|I was asked by the editor to submit a review of acp-2019-208, “Retrieval of aerosol composition directly from satellite and ground-based measurements” by Lei Li et al. The circumstances were that Reviewer #2 could not review the revision, and the editor asked me specifically to 1) evaluate the response to Reviewer #2’s original comments and 2) to offer my own assessment. I have tried to do both. |
I have gone over carefully Reviewer #2’s review and the formal response from the authors. I have also read the revised manuscript, the version with the tracked revisions, keeping in mind Reviewer #2’s comments. I did not read the manuscript intensely, as if I had been a fresh reviewer, because I was not looking to add a whole bunch of new work to the authors. As I read the pdf, I put my immediate comments as sticky notes and editor’s marks. There are 40 comments in the comments list. Most of these are minor editorial fixes, like typos. I've attached this annotated file as app-2019-208-author_response-version1_LR.pdf.
Please note that I changed “component” when it is a noun to “components” (plural), but left “component” (singular) when it is used as an adjective.
Overall, my own assessment is that the paper is quite good and publishable, “as is”, with only typos and minor edits needed. However, I do agree with Reviewer #2’s assessment on almost every point. This was a very insightful review. Most, but not all, of Reviewer #2’s comments have been satisfactorily answered. Should the authors be required to answer everything? Not my call. To fulfill my obligation to the editor, I will describe my assessment of the authors’ responses to Reviewer #2 next.
First, I will reveal myself. This is Lorraine Remer writing.
Reviewer #2 had several major comments.
1. First was a positive statement:
“the most important aspect of this work is it provides something of a
forward operator to perform more apples to apples comparisons between satellite and
models to close the radiance fields. I also agree that discussion of this point should be expanded “
This is the most important aspect of the work and it is hidden in a single statement on lines 268-271 that I’ve put into highlight in my comments. This really is the main contribution of this method and paper. I actually came to this on my own after reading lines 268-271, without understanding the previous reviewers’ comments. So, now you have it. Three independent reviewers all see the same thing. The main contribution the GRASP/Component retrieval makes is that it retrieves parameters that match models. This ability supercedes all the nuances of avoiding working from retrieved optical properties (which MISR does anyway) and LUTs.
Thus, the authors’ response to this comment is barely satisfactory.
2. Then a serious criticism:
“However, these parameters are by no means
equivalent to “composition,” which is their premise for the entire paper”
I warned the authors about this very issue when I first encountered this work as a Ph.D. dissertation over a year ago. There is no excellent solution to the semantics, but composition is definitely wrong. Component is not perfect, but it works.
Thus, the authors’ response to this comment is good.
3. Next, a suggestion:
“They should be more up front as to what
they are doing, or spend much more time explaining why they think there is
something fundamentally different in their approach”
The authors responded to this comment very well. The additional text is much welcomed and adds much to the paper.
I do have a problem with “a first attempt to derive aerosol components from satellite”. Component is not much different from type. I don’t see this work as fundamentally different from MISR. What might be said instead:
"a first attempt to derive aerosol components without Look-Up Tables from satellites" OR "a first attempt at this more realistic method to derive aerosol components from satellite"
OR "a first attempt to derive aerosol components linked to their hygroscopicity from satellite"
OR "a first attempt to derive aerosol components from satellite specifically tied to global chemical transport model quantities"
OR "a first attempt to derive aerosol components in this manner"
I happen to like the fourth statement best.
4. A question of not understanding what is going on in the algorithm:
one considers the GRASP algorithm simply has a smoothness of fit contain [sic], but
otherwise can pick any refractive index and size it wants, why not just take the
standard GRASP algorithm as it is, and after the fact match the size and refractive
index to any basis function they like?”
The authors don’t really address this in the text, as far as I can tell, except for updating the Figure. I like the new figure, and I seem to understand what is going on. I think. But obviously an informed reader (Reviewer #2) missed some key things. It might be worthwhile to add a simple statement at the beginning of the methodology to answer this question directly.
Authors response: Not satisfactory.
5. Another question concerning the algorithm:
“Is the goodness of fit always the
same from the standard GRASP algorithm, or does your predefinition of species leave
a residual? If so, how big is that residual?”
Here the authors dig in, provide a good response, and add some key information in the form of a few summary sentences in the text. One thing to think about is that I found figure R2 more informative than the scatter plots that are found in the manuscript itself. The authors may want to add R2 to the Supplementary material.
Authors response: Very good.
6. The issue of verification:
“there is really very little verification work provide
that shows that the results of the retrievals are fundamentally better than other
categorization methods (that is getting back to the question in point 2 (if this method
leaves a residual from the free running GRASP algorithm).”
The fact is that it is extremely difficult to verify components. Still, whatever is shown in the paper only tells us that the component method is “as good” as standard GRASP, not “better”, in the parameters that standard GRASP derives. I think that is ok. But it would have been nice to have a solid statement in the text outlining the verification strategy. Maybe saying that the goal was to match standard GRASP in terms of AOD, AE and SSA, which gives support to the component divisions, but doesn’t validate them. I think would satisfy Reviewer #2 better than what was said.
Response to review: Not satisfactory.