the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Algorithm to retrieve aerosol optical properties using lidar measurements on board the EarthCARE satellite
Abstract. Algorithms were developed to produce ATLID (Atmospheric Lidar) L2 aerosol products using ATLID L1 data. The algorithms estimated the following four products: (1) Layer identifiers such as aerosols, clouds, clear-skies, or surfaces (feature masks) were estimated by the combined use of vertically variable criteria and spatial continuity methods developed for the CALIOP (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation) analysis. (2) Aerosol optical properties such as extinction coefficient, backscatter coefficient, depolarization ratio, and lidar ratio at 355 nm were estimated by our developed optimization method using the Gauss-Newton method combined with the line search method developed for ground-based measurements. (3) Six aerosol types, namely smoke, pollution, marine, pristine, dusty-mixture, and dust were identified based on a two-dimensional diagram of the lidar ratio and depolarization ratio at 355 nm developed by cluster-analysis using the AERONET (AErosol RObotic NETwork) dataset with ground-based lidar data. (4) The planetary boundary layer height was determined using the improved wavelet covariance transform method for the ATLID analysis. We evaluated the algorithm’s performance using simulated ATLID L1 data generated by Joint-Simulator (Joint Simulator for Satellite Sensors), incorporating aerosol and cloud distributions from numerical models. Results from applying the algorithms to the simulated ATLID L1 data with realistic signal noise added for aerosol or cloud predominant cases revealed: (1) misidentification of aerosol and cloud layers by the feature mask algorithm was relatively low, approximately 10 %; (2) the retrieval errors of aerosol optical properties were 0.08 × 10-7 ± 1.12 × 10-7 m-1sr-1 (2 ± 34 % in relative error) for backscatter coefficient and 0.01 ± 0.07 (4 ± 27 %) for depolarization ratio; (3) aerosol type classification was generally performed well, with a 37 % of misclassification for dust. These results indicate that the algorithm’s capability to provide valuable insights into the global distribution of aerosols and clouds, facilitating assessments of their climate impact through atmospheric radiation processes.
- Preprint
(1654 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on amt-2024-100', Anonymous Referee #1, 01 Aug 2024
Review of "Algorithm to retrieve aerosol optical properties using lidar
measurements on board the EarthCARE satellite"
By T. Nishizawa et al.General Remarks
===============This paper provides a useful concise overview of the JAXA EarthCARE lidar
products and will be a useful reference for the community. I recommend publication.
There are, however, several mainly editorial issues that need to be addressed.
Title:
------This paper describes several algorithms. I suggest changing
"Algorithm" to "Algorithms" in the title.
Abstract
--------
The abstract is, in general, awkward to read. It should be re-writtenFor example:
-EarthCARE should be introduced. "ATLID (Atmospheric lidar)" does not
mean much to many readers by itself without more context.-"optimization method using the Gauss-Newton method combined..". The
numerical methods used in the optimization procedure are not
interesting enough to be included in the abstract ! It would be
more suitable to mention what is being optimized (e.g. have you
implemented an optimal estimation type procedure ? )-"algorithm's performance". Since more than one algorithm is being
treated, the phrase "The performance of the various algorithms was
evaluated"
1 Introduction:
---------------Line 53: "...extraction of the component parallel to the laser
polarization (co-polar component)..."Line 54 delete "published" == > "calibrated"
Line 55 "understandings" ==> "understanding"
Line 77: "...generate JAXA L2 products using.."
Line 79: "..cloud properly estimation.."
2 Algorithm flow and products
------------------------------Line 85 : "Initially, the algorithm" ... which algorithm ? I guess
this is referring to the "signal smoothing" step in Fig 1. ? Please
re-work this sentence.Line 114 : "..ECMWF forecast model.."
3 Algorithm
------------Layer Identification
--------------------Line 115: "Algorithm" ==> Algorithm
Line 129 : "..and linear depolarization ratio.."
Line 133-134: Pm and Pr are not defined ! Or does e.g. Pm=beta_atn_M ?
and Pr=beta_atn_R ? If this is the case, it is unnecessary and
confusing in the description. Please adjust the subsequent description
and Equations 3 and 4 to use beta_atn_M etc..Line 168: "SN's" ? Do you mean "..are not identified separately using the SNR" ?
Aerosol optical properties
---------------------------Can the authors give an indication of how computationally demanding
their approach is ? i.e. how long (and on what type of computing
system) does it take to profile a frame of Atlid data ?Line 193: If I understand correctly, the forward model being employed
is described by Eq1 1a-1c. Is there any account of lidar multiple
scattering ?Line 198: "...optimize the vertical profiles of the POP to the L1
data...". I am not sure what is meant here ? Maybe the authors
mean to say that "...optimize the difference between the observed and
forward modeled L1 profiles based on the POP profiles"Please describe how the w terms in Eq 5 are determined ? I guess they
are the log uncertainties based on the (linear) error estimations in
the alpha and beta determinations ? Are the w terms also adjusted to
control the "smoothness" of the results ?Esq. (5). It looks like there are extra "-" signs in the last three
terms of the equation. e.g. -ln(-alpha_p(z_(i+1)).Line 199: Is this really and "optimal estimation technique" ? The
method looks like some sort of forward modeling approach coupled with
smoothing constraints but I do not think it can be described as an
"optimal estimation" technique. I.E. optimal estimation involve some
sort of a prior constraint, not smoothness constraints.Line 217-220: Here the authors (finally) introduce the state-vector (x). The
discussion would be much easier to follow if this was done explicitly
at the beginning of this sub-section.Line 219 : Provide a reference for the "Armijo" rule.
Aerosol type classification (Target mask).
------------------------------------------Line 244: Please provide a reference for the "fuzzy c-means method"
PBL height
------------Line 264 : "...ratio are directly influenced...."
Line 275 : "..results in a small;.." ? Small what ? This sentence seems corrupt.
Results and Discussion
-----------------------Line 293 : "..algorithm,.." ==> "..algorithms,..".
Line 298 : Were lidar multiple-scattering (MS) effects included in the
simulations ? MS is described later in lines 359-365 but it is
unclear(to me) if these effects were incorporated into the
simulations used in this paper.Line 311 : "..highliting.." ==> "..highlighting.."
Conclusion
-----------Line 412 : "...will be released as JAXA's L2 ATLID standard products."
References
-----------Check the references to the other special issue papers and update them if appropriate.
Review of "Algorithm to retrieve aerosol optical properties using lidar
measurements on board the EarthCARE satellite"
By T. Nishizawa et al.General Remarks
===============This paper provides a useful concise overview of the JAXA EarthCARE lidar
products and will be a useful reference for the community. I recommend publication.
There are, however, several mainly editorial issues that need to be addressed.
Title:
------This paper describes several algorithms. I suggest changing
"Algorithm" to "Algorithms" in the title.
Abstract
--------
The abstract is, in general, awkward to read. It should be re-writtenFor example:
-EarthCARE should be introduced. "ATLID (Atmospheric lidar)" does not
mean much to many readers by itself without more context.-"optimization method using the Gauss-Newton method combined..". The
numerical methods used in the optimization procedure are not
interesting enough to be included in the abstract ! It would be
more suitable to mention what is being optimized (e.g. have you
implemented an optimal estimation type procedure ? )-"algorithm's performance". Since more than one algorithm is being
treated, the phrase "The performance of the various algorithms was
evaluated"
1 Introduction:
---------------Line 53: "...extraction of the component parallel to the laser
polarization (co-polar component)..."Line 54 delete "published" == > "calibrated"
Line 55 "understandings" ==> "understanding"
Line 77: "...generate JAXA L2 products using.."
Line 79: "..cloud properly estimation.."
2 Algorithm flow and products
------------------------------Line 85 : "Initially, the algorithm" ... which algorithm ? I guess
this is referring to the "signal smoothing" step in Fig 1. ? Please
re-work this sentence.Line 114 : "..ECMWF forecast model.."
3 Algorithm
------------Layer Identification
--------------------Line 115: "Algorithm" ==> Algorithm
Line 129 : "..and linear depolarization ratio.."
Line 133-134: Pm and Pr are not defined ! Or does e.g. Pm=beta_atn_M ?
and Pr=beta_atn_R ? If this is the case, it is unnecessary and
confusing in the description. Please adjust the subsequent description
and Equations 3 and 4 to use beta_atn_M etc..Line 168: "SN's" ? Do you mean "..are not identified separately using the SNR" ?
Aerosol optical properties
---------------------------Can the authors give an indication of how computationally demanding
their approach is ? i.e. how long (and on what type of computing
system) does it take to profile a frame of Atlid data ?Line 193: If I understand correctly, the forward model being employed
is described by Eq1 1a-1c. Is there any account of lidar multiple
scattering ?Line 198: "...optimize the vertical profiles of the POP to the L1
data...". I am not sure what is meant here ? Maybe the authors
mean to say that "...optimize the difference between the observed and
forward modeled L1 profiles based on the POP profiles"Please describe how the w terms in Eq 5 are determined ? I guess they
are the log uncertainties based on the (linear) error estimations in
the alpha and beta determinations ? Are the w terms also adjusted to
control the "smoothness" of the results ?Esq. (5). It looks like there are extra "-" signs in the last three
terms of the equation. e.g. -ln(-alpha_p(z_(i+1)).Line 199: Is this really and "optimal estimation technique" ? The
method looks like some sort of forward modeling approach coupled with
smoothing constraints but I do not think it can be described as an
"optimal estimation" technique. I.E. optimal estimation involve some
sort of a prior constraint, not smoothness constraints.Line 217-220: Here the authors (finally) introduce the state-vector (x). The
discussion would be much easier to follow if this was done explicitly
at the beginning of this sub-section.Line 219 : Provide a reference for the "Armijo" rule.
Aerosol type classification (Target mask).
------------------------------------------Line 244: Please provide a reference for the "fuzzy c-means method"
PBL height
------------Line 264 : "...ratio are directly influenced...."
Line 275 : "..results in a small;.." ? Small what ? This sentence seems corrupt.
Results and Discussion
-----------------------Line 293 : "..algorithm,.." ==> "..algorithms,..".
Line 298 : Were lidar multiple-scattering (MS) effects included in the
simulations ? MS is described later in lines 359-365 but it is
unclear(to me) if these effects were incorporated into the
simulations used in this paper.Line 311 : "..highliting.." ==> "..highlighting.."
Conclusion
-----------Line 412 : "...will be released as JAXA's L2 ATLID standard products."
References
-----------Check the references to the other special issue papers and update them if appropriate.
Review of "Algorithm to retrieve aerosol optical properties using lidar
measurements on board the EarthCARE satellite"
By T. Nishizawa et al.General Remarks
===============This paper provides a useful concise overview of the JAXA EarthCARE lidar
products and will be a useful reference for the community. I recommend publication.
There are, however, several mainly editorial issues that need to be addressed.
Title:
------This paper describes several algorithms. I suggest changing
"Algorithm" to "Algorithms" in the title.
Abstract
--------
The abstract is, in general, awkward to read. It should be re-writtenFor example:
-EarthCARE should be introduced. "ATLID (Atmospheric lidar)" does not
mean much to many readers by itself without more context.-"optimization method using the Gauss-Newton method combined..". The
numerical methods used in the optimization procedure are not
interesting enough to be included in the abstract ! It would be
more suitable to mention what is being optimized (e.g. have you
implemented an optimal estimation type procedure ? )-"algorithm's performance". Since more than one algorithm is being
treated, the phrase "The performance of the various algorithms was
evaluated"
1 Introduction:
---------------Line 53: "...extraction of the component parallel to the laser
polarization (co-polar component)..."Line 54 delete "published" == > "calibrated"
Line 55 "understandings" ==> "understanding"
Line 77: "...generate JAXA L2 products using.."
Line 79: "..cloud properly estimation.."
2 Algorithm flow and products
------------------------------Line 85 : "Initially, the algorithm" ... which algorithm ? I guess
this is referring to the "signal smoothing" step in Fig 1. ? Please
re-work this sentence.Line 114 : "..ECMWF forecast model.."
3 Algorithm
------------Layer Identification
--------------------Line 115: "Algorithm" ==> Algorithm
Line 129 : "..and linear depolarization ratio.."
Line 133-134: Pm and Pr are not defined ! Or does e.g. Pm=beta_atn_M ?
and Pr=beta_atn_R ? If this is the case, it is unnecessary and
confusing in the description. Please adjust the subsequent description
and Equations 3 and 4 to use beta_atn_M etc..Line 168: "SN's" ? Do you mean "..are not identified separately using the SNR" ?
Aerosol optical properties
---------------------------Can the authors give an indication of how computationally demanding
their approach is ? i.e. how long (and on what type of computing
system) does it take to profile a frame of Atlid data ?Line 193: If I understand correctly, the forward model being employed
is described by Eq1 1a-1c. Is there any account of lidar multiple
scattering ?Line 198: "...optimize the vertical profiles of the POP to the L1
data...". I am not sure what is meant here ? Maybe the authors
mean to say that "...optimize the difference between the observed and
forward modeled L1 profiles based on the POP profiles"Please describe how the w terms in Eq 5 are determined ? I guess they
are the log uncertainties based on the (linear) error estimations in
the alpha and beta determinations ? Are the w terms also adjusted to
control the "smoothness" of the results ?Esq. (5). It looks like there are extra "-" signs in the last three
terms of the equation. e.g. -ln(-alpha_p(z_(i+1)).Line 199: Is this really and "optimal estimation technique" ? The
method looks like some sort of forward modeling approach coupled with
smoothing constraints but I do not think it can be described as an
"optimal estimation" technique. I.E. optimal estimation involve some
sort of a prior constraint, not smoothness constraints.Line 217-220: Here the authors (finally) introduce the state-vector (x). The
discussion would be much easier to follow if this was done explicitly
at the beginning of this sub-section.Line 219 : Provide a reference for the "Armijo" rule.
Aerosol type classification (Target mask).
------------------------------------------Line 244: Please provide a reference for the "fuzzy c-means method"
PBL height
------------Line 264 : "...ratio are directly influenced...."
Line 275 : "..results in a small;.." ? Small what ? This sentence seems corrupt.
Results and Discussion
-----------------------Line 293 : "..algorithm,.." ==> "..algorithms,..".
Line 298 : Were lidar multiple-scattering (MS) effects included in the
simulations ? MS is described later in lines 359-365 but it is
unclear(to me) if these effects were incorporated into the
simulations used in this paper.Line 311 : "..highliting.." ==> "..highlighting.."
Conclusion
-----------Line 412 : "...will be released as JAXA's L2 ATLID standard products."
References
-----------Check the references to the other special issue papers and update them if appropriate.
Citation: https://doi.org/10.5194/amt-2024-100-RC1 -
RC2: 'Comment on amt-2024-100', Anonymous Referee #2, 13 Sep 2024
The manuscript describes the JAXA algorithm to derive lidar L1 and L2 data from EarthCARE’s Atmospheric Lidar (ATLID). The paper provides an important contribution with respect to JAXA’s EarthCARE data and should be published after addressing some mainly minor or editorial points.
Abstract
L15: ‘… were estimated by our developed optimization method…’ should it be rather be ‘… calculated’
L27f: ‘37% of misclassification for dust’ – should not be addressed in the Abstract, but what causes this misclassification and why is it so high for dust?
Introduction
The introduction has no clear thread but appears to me to have several jumps in topics. This is often confusing. The authors should consider rephrasing the introduction to achieve a more clear structure and argumentation. Some examples:
In large part it is written as it was dealing with satellite lidars ATLID and CALIPSO, but in ll59-61 retrievals are mentioned to derive microphysical and optical properties from wavelength combinations that are neither used by CALIPSO nor by ATLID. This is a bit confusing.
L61: ‘Furthermore, the extinction coefficients…’ – further to what. Are the extinction coefficients not part of the optical properties? The authors should give some more information here.
L77: Based on the above background – Not clear to me what is meant with above background.
Algorithm flow and products
Figure 1: Are all the variables and abbreviations explained in the text? Please check.
L106: ‘…POP is of aerosol or cloud origin is …’ – tow times ‘is’
Layer identification (Feature mask)
This paragraph is hard to read and understand as some of the parameters are not introduced or their meaning is not introduced. The authors should consider to rephrase this paragraph in the sake for the reader to better understand. They should check carefully that all parameters are described.
L139 and following: What is meant with ‘significant’?
LL158-160: Not clear to me. What is Pp?
L168: ‘aeroso.l’ typo
Aerosol type classification
This paragraph is not fully clear to me. E.g. the required optical properties and size distributions from AERONET measurements. Are they used as a general input for different aerosols? Or is the closest measurement in space and time required for the algorithm? The authors should think of rephrasing part of this paragraph.
L232: For sure these are the main aerosol types, but what about e.g. volcanic aerosols? Can the authors give a short assessment about the uncertainties in retrieving the radiative (and climate) impact, as well as the impact on Aerosol-Cloud-Interaction, if volcanic aerosols are classified as dust.
L243: Is the spheroid model sufficient for dust?
LL157-159: There are also studies about long-range transported dust and on Arabian dust. The authors should consider to discuss them as well.
Planetary boundary layer height
The first paragraph reads more like a general introduction to this topic. The authors should consider to move this to the introduction to motivate the need to develop the corresponding algorithms.
L275: What is mean by ‘results in a small’?
Results and Discussion
Figure 5: Why do the cases for 0.3 km and 1 km horizontal resolution data look so different? Have different cases been used? Or rather, it looks like the example for 0.3 km is just the excerpt for the part form 0 to about 1.31 in the 1 km data. If this is the case it is inconsistent with Figure 3.
LL361-368: The authors should consider to move that to the retrieval part. – Is multiple scattering considered in their algorithm?
L372: What is meant by ‘layers’ here?
LL372-374: What is the main reason for the misidentification?
Figure 9: It is interesting to see, that dust is quite well detected (although many false classification parts), but pristine is not really captured well. It would be interesting to see; which aerosols have a high certainty to be classified and which tend to be more often misclassified. And what are the aerosols they are mainly classified at.
LL383-385: The authors already mentioned that that the cloud algorithm is presented in separate publications. To include it in four lines in this manuscript does not give a meaningful description of the cloud algorithm nor a meaningful contribution to this manuscript. The authors should consider to remove it.
General remark: It would have been interesting to see also a case with different aerosol types just only one example with dust as the dominant aerosol type.
Conclusion
L401: It has to be made clear that this is not real ATLID L1 data but simulations. Please also check throughout the manuscript.
Conclusion and Abstract are very similar. The authors should think about rewriting the one or the other. Instead of repeating it would have been interesting to give a discussion on the main contributors to the uncertainties.
L413: What are the standard algorithms?
L413f: ‘we have been developing an algorithm to estimate…’ – Was that shown in this manuscript? If it is of importance to mention it here, it should be better introduced and motivated.
Citation: https://doi.org/10.5194/amt-2024-100-RC2
Viewed
HTML | XML | Total | BibTeX | EndNote | |
---|---|---|---|---|---|
326 | 84 | 26 | 436 | 14 | 13 |
- HTML: 326
- PDF: 84
- XML: 26
- Total: 436
- BibTeX: 14
- EndNote: 13
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1