Preprints
https://doi.org/10.5194/egusphere-2024-2800
© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
https://doi.org/10.5194/egusphere-2024-2800
© Author(s) 2024. This work is distributed under
the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
16 Dec 2024
| 16 Dec 2024
Status: this preprint is open for discussion and under review for Atmospheric Measurement Techniques (AMT).
Satellite Aerosol Composition Retrieval from a combination of three different Instruments: Information content analysis
Abstract. This study focuses on the information content for retrieving Aerosol Optical Depth (AOD) and its components from satellite measurements. We utilize an optimal estimation retrieval algorithm with data from three satellite-based instruments: SLSTR on Sentinel 3A/3B, IASI and GOME-2 on Metop A/B/C. Data are averaged to a common 40x80 km2 grid, temporally aligned within a 60-minute window and cloud masked. A simulation study will be carried out to analyse the information content of the instrument combination, identify retrievable parameters and initiate the development of a uniform retrieval algorithm for the AOD and aerosol components. The simulation study for the information content analysis is implemented using the radiative transfer model SCIATRAN and uses MERRA-2 reanalysis data for AOD and mass mixing ratios of different aerosol components. The study shows 6 to 15 degrees of freedom for the determination of aerosol components dependent on AOD and the underlying surface. The results will be used for the development of a synergistic multi-sensor retrieval algorithm for AOD and its components in cloud-free atmospheres across various surface types.
How to cite. Stöffelmair, U., Popp, T., Vountas, M., and Bösch, H.: Satellite Aerosol Composition Retrieval from a combination of three different Instruments: Information content analysis, EGUsphere [preprint], https://doi.org/10.5194/egusphere-2024-2800, 2024.
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 preprint. The responsibility to include appropriate place names lies with the authors.
Download & links
Download & links
- Preprint
(7528 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 20 Jan 2025)
Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor
| : Report abuse
-
RC1: 'Comment on egusphere-2024-2800', Anonymous Referee #2, 17 Dec 2024
reply
Review of the manuscript titled “Satellite Aerosol Composition Retrieval from a combination of three different Instruments: Information content analysis” by Stöffelmair et. al.,
This manuscript focuses on retrieval of AOD and its components from the valuable apace-borne instruments such as SLSTR on Sentinel 3A/3B, IASI and GOME-2 on Metop A/B/C. Further, this study has used SCIATRAN and uses MERRA-2 reanalysis data to study information content of the conducted retrieval and suggested that 6 to 15 degrees of freedom for the determination of aerosol components dependent on AOD and the underlying surface.
This manuscript is well written and all the analysis are discussed in a very appropriate way. The information content analysis presented in this study will be meaningfully supporting the future AOD retrievals over various regions of the globe. I would recommend this manuscript for publication after these minor comments are addressed.
Minor comments:
1. Change the title, as in this study the satellite is being used to detect the AOD and its components, thus change the title as follows “Aerosol Composition Retrieval from a combination of three different space-borne Instruments: Information content analysis”. This title will be more appropriate.
2. Please correct the line 4 as “A simulation study has been carried out to analyse the information…”
3. The introduction is very well written and easy to follow.
4. The theory of information content and Optimal Estimation is also discussed in a very understandable way.
5. Please add a paragraph focusing only why SCIATRAN in particular is considered in this study and what are the advantages section 3.1 “ Radiative transfer forward model”.
6. Similarly, please add a paragraph focusing only why only these satellite instruments are used in this study and advantages of selecting these instruments in section 3.2 “Satellite Measurements and Observation Vector”.
7. Please improve the titles of all the figures, so that the readers can easily follow the figures.
8. Discuss why there is no data information available over Antarctica, may be add two to three sentences in the result section.
9. Separate the “Discussion and Conclusion” sections into two sections and also in discussion section section add a paragraph about how the information content analysis can be useful for the exiting and future satellite missions.At the end, I would like to mention that this manuscript is short, concise, and the results are well discussed and contains valuable information for the enhancement of current and future retrieval process of AOD and associated components. I wish the authors in advance Merry Christmas and successful start to the new year!
Citation: https://doi.org/10.5194/egusphere-2024-2800-RC1 -
RC2: 'Comment on egusphere-2024-2800', Anonymous Referee #3, 06 Jan 2025
reply
Review of "Satellite Aerosol Composition Retrieval from a combination of three different Instruments: Information content analysis", by Stoeffelmair et al.This paper presents an information content analysis of simulated top-of-atmosphere measurements spanning the spectral range from the UV through to the thermal infrared for the determination of atmospheric aerosol properties (and surface albedo). The simulated measurements are based on the combination of the GOME-2, SLSTR and IASI instruments and the authors state that their analysis is a step towards the development of a retrieval utilising this combination of instruments to retrieve information on aerosol composition, which is an important in determining the direct and indirect radiative impacts of aerosol.Overall, the analysis performed is sound and worthy of publication, however the simplistic representation of the instruments involved, representation of measurement and forward model uncertainty and surface reflectance limit the applicability of the results in predicting the performance and information content of an actual retrieval system applied to real measurements. Thus, I feel that the primary conclusion the authors draw from the work - that this combination of instruments can provide considerable constraint on aerosol composition (in addition to AOD and surface albedo), needs tempering somewhat with the following considerations:
-
The assumption of Lambertian surface reflectance removes one of the key sources of uncertainty when using visible and near-IR measurements for aerosol retrieval. Simulations made using this assumption will thus overestimate the information content of SLSTR measurements available for constraining aerosol properties.
-
The use of simple instrument noise estimates to set the covariance matrices used in the analysis ignores forward model error and correlations between the elements of the measurement vector, both of which will decrease the information available to constrain aerosol properties.
-
The authors also seem to neglect the vertical distribution of aerosol, which can have an impact on TOA radiance significantly exceeding aerosol composition in both the UV and TIR. Dealing with this in a real retrieval scheme will again reduce the information available for constraining aerosol composition.
-
The simplistic way that the measurements from the individual instruments is modelled (neglecting the problems of co-location and matching the very different viewing geometries and spatial sampling of the instruments cited, as well as not attempting to accurately model the spectral response of the real instruments) also neglects important sources of error which would be present in a real retrieval scheme and further abstracts the simulations performed from what might be expected from real measurements.
-
The authors never address another significant source of error and complexity in any aerosol retrieval scheme, which will be even more difficult to deal with when combining three instruments with completely different sensitivities and sampling - clouds!
To be clear, none of these points negate the value of the analysis presented and the conclusion that a combination of measurements like those provided by GOME-2, SLSTR and IASI do provide information on aerosol composition, which is largely lacking in the current generation of satellite-based aerosol products, and which is of great importance in better constraining the role of aerosol in climate. However, in its current form, I feel the paper somewhat over sells the potential of the proposed retrieval approach and I believe the authors need to be more up-front about the limitations of their analysis. There is no way a scheme using this combination of instruments will provide up to 15 independent pieces of information on aerosol composition (and, indeed, current schemes utilising the individual instruments do not match the performance suggested by Figures 1 and 2).For this reason, I believe this paper should be published in AMT, provided the text is modified to make the limitations of the analysis more clear and to more explicitly state the qualitative nature of the results and conclusions (as opposed to a quantitative analysis of the information content of a scheme combining these three instruments). It would also be interesting and a lot more informative if the authors spend some time to investigate the impact of their assumptions and simplifications on the information content. What is the dependence of the degrees of freedom on the assumed measurement covariance and the relative weighting of each instrument, for instance, or what impact do differences in the area sampled by each instrument have?In addition to this general recommendation I also have the follow specific corrections and points on the text:Pg1, ln4: Inconsistent use of tense - a simulation study "has been", or "is" carried out (rather then "will be").Pg1, ln14: A brief definition of direct, semi-direct and indirect effects should be provided on their first use.Pg1, ln17: There is a unnecessary ellipsis (...) here.Pg1, ln21: "also depends" rather than "depends also".Pg1, ln21: Final clause of this sentence, beginning "because different aerosol...", is superfluous and can be deleted.Pg2, ln28: Combine sentence beginning "Observational data" with the previous one and reword to ", so observational data are important for validation and assimilation purposes."Pg2, ln29: Replace the two sentences starting "It is not sufficient..." with "It is not sufficient to constrain just the quantity and distribution of aerosol, composition information is also needed if we want to reduce uncertainties on climate forcing due to aerosol. Hence, there is an important climate research need for global monitoring of aerosol composition from satellite measurements."Pg2, ln35: "there is not enough" (rather than "there are not enough").Pg2, ln36: Insert "properties" after the word "surface".Pg2, ln39: Insert "and" between "AOD" and "composition".Pg2, ln40: "will be developed" rather than "shall be applied".Pg2, ln44: The term "aerosol component" needs to be defined on first use (at least within the body of the text).Pg2, ln44: The paragraph on this line does not scan well and needs to be rewritten. If the authors are trying to justify the need for their proposed algorithm (which, I guess, is essentially an extension of the idea explored with SYNAER by extending to include the thermal-IR), when the GRASP algorithm exists, this can surely be done more succinctly and clearly.Actually, rather than GRASP, the authors should be more concerned with explaining how their proposed retrieval scheme will differ from/improve upon the PMAP (Polar Mulit-sensor Aerosol Optical Properties) algorithm, which is operationally run by EUMETSAT and combines measurements from GOME-2, AVHRR and IASI. This algorithm needs to at least be referenced (perhaps PMAP is a direct development of the work of Hasekamp and Landgraf (2005), which the authors reference later in this section, but I'm not sure!)Pg2, ln57: "additional channels in the visible range" - additional to what?Pg2, ln58: Reword sentence to "IASI is mostly sensitive to mineral dust and larger particles". Also, a more general point to note is that IASI and GOME-2 are also sensitive to elevated stratospheric sulphate aerosol loadings, when compared to SLSTR.Pg3, ln68: Replace sentence starting "This is made possible..." with "The algorithm proposed here has the potential to be applied to predecessor instruments: (A)ATSR(-2) for SLSTR, GOME and SCIAMACHY for GOME-2 and HIRS for IASI, which provide temporal..."Also, I would imagine imminently upcoming instruments also lend themselves to this retrieval approach. Sentinel-4 perhaps?Pg3, ln76: The paragraph starting here would more sensibly be placed earlier in this section.Pg3, ln87: Here the topic of information content and degrees of freedom is launched into without any explanation of what is meant by these phrases. The mathematical definitions can wait until later in the paper, but a simple explanation of their meaning is needed here.Pg4, ln113: Replace "the reflectance" with "spectral reflectance and brightness temperatures" (or whatever is appropriate, but you are surely not using top-of-atmosphere reflectance as a measurement in the thermal-IR).Pg4, ln116: Both "cost function" and "minimised" need to be defined.Pg4, ln120 / equation (2): There is a lot to unpick here. Firstly, I'm not sure why you're introducing an iterative update to a state vector, since the paper is not describing a retrieval or optimisation scheme. That not withstanding, you also do not define S_eplison or where your initial guess at x_i might come from. You might also consider making it clear that superscript "T" and "-1" refer to matrix transpose and inversion respectively.Pg4, ln122: In practice the Jacobian matrix is made up of the derivatives of the forward model wrt to the state vector, not the measurements.Pg4, ln123: Define x-hat.Pg4, ln125 / equation (3): Note that this equation is the linear approximation of error covariance, and thus will only be valid if evaluated at the true state for a non-linear system.Pg5, ln128 / equation (4): The averaging kernel "A" is a key concept/quantity for this paper, so it's probably worth naming. Also, further explanation of what it represents would be desirable. For instance, you note that the diagonal elements denote the sensitivity of a retrieved parameter to its true value, but what do off diagonal elements of this matrix represent?Pg5, ln141: Remove "for radiative transfer and".Pg5, ln142: Replace "observations from UV to TIR" with "observations across the UV to TIR".Pg5, ln149: "the MERRA-2 model comprises precomputed" (plural).Pg6, ln172: I'm not sure what you are referring to by "infrared camera". If you mean the sensor which converts the incoming thermal radiation to an electric current, I think that is an implicit component of a Michelson interferometer.Pg6/7 - Section 3.2: Several details aren't clear from this section and should be explained:- Are radiative-transfer calculations performed at the native resolution of each simulated instrument and then averaged onto the GOME-2 grid, or are all calculations performed on the GOME-2 grid from the start? (I assume the latter).
- For IASI, are you essentially simulating L1C data by performing the radiative transfer calculations assuming a flat instrument response function?
- In the case of GOME-2, are you simulating realistic GOME-2 spectra, or more simplified "GOME-2 like" spectra. Saying you use "a wavelength step of approximately 10 nm" is quite vague.
- I feel the description of the measurement errors/uncertainties should be included in the description of the observation vector.
Pg6, ln173: A Michelson interferometer measures an interferogram (hence the name), which is converted to a spectrum through a Fourier transform operation, not the other way around.Pg7, Section 3.3: I am surprised that you don't say how many elements there are in your state vector in total. This is key parameter, as it defines the maximum DGF value for your model. Also, if I understand correctly, the only variation in aerosol height profile in you model is through differences in the profile of each component in the MERRA-2 database? In this case, you're sensitivity to aerosol component is actually a mixture of the composition/optical properties of each component and it's height profile. Please explain.Pg7, Section 3.4: Values of the a priori state vector are not relevant to your analysis (x_a doesn't appear in equation 4). What is relevant is the a priori covariance matrix, but you don't mention this anywhere. Please correct.Pg7, ln206: Replace "aerosol retrieval from the combination of three instruments" with "the forward model arrangement described above".Pg7, paragraph beginning on ln212: I don't understand the procedure described here. Firstly, why regrid the MERRA-2 mixing ratios from their native resolution to to 1x1 degree lat-lon grid?Also, what is the purpose of calculating monthly means of mass-mixing ratios and then normalising them? Please make it clear what data MERRA-2 actually provides you and what you are converting this into using these calculations.Pg8, paragraph beginning on ln220: So, if I understand correctly, you are ignoring the range of viewing angles observed by the different instruments, and the temporal difference between the Sentinel-3 and Metop platforms? This is a substantial simplification of the actual measurement system and should be noted as such.Pg9, paragraph beginning on ln 224: This description of measurement uncertainties belongs in section 3.2. Here would be the place to include forward model error description, if you'd included any.Pg9, ln232: What is described here is not interpolation or regridding. You are simply sub-sampling the data, with one 1x1 degree box extracted from each 10x10 degree region.Pg9, ln245: Am I correct in think the DGF for "aerosol components" is simply the total of the diagonal elements of the averaging kernel corresponding to the MERRA-2 scaling factors for each component? Please be more explicit.Pg11, Figure 5: Some visual categorisation of the aerosol components into broad types (like dust, sea-salt) would be helpful here - maybe through colouring the labels?Pg12, Figure 6: Similar comment to Figure 5.Pg13, ln300: Where does this mention of soil-type come from? Do you mean surface albedo?Pg13, ln304: I don't think you can claim you've used realistic measurement noise. In general, this paragraph does not go far enough in acknowledging the limitations of your analysis, particularly with regard to simulating a retrieval scheme applied to real-world measurements.Pg13, ln307: Spelling "varying".Pg13, ln310: Replace "data of the three instruments" with "data from the three instruments".Citation: https://doi.org/10.5194/egusphere-2024-2800-RC2 -
Viewed
Total article views: 172 (including HTML, PDF, and XML)
Cumulative views and downloads
(calculated since 16 Dec 2024)
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 133 | 32 | 7 | 172 | 3 | 4 |
- HTML: 133
- PDF: 32
- XML: 7
- Total: 172
- BibTeX: 3
- EndNote: 4
Viewed (geographical distribution)
Total article views: 161 (including HTML, PDF, and XML)
Thereof 161 with geography defined
and 0 with unknown origin.
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
1
Latest update: 17 Jan 2025
Ulrike Stöffelmair
CORRESPONDING AUTHOR
German Aerospace Center (DLR)
University of Bremen, Institute of Environmental Physics (IUP)
Thomas Popp
German Aerospace Center (DLR)
Marco Vountas
University of Bremen, Institute of Environmental Physics (IUP)
Hartmut Bösch
University of Bremen, Institute of Environmental Physics (IUP)
Short summary
The aerosol composition has a large influence on the climate system. This study uses realistic simulated scenarios to look at the information content of the combination of three satellite-based instruments (SLSTR, IASI and GOME-2). It shows that it is possible to retrieve 6 to 15 different aerosol components in addition to the AOD and different surface parameter. The results will be used for the development of a synergistic multi-sensor retrieval algorithms.
The aerosol composition has a large influence on the climate system. This study uses realistic...