Application of the Complete Data Fusion algorithm to the ozone profiles measured by geostationary and low-Earth-orbit  satellites:  a feasibility study

Zoppetti, Nicola; Ceccherini, Simone; Carli, Bruno; Del Bianco, Samuele; Gai, Marco; Tirelli, Cecilia; Barbara, Flavio; Dragani, Rossana; Arola, Antti; Kujanpää, Jukka; van Peet, Jacob C. A.; van der A, Ronald; Cortesi, Ugo

doi:https://doi.org/10.5194/amt-14-2041-2021

Articles | Volume 14, issue 3

https://doi.org/10.5194/amt-14-2041-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

https://doi.org/10.5194/amt-14-2041-2021

© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

Articles | Volume 14, issue 3

Research article

|

12 Mar 2021

Research article |

| 12 Mar 2021

Application of the Complete Data Fusion algorithm to the ozone profiles measured by geostationary and low-Earth-orbit satellites: a feasibility study

Nicola Zoppetti, Simone Ceccherini, Bruno Carli, Samuele Del Bianco, Marco Gai, Cecilia Tirelli, Flavio Barbara, Rossana Dragani, Antti Arola, Jukka Kujanpää, Jacob C. A. van Peet, Ronald van der A, and Ugo Cortesi

Download

Final revised paper (published on 12 Mar 2021)
Supplement to the final revised paper
Preprint (discussion started on 18 Dec 2019)
Supplement to the preprint

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC1: 'referee comment', Anonymous Referee #1, 14 Jan 2020
- AC1: 'reply to reviewer comments', Nicola Zoppetti, 16 Mar 2020
RC2: 'Review of Zoppetti et al paper submitted to AMTD', Anonymous Referee #2, 04 May 2020
- AC2: 'reply to referee #2 comments', Nicola Zoppetti, 26 Jun 2020

Peer-review completion

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

AR by Nicola Zoppetti on behalf of the Authors (26 Jun 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (03 Jul 2020) by Thomas von Clarmann (deceased)

RR by Anonymous Referee #1 (07 Jul 2020)

Suggestions for revision or reasons for rejection

The authors have done a great effort to answer all reviewer comments appropriately. This manuscript has significantly improved as a result. The writing style however can still be considered somewhat sloppy at some instances. Technical corrections are therefore suggested below, but the authors are encouraged to have the final version of their manuscript revised by a native English speaker.
Technical corrections:
Lines 30-33: Rewrite for clarity. Statistical analysis of what? Start a new sentence afterwards: “The grid box size was also...”
Line 35: “coordinated by the European Commission”
Line 48: “contributing missions” is unclear here. What is contributing to what? Other missions contributing to the Sentinels would be misleading...
Line 49: “encompass from” is a wrong wording.
Line 53: “The three approaches differ…”
Line 60: Two times particular(ly).
Line 65: Mismatch between single and plural.
Line 65: “which collects all the available information content” sounds vague. Something like “which optimizes the DFS of the fused product” and including a reference to one of the technical papers?
Line 70: “the algorithm” – which one is intended here; or should this be plural, referring to all previously mentioned?
Line 77: “and discuss the use of the profiles average as fusion technique” - This addition is misleading, as if averaging is the main fusion technique. Suggestion: "and discuss the differences between CDF and mere averaging"
Lines 89-91: Duplication of following?
Line 149: “specifications used for the simulation”
Line 159: “Eqs.”
Line 163: “Concerning the profile and the error the CDF…”
Lines 169-171: Could you briefly elaborate on how this 5% choice is made based on grid cell size? How should other users of the CDF method make such decisions for different grids? Could you provide a few references?
Eq. (8): Remove “= …”
Lines 208-209: “used in this article”
Line 231: “allows seeing”
Line 249: Remove “as suggested by one of the reviewers”
Line 254: Replace “contribute of the measures” by something like “contribution of the (simulated) measurements”
Lines 359-360: Rewrite “even if further analysis are needed especially for what concern the coincidence error to be applied fusing data” for clarity.

Hide

RR by Anonymous Referee #2 (27 Jul 2020)

Suggestions for revision or reasons for rejection

The manuscript by Zoppetti et al. has improved in clarity for some of the aspects remarked by the reviewers. However, major revisions are still needed, and the manuscript cannot be published in the current state. My major concern is that there are mistakes in the characteristics of the instruments onboard MTG-S and EPS-SG together with the instruments Sentinel 4 and 5, whose measurements are simulated for demonstrating the performance of the Fusion algorithm. These instrumental characteristics are greatly important in this work. They must be corrected and all the results using these simulated measurements (Figures 3, 4, 6 and 7) should be done again with the correct configuration. These major errors are the following:

1) IRS onboard MTG-S (called in the paper S4-TIR): this sensor is NOT IASI-NG (nor IRS2b) which is described by Crevoisier et al. (2014). Crevoisier et al. (2014) do NOT describe the instrument that will be onboard the geostationary satellite MTG-S together with UVN/Sentinel 4, but they only describe IASI-NG which will be onboard the low orbit EPS-SG satellite together with Sentinel 5/UVNS. IASI-NG is supposed to be S5-TIR in the paper. EUMETSAT describes IRS onboard MTG-S here: https://www.eumetsat.int/website/home/Satellites/FutureSatellites/MeteosatThirdGeneration/MTGDesign/index.html
“The Infrared Sounder (IRS) on MTG-S … with a hyperspectral resolution of 0.625 cm-1 wave-number...”
This means that IRS onboard MTG-S (in your paper S4-TIR) will have an even coarser spectral resolution than IASI (0.5 cm-1) and much coarser resolution than IASI-NG (0.25 cm-1). The SNR for IRS onboard MTG-S will be much lower than that for IASI-NG and similar to IASI. IASI, with a finer spectral resolution than IRS/MTG-S, enables the retrieval of ozone with 3,5 DOFs at most. Therefore, it is not possible to obtain 5 degrees of freedom for the retrieval of the ozone total column derived from IRS onboard MTG-S, as illustrated in Figure 3 of the paper.
In consequence, all simulations and retrievals using S4-TIR must be done again with the correct instrumental characteristics of this instrument. Figures 3, 4, 6 and 7 should be revised with correct simulations of IRS/MTG-S measurements.
2) IASI-NG (called in the paper S5-TIR): the spatial sampling for IASI-NG (called in your paper S5-TIR) is not correct. The spatial sampling for IASI-NG is the following (the same as for IASI): circular pixels of 12 km of diameters whose centers are spaced by 25 km at nadir. Figure 1 of the paper shows 5 pixels within 0.5 ° of latitude and that is not possible. One may have at best 3 pixels and typically 2 pixels within 0.5 ° (which is roughly 55 km) in both directions (along and across the satellite track). Therefore, it is not possible to have so many IASI-NG pixels within a box of 0.5° x 0.625°, but half of them. Therefore, the number of measurements used for the fusion for the given box is not correct. The spacing of the IASI-NG pixels must be corrected and the results should be revised.
3) UVNS (called in the paper S5-UV): The horizontal sampling of UVNS pixels is not correct either. As announced by ESA https://earth.esa.int/web/eoportal/satellite-missions/c-missions/copernicus-sentinel-5, the horizontal sampling of UVNS is 7 km x 7 km which is much smaller than the box of 0.5 x 0.625° in Fig. 1. Therefore, it is not possible that no pixel is available for UVNS (as indicated in the current version of the manuscript).
For avoiding all these errors, I strongly recommend using the real names of the instruments: UVNS/Sentinel 5, UVN/Sentinel 4, IASI-NG/EPS-SG and IRS/MTG-S, or at least part of their real names. In my opinion, creating new names that are only used in this paper only add confusion. In addition, “Sentinel” only refers to the UV-VIS instruments and not the TIR. Of course, the spectral resolution and horizontal sampling for each of them must be corrected.

Another major point concerns the explanation of the results of Figure 3 of the manuscript. In the revision, a new paragraph explains the gain of sensitivity in the Fused product with respect to the L2 retrievals. However, they do not explain two key points: At the lowest point (1000 hPa) and near the highest (0.5 hPa) the AK for the fused product is much stronger than the ones of the L2 products. At 1000 hPa, a local minimum is seen for the AK of all the L2 products (S4-TIR, S5-TIR and S4-UV), however the AK shows a relative maximum more than twice larger than the best case of the L2 products. At 0.5 hPa, the AK of the Fused product is 5 or 6 times larger than the individual L2 products. Elsewhere, the AK enhances by about 30 or 50 %, but not by a factor 2 or 6. Could you explain in the detail these features? Is this only a numerical feature of this single profile? What are the physical reasons for such an extremely large enhancement of sensitivity? and where would the information come from? Moreover, why vertical resolution in Fig. 3 right panel (FUS product) is the finest at 1000 hPa (above it is coarser)? This is never the case for any L2 product of ozone derived by optimal estimation or Tikhonov-Philips regularization. I cannot understand such a performance of the Fusion algorithm, which is not consistent with those of the original products. This should be clearly and thoroughly demonstrated.

Additional comments:

1) Title: it is written “the Copernicus Atmospheric Sentinel missions” but it is not precise. The paper only refers to EPS-SG and MTG-S missions. There are other Sentinel missions that are used for the atmosphere which are not analyzed in the paper: Sentinel 5 Precusor/TROPOMI, Sentinel 3 SLSTR and there will probably be other atmospheric sentinel missions in the future. Therefore, the title should be revised: “the Copernicus Atmospheric Sentinel missions” should be replaced by EPS-SG and MTG-S missions.
2) Section 3.2: Please correct “contribute of the measures” … by “contribution of measurements”
3) Keep the same nomenclature for averaging kernels: AK or AKM, but do not use both.
4) Although requested in the previous review, authors provided only scarce additional information on “Atmospheric scenario” in section 2.1. A more detailed description of the used models should be provided.
5) This sentence and others are not correct “… ozone profiles measured in the UV region” Ozone is not measured in a spectral region. Please replace by “ozone profiles derived from measurements in the UV region”
6) In the reply, please cite every change in the response to the reviewer’s documents and the lines where they are. Otherwise, it is very difficult to track the changes done.
7) Last sentence of abstract may be cut into two.

Hide

ED: Reconsider after major revisions (28 Jul 2020) by Thomas von Clarmann (deceased)

AR by Nicola Zoppetti on behalf of the Authors (14 Oct 2020) Author's response Manuscript

ED: Referee Nomination & Report Request started (16 Oct 2020) by Thomas von Clarmann (deceased)

RR by Anonymous Referee #2 (05 Nov 2020)

Suggestions for revision or reasons for rejection

In this new revised version of the manuscript, important changes have been done. Particularly, the fact of considering generic GEO and LEO satellites is indeed the only way to keep current results. However, this “generalization” to a GEO and LEO satellites is not complete and it gives important ambiguity.
The differences in the instrument characteristics considered by Zoppetti et al. paper and the real ones from Sentinel-4/MTG-S and EPS-SG (SNR and spatial sampling, which are key characteristics) do not come from recent changes, because these instruments have been defined like that for at least 10 years (much before 2019). Spatial sampling of most of the instruments are not correct and I repeat that it is impossible that the total column degrees of freedom for ozone from MTG-S/IRS is 5, the simulations from Zoppetti et al. do not correspond to this instrument (before or after 2019). Also, the pixels from IASI-NG were never considered as those described by your paper. It is not a problem of update of specifications.
It is mandatory that this work is clearly dissociated with respect to Sentinel 4 and 5. This is a fundamental correction that should be done to this manuscript and figures. For this, the following changes should be done:
1) The name of the instruments can NOT be S4-TIR, S4-UV, S5-TIR and S5-UV. They must change to general names as GEO-TIR, GEO-UV, LEO-TIR and LEO-UV. Labeling in the whole manuscript and figures should be changed and S4, S5 should not be mentioned.
2) The sentence “according to the specifications of the Sentinel 4 and 5 missions of the Copernicus programme” in the abstract (line 30) should be withdrawn since it is not correct.
3) The sentence “according to the specifications of the atmospheric Sentinels” in line 74 should be changed to “according to specifications similar to those of the atmospheric Sentinels”.
4) As explained before, the sentence (lines 89-90) “It is important to note that the specifications used to simulate L2 products were the state of the art at the time of their usage in 90 the AURORA project context, but are now out of date” is not correct. This should be removed.
5) It should be clearly stated “The spatial sampling (spacing between pixels and shape) and in some cases the signal-to-noise ratio (the geostationary TIR instrument) of the instruments used in the present paper (both geostationary and low orbit ) are different from those of that will be onboard the Sentinel 4/MTG-S and EPS-SG”.
6) The sentence “These specifications are now partially outdated” in line 163 should be removed.
7) The sentence “In table 1 and in the next sections of the paper, we will refer to UVNS/MetOp-SG as S5-UV1, to UVN/MTG as S4-UV1, to IASI-NG/MetOp-SG as S5-TIR and to IR/MTG as S4-TIR” should be modified and the name of the instruments CAN NOT use S4 and S5 since it creates confusion.
8) The sentence “S4-S5” in line 251 and also in Table 1 should be replaced by “GEO-LEO”… and that all through the manuscript.
9) Be aware that using IRS1b from Crevoisier et al 2014 for IRS/MTG-S is NOT correct. The SNR and spectral resolution of IRS1b are NOT those of IRS/MTG-S. This information creates confusion. Table 1 should only describe the GEO and LEO instruments in the present paper without introducing a confusion erroneously referring to actual satellite missions.
10) The statement “to quantify the possible benefits of its application to Sentinel data” should be removed from the conclusion (line 395).

Hide

ED: Reconsider after major revisions (05 Nov 2020) by Thomas von Clarmann (deceased)

AR by Nicola Zoppetti on behalf of the Authors (10 Nov 2020) Author's response Manuscript

ED: Publish subject to minor revisions (review by editor) (10 Nov 2020) by Thomas von Clarmann (deceased)

AR by Nicola Zoppetti on behalf of the Authors (10 Nov 2020) Author's response Manuscript

ED: Publish as is (11 Nov 2020) by Thomas von Clarmann (deceased)

AR by Nicola Zoppetti on behalf of the Authors (11 Nov 2020)

Short summary

The new platforms for Earth observation from space will provide an enormous amount of data that can be hard to exploit as a whole. The Complete Data Fusion algorithm can reduce the data volume while retaining the information of the full dataset. In this work, we applied the Complete Data Fusion algorithm to simulated ozone profiles, and the results show that the fused products are characterized by higher information content compared to individual L2 products.