|Review of the revised manuscript “First assessment of Aeolus L2A particle backscatter coefficient retrievals in the Eastern Mediterranean” by Gkikas et al.|
The authors responded to most of the reviewers’ comments. There was an effort to make the paper lighter but long, heavy sentences, of the kind you need to read twice to be sure you understand, are still present. I propose some simplifications in the specific comments.
Following reviewer #2, I think it is important to mention that the SCA mid-bin backscatter is not expected to perform better than the normal SCA, the mid-bin algorithm is designed to reduce the extinction errors. The authors are aware of this, because it is mentioned on l. 397-405. But the comparison of performance between the two results is still at the heart of the paper. I am not sure the paper is clear enough on this point. From his/her comments, it seems that even reviewer #1 misunderstood this.
Also, it is important to note that the SCA performance is not the performance of Aeolus itself. It’s one of the optical properties products and, hopefully, there is still room for improvement.
Getting back to the main subject of the paper, the validation of satellite observations from ground-based measurements, I think the work is useful and presents an important contribution to a difficult problem. A meta-analysis, exploiting more numerous profiles from more locations would be very useful (e.g. co-ordinated at EARLINET level? Ok, it is mentioned in the conclusion!).
On the scientific side, I think most limitations of the study are properly acknowledged.
l. 27: “along with wind HLOS profiles” could be removed to make the sentence lighter. Or simply replaced by “also”.
l. 31 “(capital of Greece)” the administrative status is not a geographical indication. Centre of Greece?
Several occurrences in the text of “the performance downgrades” (l.45, l.53, l.1157). I would replace “downgrade” with “degrade”. I am being picky but this bothered me. You wouldn’t use “upgrade” if the performance improved, would you?
l. 287 named by -> named after?
l.295 The orbital velocity is actually closer to 7.7 km/s
l.298: The precision that solar panels are facing the sun is not useful. If you want to keep this sentence, you should focus on the telescope rather than on the panels. Something like “The telescope is pointed to the right of the flight direction, aiming into the night hemisphere”. As you are not discussing the influence of solar background signal, you could omit this sentence.
l. 300 “HSRL lidar” contains lidar twice. “HSRL” alone or “HSR lidar”?
l. 309 As you want to focus on the description of the geometry here, you may simply say that “Aeolus provides observations at the slant nadir angle of 35 degrees”.
l.315 You could use a simpler structure: “Therefore, the discrimination between aerosol and clouds, as well as the typing of aerosols, is challenging”.
l. 336: I would call the product “(particulate) optical properties product”, to stress that it focusses purely on optical properties, without any classification or any chemistry consideration.
l. 344: Is it correct to replace “backscatter (either molecular or particulate) where the squared one-way transmission through the atmosphere is taken into account” by “attenuated backscatter”?
l. 346-354: I am not sure that this description of the cross-talk correction with the coefficients name is needed. This paragraph could probably be summarized in one sentence.
l. 354-405: This is a long introduction to explain the interest of the “SCA mid-bin” product. As a reader, I would prefer to be pointed to the ATBD or to Flament et al. 2021. It would save one page of reading and allow the reader to focus on the actual comparison.
- Do you have a reference on comparisons of the MLE to ground based lidars? Ehlers et al. only provide comparisons with simulated and CALIPSO data.
- Also, the sentence spanning l. 410 to l. 413 is a long one again. “Comparison against ground based observation showed that the precision of the extinction and LR retrieved by the MLE is much better than the one from the SCA”.
- What are “SCA end-to-end simulated optical products”? It is not clearly described.
l. 566: It is not clear to me whether this is the standard deviations with respect to time or space. I would think time.
l. 588 “cycles” should be “circles”?
l.597-598: “strong horizontal variability” fig 1-iii, 1-iv and S1 say the opposite: strong spatial homogeneity around Antikythera (From S1-ii, I understand that the time series of average AOD within each circle are very strongly correlated in ANT. So, the average within any given circle is close to the average within another one, most of the time.)
l.611 and Figure 2: the green arrow is not much better than the orange one: we cannot see in which direction it is pointing (Maybe because of image compression in the manuscript I received?). It doesn’t really matter, all the tracks pictured are descending anyway, this can be said in the text.
l. 663 “entire” should be replaced by “entirely”
l.721 “it is revealed” doesn’t sound idiomatic
l. 722-747 I wouldn’t use “underestimated” (l.723) or “negative biases” (l.730) which suggests a deficiency of the Aeolus instrument or processing. It is only a deficiency in the nomenclature, I agree that the variables in the L2A should probably be named “circular co-polar backscatter” but we should focus on comparing apples with apples.
Here, it is just a matter of presentation: I would explain the importance of the correction first, and then present the results.
l.816-819: “due to overlying noise (i.e. negative backscatter)”: I don’t agree with this explanation. If you mean that errors from the overlying layers are propagated downwards, this is wrong. Within any profile, backscatter retrievals of the SCA are independent from each other.
l.841-842: I do not see the “layer” between 5.5 and 8 km on Figure 3-iv. I think it is important to stress that we should not over-interpret single profiles. The noise is very large, and errors are underestimated.
l. 947: This mostly shows that thin range bins close to the ground are detrimental to optical properties retrieval because of the low SNR they induce (cause (ii) below). The signal from the surface should only leak into the first bin above the ground.
l. 964-981: How many profiles are used for comparison after filtering?
Figure 5 and this paragraph really underline the importance of proper cloud filtering. The results of figure 4 are only useful to show the contrast.
l. 1078-1080: Out of curiosity: how invalid can the underlying assumptions be? Is there a quantitative assessment somewhere? That would also be an important step for later studies, because few ground lidars actually have the circular polarization capability (could this assessment be a task for eVe?).
l. 1100-1102: this consideration was not discussed before and only appears here.
L. 1141: again this “underestimation” is more a problem of misinterpretation.
l. 1159: why not say: “Our analysis reveals that …”?
Figure 1: the legend doesn’t contain a description for sub-figure 1-v