CALIOP and TROPOMI (Figures 9a and 12 a) and the wide spread of differences (+/- 3 km, Fig 13) is not explained in a satisfactory way. The amplitude of these differences and the absence of correlation across the (~3 km) range of AEH values shown in the scatterplots go beyond what can be explained by the AEH uncertainty estimate of 0.74 – 1.28 km. More work is necessary to identify the root causes for these discrepancies. It seems that improvements to the algorithm are necessary to make the product useful. This should be more clearly stated in the conclusion.

3. It is stated that AEH retrieval algorithm takes surface reflectance data from spectral surface reflectance product L2SFC, and AOD data from the L2AERAOD product generated using a different surface reflectance (Line 154 ff). Please clarify whether potential inconsistencies in surface reflectance are problematic for the AEH retrieval. It is recommended to test the AEH algorithm with different input ensuring a consistent treatment of surface reflectance.

Detailed Comments

Typo line 133: onboarding ==> on board

Rephrase line 133-134: “The GEMS is onboarding the Geostationary Korea multipurpose satellite 2B (GK2B) as orbiting at 128.2⁰E, …” ==> “The GEMS instrument on board the Geostationary Korea multipurpose satellite 2B (GK2B) located at 128.2⁰E, …”

Use a dash instead of a tilde for ranges throughout the manuscript

The term “standard spatial resolution” in line 137 is not clear. I guess this is the resolution at a reference location. Rephrase.

Line 139: Fix the sentence “The spectral resolution and sampling are respectively 0.6 nm with full-width and half-maximum (FWHM) and 0.2 nm with spectral range of 300~500 nm.” I suggest something like “The spectral range … is covered with a spectral resolution (as defined by the full-width and half-maximum (FWHM) of the spectral response function) of 0.6 nm and a spectral sampling distance of 0.2 nm”.

Line 142: unclear formulation “with considering the aerosol types”. Are AOD and SSA provided for a set of considered aerosol types?

Line 145: unclear formulation “classification methods based on the UV and visible AIs”. I suggest “AI based classification methods exploiting measurements in the UV and visible”

Line 148: unclear formulation: “Overall, the error for AEH is ranged from 739~1276 m under the AOD error of 0.2, particle size error of 20%, SSA error of 10%, and surface albedo error of 0.02 (Park et al., 2016).” I suggest “Overall, uncertainties of 0.2 in AOD, of 20% in particle size, of 10% in SSA, and 0.02 in surface albedo cause AEH errors in the range 739 – 1276 m in the (Park et al., 2016).”

Line 150: unclear formulation: “The main variables causing errors for AEH retrieval can be obtained from the L2AERAOD results.” Not understood. Remove?

Line 150: unclear formulation: “Therefore, the L2AERAOD results for AOD at 550 nm and SSA at 443 nm were adopted as input data for aerosol properties” The logic link implied by “Therefore” is not understood. Please state that AOD at 550 nm and SSA at 443 nm are used as input to AEH retrieval. Please justify why these two wavelengths are selected, specifically why parameters at different wavelength are used.

Line 164 - 169: The discussion of the input surface data is unclear. Please clarify the meaning of “use the minimum reflectance”. What is BSR? Is the Black Sky Albedo given in the L2SFC product? Is this parameter taken as the Lambertian equivalent reflectance in the AEH algorithm?
Line 164 – 169: garbled sentence. I suggest something like “The S5P satellite flies on a polar orbit and crosses the equator at 13:30 (local time ascending node)”.

Line 176: the “product … provides”

Line 177: “in THE free”

Line 179 – 181: garbled sentence. I suggest something like “The spectral fit employs a fast forward model based on a neural network.”

Line 182: “IS used”

Line 179 – 181: fix the garbled sentence to something like

Line 181 – 186: garbled sentence. I suggest something like “After cloud masking, an optimal estimation method is used to retrieve the ALH and AOD. Other aerosol parameters…”

Line 189: “The main purpose ...”

Line 191: Add “for the layer height and for the associated pressure respectively” when reporting 0.5 km and 50 hPa and 1 km or 100 hPa.

Line 202 introduce acronym

Line 206 – 207: garbled sentence. I suggest something like “The CALIPSO satellite flies on a ascending node orbit in the A-train constellation.”

Line 213: whitespace between number and unit: 90m ==> 90 m

Line 222 typo: CLIOP

Line 224: “Furthermore, the AEH adopted the same method as CALIOP for estimating the ALH, with only changes in the height definition.” Is unclear. Please rephrase.

Line 229 ”is THE observation area”

Line 231-232 “From Park et al. (2020), the spatial scales for AOD validation are 30~40 km.” Is unclear. Please rephrase.

Line 232: A part of the sentence is missing. Is it “To ensure that the number of retrieved observation pixels THAT CAN BE ANALYSED IN THE PRESENT STUDY IS SUFFICNETLY HIGH, …”?

Line 233: Please specify to which level the limits were relaxed. Make at least a qualitative statement on the impact of this relaxation on the robustness of the present study.

Line 251: garbled sentence. I suggest something like “For this reason, cloud contamination remains a critical source of uncertainty in the retrieved ALH”

Line 251: I suggest fixing formulation as “the UVAI is used to detect the presence of elevated absorbing aerosol.”

Line 254-256: Sentence makes no sense. Rephrase / remove.

Line 256: unclear formulation. I suggest ”The pixels for which with an AOD at 443 nm taken from the GEMS aerosol product exceed 0.3 are selected for AEH processing. The UVAI is not used for this selection.”

Line 262: definition is not cleanly formulated. I suggest ”…into the aerosol layer. It is defined such that the integral of the vertical aerosol extinction profile from the surface to the AEH is equal to (1-e-1) times the AOD, as defined by Park et al. (2016).

Line 270: Please clarify in which sense the GDF a “modified” Gaussian? The AEC profile shown in
Figure 1 looks pretty much like a standard Gaussian without modification.

Line 272 “..is SET to 1 km.” or “…is ASSUMED to be 1 km.”

Line 272 “A schematic ..”

Line 275 – 277 Definition of ALH is not clean. I suggest “the ALH is defined such that the integral of the vertical aerosol extinction profile from the surface to the ALH is equal to 0.5 times the AOD. I assume this is the definition used by Nanda et al. (2016); if confirmed please add this statement.

Line 277: “is same to the” ==> “equals the”

Line 288: “for the vertical profile condition as shown” ==> “of the profile in the conditions shown”
Figure 1: I suggest adding the ALH definition to the figure linking it to the AEC profile. Please introduce acronym AEC.

Figure 2: Please distinguish on-line processing steps (from L1C to AEH) from off-line processing steps (computation of Look-up-Tables). I recommend splitting the chart in two. There are 2 LUTs: LUT1 SCD = F(alpha, tau, z,…) and LUT2 SCD = F(z). I assume that LUT2 is extracted from LUT1. Please clarify. Is this extraction done on-line during L1C to AEH processing?

Figure 2: The shapes of the flow chart objects are not used in a logic fashion. Please use shapes distinguishing data, data bases, and processes.

Figure 2: Please use arrows in a logic way identifying data that enter processes: Atmospheric profiles must enter the RTM calculation process. The slit function must enter a process (convolution). Cross sections must enter DOAS analyses.

Figure 2: LUTs, AOD and SSA from the GEMS product, and surface reflectance from the L2SFC product must be used in the on-line processing from L1C to AEH.

Line 285-290 : LUT generation description is unclear. I suggest simplifying to “The AEH algorithm for GEMS employs a look-up table (LUT) that contains O2-O2 SCD values for many scenarios with a variety of … ”

Line 285-290 : LUT generation: I miss the AEH in the list of dimensions of the LUTs. (AEH is included in this list according to Table 3.)

Line 291: simplify to “the radiance is simulated monochromatically and then convolved with the GEMS instrument spectral response function”.

Line 291: statement is in conflict with the logic of the chart in Fig 2. The DOAS processing provides a O2-O2 SCD, not an AEH. The latter is obtained from the O2-O2 SCD using a LUT. Please clarify, rewrite.

Line 294 ff: the description of DOAS is confusing. Please simplify to the refer to Platt (1994).

Line 302 ff: The justification of the choice to exploit O2-O2 signatures is hard to understand. I suggest simplifying to “The AEH is retrieved exploiting the O2-O2 absorption signature. The O2 A band signature exploited in ALH algorithms (references) is not covered by GEMS.”

Line 325: Conclusion on fit parameters is confusing. I suggest simplifying to “Smallest AEH fitting errors are obtained by a DOAS fit with a 2nd order polynomial and no offset. These settings are used in the GEMS AEH algorithm.”

Line 330: Sentence does not swing. I suggest “Park et al () have shown that the largest contributor to the AEH uncertainty is associated with the uncertainty in SSA.”

Line 332: “uncertainty by the aerosol optical properties” ==> “uncertainty due to error in the aerosol optical properties”

Line 339: aerosol type is DEFINED by the radiative absorptivity and PARTICLE size,

Line 340 “based on the method from Lee et al. (2010)” not understood. Are aerosol types defined following Lee et al. (2010)?

Line 341: Absorbing and non-absorbing particles are considered for the fine mode of the size distribution.

Line 344: remove “Otherwise”

Line 345: “assumed as the fixed value” ==> “assumed to be spectrally constant”

Line 345 “Although the center of O2-O2 absorption” not understood. Is this about spectral calibration? Are imperfections of the latter ignored?

Line 348: “we performed the slit response function of GEMS and sampling specification” ==> “the convolution with the GEMS slit response function was applied and the spectra were sampled on the GEMS L1C spectral grid.”

Line 350: remove duplication “For O2-O2 absorption, the absorption cross section used for the radiative transfer model calculation is considered the temperature dependent absorption cross section” ==> “Radiative transfer calculations were performed accounting for the temperature dependence of absorption cross sections”.

Figure 3 caption should be “O2-O2 SCD dependence as a function of AOD and AEH for (a)…”

Line 360: please change “Figure 3 shows … “ accordingly.

Line 368 “Although .. uncertainties ” not understood.

Line 369 – 373 confusing. Rewrite or remove.

Line 384: “in AOD < 0.4” ==> “in cases with AOD < 0.4”

Figure 4 caption “uncertainty by the AOD error of 0.2 according to (a)…” ==> “uncertainty due to an AOD error of 0.2 for cases with (a) …”

Figure 5 caption same as Figure 4 caption

Line 397: Confusing sentence. I guess what is meant is “On average, the associated AEH bias takes values up to 0.5 km”

Line 399 ff “Based on the changes … for AEH retrieval” is confusing. Rewrite or remove.

Line 403: “the L2SFC product is used in operation” ==> “the L2SFC product is produced operationally”?

Line 404: explain why a climatological minimum LER is used instead of the data from the operational L2SFC product.

Line 406 “The data quality” not understood.

Line 412: “From Park et al.,” ==> “Pixels with low AOD values… due to … (see also Park..) ”

Line 415 – 417 ”In addition, the AE” not understood. Is the quality flag from the AOD product copied into the AEH product?

Line 417 what is meant with “significant averaging kernel for optimal estimation for aerosol property retrievals”? significant sensitivity of AEH to AOD?

Line 418 : “During this case study, a dust plume …” ==> “A case was analysed in which a dust plume …”

423 “shows similar pattern ranging” ==> “shows similar patterns and takes values ranging”

424 “between 1.0 to 2.0 km” ==> “between 1.0 and 2.0 km”

Section Case Studies: the repetitive specification “at 443” for AOD and SSA should be avoided. I recommend clarifying upfront that all SSA and AOD values discussed in this section are reported at this reference wavelength.

430 “From Rana et al. ” ==> “According to Rana et al. ”

431 “clusterS”

433 “was a result that actually exists, and it was not a result with high uncertainty due to…” ==> “in considered physical rather than an artifact due to …”

444 “were 0.5~1.5 km” ==> “range between 0.5 - 1.5 km”

444 – 446 Confusing sentence “Given the difference in definition for the aerosol height parameters between ALH and AEH, relatively high height values were retrieved from GEMS compared to TROPOMI” ==> “AEH values retrieved from GEMS were higher that the ALH values retrieved from TROPOMI. The opposite was expected in view of the definitions of the two parameters”.

446-449 remove

It is stated that “A dust plume was transported from China to South Korea during this period” (Line 441) and that this “dust transport case was successfully retrieved” (line 451). Please clarify whether independent data are available that confirm that this dust transport event was real. Only then one can conclude that the dust transport case was successfully retrieved; otherwise, one can only conclude that the AEH signature looks physical.

510 “is strong surface type dependence” ==> “exhibits a strong surface type dependence”

514 A correlation coefficient of 0.993 between GEMS AEH and TROPOMI ALH is reported, which is much higher that what one would expected from the plotted data. Please check/confirm. If confirmed please explain.

525 Unclear sentence “sensitivity of O2-O2 SCD was weak in cases of high AEH because of the vertical distribution of air molecules”. Which sensitivity of O2-O2 SCD is meant? Why past tense? I guess what you want to say is that AEH is weakly sensitive to the true layer height in high layer cases. This is a general feature of the algorithm and is related to the fact that the O2-O2 abundance scales with the square of the pressure.

666 Rephrase “…, the number of pixels observed at 03:00 UTC was insufficient for the identification of diurnal variation.” Observations at one fixed time never allow studying diurnal variation.

Hide

ED: Reconsider after major revisions (04 Nov 2024) by Ben Veihelmann

AR by Sang Seo Park on behalf of the Authors (22 Jan 2025) Author's response Author's tracked changes Manuscript

ED: Publish subject to minor revisions (review by editor) (23 Jan 2025) by Ben Veihelmann

AR by Sang Seo Park on behalf of the Authors (26 Jan 2025) Author's response Author's tracked changes Manuscript

ED: Publish as is (27 Jan 2025) by Ben Veihelmann

AR by Sang Seo Park on behalf of the Authors (07 Feb 2025) Manuscript

Short summary

An operational aerosol effective height (AEH) retrieval algorithm for the Geostationary Environment Monitoring Spectrometer (GEMS) was developed that solely uses the O²–O² absorption band considering aerosol and surface properties. AEH retrievals are only performed when aerosol optical depth is larger than 0.3. The retrieval results show significant estimations by comparing the aerosol height from the Cloud–Aerosol Lidar with Orthogonal Polarization and the Tropospheric Monitoring Instrument.