the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Mar 2025
Extension of the Complete Data Fusion algorithm to tomographic retrieval products
Abstract. In data analysis of atmospheric remote sensing, the combination of complementary measurements of the same atmospheric state from different sensors operating with different geometries and/or in different spectral ranges is a powerful technique to advance the knowledge of tropospheric and stratospheric processes. The Complete Data Fusion (CDF) is an a posteriori method used so far to combine only one-dimensional atmospheric products (vertical profiles) from simultaneous and independent remote sensing observations of the same air mass. In this study, we demonstrate the applicability of the CDF algorithm to two-dimensional products and show its first application to simulated ozone datasets from the future Infrared Atmospheric Sounding Interferometer New Generation (IASI-NG) mission and the Changing-Atmosphere Infrared Tomography (CAIRT) ESA’s Earth Explorer 11 candidate mission, in nadir- and limb-viewing observational geometry, respectively. We present the analysis of the performance of the CDF in three (one one-dimensional and two two-dimensional) case studies considering different configurations for the acquisitions of the two sensors, evaluating for each the number of degrees of freedom, the Shannon information content, the total errors and the spatial resolution. Furthermore, we quantitatively compare the 1D-CDF and the 2D-CDF performances, demonstrating that the exploitation of tomographic capabilities of atmospheric sensors allows advanced data fusion techniques, like 2D-CDF, to maximize the information extracted from complementary datasets.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Atmospheric Measurement Techniques.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(1029 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2025-1283', Anonymous Referee #1, 30 Apr 2025
Tirelli et al. in this work propose an extension of the Complete Data Fusion framework to tomographic atmospheric retrieval products that are both vertically and horizontally resolved. In preparation for ESA’s EE11 candidate mission CAIRT, this work provides promising new results. Their presentation, however, could be improved, in terms of readability as well as on the technical implementation of the multidimensional CDF.
General comments:
Abstract and introduction (especially lines 32-33): The statement that CDF has been “used so far to combine only one-dimensional atmospheric products (vertical profiles) from simultaneous and independent remote sensing observations of the same air mass” might be confusing, given that several 1D profiles within a certain spatiotemporal extent (e.g. L3 data grid) can be fused, so (exactly) the same air mass sounds too strong. Brief Section 2.1 is clearer on this and might therefore rather be part of the introduction. With that, the difference should be explained between the tomographic retrieval and an along-track curtain of nadir observations, in terms of retrieval and its information characteristics (i.e. the former having vertical and horizontal information smoothing and covariances).
Section 6: Having subsections for the discussion of the three cases would improve readability, but, more importantly, schematic drawings of the spatial configuration of the measurements that are combined in the three cases would provide a better understanding. Moreover, case 1 does not seem to be fully representative, not only because of its dimensionality, but also as the fused product is fully dominated by CAIRT (Figure 1). This should be appropriately discussed (instead of being just mentioned in lines 287-288), also in view of the DFS differences between Tables 1 and 2, which are minor for CAIRT (how?) but substantial for IASI-NG.
It seems inappropriate and misleading to indicate the fused DFS to have a ‘gain’ or X times higher DFS than the individual products, as this largely depends on the number of profiles involved, rather than on the fusion performance.
The CAIRT AKM diagonal in Figure 2 (middle-left) looks interesting but lacks explanation. Could you elaborate on the ‘smile’ shape of the information between 10 and 20 km, and on the offset of the vertical information towards positive along-track distances (closer to CAIRT)? It could then be clarified from the configuration difference why these features are missing in Figure 4 for the third case. Here, the color scale should be updated to avoid saturation. Why keep the same color scale for the AKM diagonals between Figures 2 and 4, and not for the errors, wherefrom it would be clearer that the IASI-NG total errors in Figures 2 and 4 are the same?
Lines 338-339: “Thus, the results shown in Fig. 5 are the same for all the 21 ALT positions.” Is this exactly the same, or by approximation?
Please provide a physical or information-wise interpretation of the synergy factors defined in Equations (21) and (22).
Lines 396-400: Although an attempt is made to explain the origin of the IASI-NG horizontal resolution estimates, this is not clear from the current description.
Technical corrections:
Line 19: Referring only to Aires (2011, 2012) here seems too limited.
Line 44: The meaning of “referred to different true profiles” is not clear here. Does this mean “not covering exactly the same air mass” here?
Appropriate referencing is missing in Section 3.1.
Lines 148-149: To my understanding, the age of air does not equal “the time needed by a tropospheric air parcel to reach the stratosphere”
Line 196: “monochromatic radiance” There should be several of these at least? Please explain.
Lines 204-213: This paragraph contains several unexplained abbreviations.
Line 211: Why 1.4 km? Please explain.
Line 223: Although well-explained, ALT is a somewhat confusing abbreviation for the along-track position, as it is often used for the vertical (altitude) dimension.
Line 342: Which layers?
Data availability: Possibly the reader can already be referred to the foreseen IASI-NG and CAIRT data archives, if already existing?
Citation: https://doi.org/10.5194/egusphere-2025-1283-RC1 - AC1: 'Reply on RC1', Cecilia Tirelli, 20 Aug 2025
-
RC2: 'Comment on egusphere-2025-1283', Anonymous Referee #2, 02 Jul 2025
Tirelli et al. proposes the extension of the Complete Data Fusion framework to extract ozone both vertically and horizontally by exploiting synthetic measurements taken by IASI-NG and the propose CAIRT instruments. The synergistic methods shown here show the potential for useful improvements in the retrieved products. However, clearer descriptions of the instruments and simulation methodology is needed to interpret the results.
General Comments
The description of the vectors in Section 2.3 could be support with a diagram to highlight the difference between the retrieval products from the limb measurements and from the nadir measurements. It would also be good in this section or earlier to specify what your retrieval targets are.
Section 3.1 requires references for the instrument details. While all the relevant information is there, this section is hard to follow as the instruments are introduced and key information about them is then merged in the second paragraph. I would recommend that this section be separated as follows: introduction to why CAIRT (aims, details, advantages etc), introduction IASI-NG (aims, details advantages etc.), why the synergy is useful (similar to your current last paragraph).
In Section 3.2 it is unclear why the ERS are being introduced, how many simulations you are performing, and what was performed in Ererra, 2023 vs in this work. It would be better to include an introduction paragraph for this section to describe the use of these simulations and to justify why two different radiative transfer models are being used. Figures of the ozone profiles should also be included.
Section 6: While the results look promising, it is difficult to interpret the improvements in DFS and the AKMs without information about the ozone profiles. It is not explained why the measurements in these three cases are suspected to be representative of the CAIRT and IASI-NG fusion. More detail is needed about the acquisitions themselves and why they have been averaged in this way. In this section you also refer to the IASI-NG and CAIRT measurements, please be specific that these are synthetic measurements or simulations.
Technical Comments
Line 21: It’s not necessary to cite all of these references, only cite the key ones.
Line 35: It is not initially clear what the features (that you explain in the next sentence) are. Please combine these two sentences and rewrite.
Line 47: Can you define what Multi Target Retrieval is?
Line 90: Define Sn,I and Ss,i
Line 113: This looks like the same citation twice.
Line 127: Equation 14 and the description above could be better explained. If my understanding is correction, this could be rewritten like the below.
X^t = [pjk] where j=[1,m] and k=[1,n]
Line 137: I would suggest moving this to the supplementary or removing it. This does not seem particularly important for the paper.
Line 163: Change your definition of ranges to ‘…645 to 2760 cm1 (15.5 to 3.6 microns)’ for easier reading
Line 185: Please specify if KLIMA is a line-by-line or fast rt model here, what version of the model you are using, and what HITRAN database it is based on.
Line 220: It would be good to see a plot of the NESR and target ARA for this and IASI-NG in Section 3.1.
Line 277: This information should be presented more clearly in a table in Section 4.
Figure 1/3: The light blue is hard to see and the blue is not clear on the figure. Where has your aprori error come from? What do the profiles look like?
Line 351: The description of a synergy factor should be moved to Section 5 where you introduced other quantifiers of performance.
Citation: https://doi.org/10.5194/egusphere-2025-1283-RC2 - AC2: 'Reply on RC2', Cecilia Tirelli, 20 Aug 2025
Status: closed
-
RC1: 'Comment on egusphere-2025-1283', Anonymous Referee #1, 30 Apr 2025
Tirelli et al. in this work propose an extension of the Complete Data Fusion framework to tomographic atmospheric retrieval products that are both vertically and horizontally resolved. In preparation for ESA’s EE11 candidate mission CAIRT, this work provides promising new results. Their presentation, however, could be improved, in terms of readability as well as on the technical implementation of the multidimensional CDF.
General comments:
Abstract and introduction (especially lines 32-33): The statement that CDF has been “used so far to combine only one-dimensional atmospheric products (vertical profiles) from simultaneous and independent remote sensing observations of the same air mass” might be confusing, given that several 1D profiles within a certain spatiotemporal extent (e.g. L3 data grid) can be fused, so (exactly) the same air mass sounds too strong. Brief Section 2.1 is clearer on this and might therefore rather be part of the introduction. With that, the difference should be explained between the tomographic retrieval and an along-track curtain of nadir observations, in terms of retrieval and its information characteristics (i.e. the former having vertical and horizontal information smoothing and covariances).
Section 6: Having subsections for the discussion of the three cases would improve readability, but, more importantly, schematic drawings of the spatial configuration of the measurements that are combined in the three cases would provide a better understanding. Moreover, case 1 does not seem to be fully representative, not only because of its dimensionality, but also as the fused product is fully dominated by CAIRT (Figure 1). This should be appropriately discussed (instead of being just mentioned in lines 287-288), also in view of the DFS differences between Tables 1 and 2, which are minor for CAIRT (how?) but substantial for IASI-NG.
It seems inappropriate and misleading to indicate the fused DFS to have a ‘gain’ or X times higher DFS than the individual products, as this largely depends on the number of profiles involved, rather than on the fusion performance.
The CAIRT AKM diagonal in Figure 2 (middle-left) looks interesting but lacks explanation. Could you elaborate on the ‘smile’ shape of the information between 10 and 20 km, and on the offset of the vertical information towards positive along-track distances (closer to CAIRT)? It could then be clarified from the configuration difference why these features are missing in Figure 4 for the third case. Here, the color scale should be updated to avoid saturation. Why keep the same color scale for the AKM diagonals between Figures 2 and 4, and not for the errors, wherefrom it would be clearer that the IASI-NG total errors in Figures 2 and 4 are the same?
Lines 338-339: “Thus, the results shown in Fig. 5 are the same for all the 21 ALT positions.” Is this exactly the same, or by approximation?
Please provide a physical or information-wise interpretation of the synergy factors defined in Equations (21) and (22).
Lines 396-400: Although an attempt is made to explain the origin of the IASI-NG horizontal resolution estimates, this is not clear from the current description.
Technical corrections:
Line 19: Referring only to Aires (2011, 2012) here seems too limited.
Line 44: The meaning of “referred to different true profiles” is not clear here. Does this mean “not covering exactly the same air mass” here?
Appropriate referencing is missing in Section 3.1.
Lines 148-149: To my understanding, the age of air does not equal “the time needed by a tropospheric air parcel to reach the stratosphere”
Line 196: “monochromatic radiance” There should be several of these at least? Please explain.
Lines 204-213: This paragraph contains several unexplained abbreviations.
Line 211: Why 1.4 km? Please explain.
Line 223: Although well-explained, ALT is a somewhat confusing abbreviation for the along-track position, as it is often used for the vertical (altitude) dimension.
Line 342: Which layers?
Data availability: Possibly the reader can already be referred to the foreseen IASI-NG and CAIRT data archives, if already existing?
Citation: https://doi.org/10.5194/egusphere-2025-1283-RC1 - AC1: 'Reply on RC1', Cecilia Tirelli, 20 Aug 2025
-
RC2: 'Comment on egusphere-2025-1283', Anonymous Referee #2, 02 Jul 2025
Tirelli et al. proposes the extension of the Complete Data Fusion framework to extract ozone both vertically and horizontally by exploiting synthetic measurements taken by IASI-NG and the propose CAIRT instruments. The synergistic methods shown here show the potential for useful improvements in the retrieved products. However, clearer descriptions of the instruments and simulation methodology is needed to interpret the results.
General Comments
The description of the vectors in Section 2.3 could be support with a diagram to highlight the difference between the retrieval products from the limb measurements and from the nadir measurements. It would also be good in this section or earlier to specify what your retrieval targets are.
Section 3.1 requires references for the instrument details. While all the relevant information is there, this section is hard to follow as the instruments are introduced and key information about them is then merged in the second paragraph. I would recommend that this section be separated as follows: introduction to why CAIRT (aims, details, advantages etc), introduction IASI-NG (aims, details advantages etc.), why the synergy is useful (similar to your current last paragraph).
In Section 3.2 it is unclear why the ERS are being introduced, how many simulations you are performing, and what was performed in Ererra, 2023 vs in this work. It would be better to include an introduction paragraph for this section to describe the use of these simulations and to justify why two different radiative transfer models are being used. Figures of the ozone profiles should also be included.
Section 6: While the results look promising, it is difficult to interpret the improvements in DFS and the AKMs without information about the ozone profiles. It is not explained why the measurements in these three cases are suspected to be representative of the CAIRT and IASI-NG fusion. More detail is needed about the acquisitions themselves and why they have been averaged in this way. In this section you also refer to the IASI-NG and CAIRT measurements, please be specific that these are synthetic measurements or simulations.
Technical Comments
Line 21: It’s not necessary to cite all of these references, only cite the key ones.
Line 35: It is not initially clear what the features (that you explain in the next sentence) are. Please combine these two sentences and rewrite.
Line 47: Can you define what Multi Target Retrieval is?
Line 90: Define Sn,I and Ss,i
Line 113: This looks like the same citation twice.
Line 127: Equation 14 and the description above could be better explained. If my understanding is correction, this could be rewritten like the below.
X^t = [pjk] where j=[1,m] and k=[1,n]
Line 137: I would suggest moving this to the supplementary or removing it. This does not seem particularly important for the paper.
Line 163: Change your definition of ranges to ‘…645 to 2760 cm1 (15.5 to 3.6 microns)’ for easier reading
Line 185: Please specify if KLIMA is a line-by-line or fast rt model here, what version of the model you are using, and what HITRAN database it is based on.
Line 220: It would be good to see a plot of the NESR and target ARA for this and IASI-NG in Section 3.1.
Line 277: This information should be presented more clearly in a table in Section 4.
Figure 1/3: The light blue is hard to see and the blue is not clear on the figure. Where has your aprori error come from? What do the profiles look like?
Line 351: The description of a synergy factor should be moved to Section 5 where you introduced other quantifiers of performance.
Citation: https://doi.org/10.5194/egusphere-2025-1283-RC2 - AC2: 'Reply on RC2', Cecilia Tirelli, 20 Aug 2025
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 458 | 52 | 21 | 531 | 15 | 38 |
- HTML: 458
- PDF: 52
- XML: 21
- Total: 531
- BibTeX: 15
- EndNote: 38
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1