Review of “Comparison of diurnal aerosol products retrieved from combinations of micro-pulse lidar and sun-photometer over KAUST observation site” by Lopatin et al.

This paper compares several lidar-photometer retrievals at a site near the Red Sea on the Arabian Peninsula. The input dataset are the MPLNET backscatter lidar measurements, the AERONET aerosol optical depths, and sometimes the AERONET almucantar radiances. The authors use GRASP with and without a volume depolarization ratio constraint applied to MPLNET measurements and the MPLNET retrieval products. The products that they compare are the aerosol AODs, extinctions, and column lidar ratios. They also compare the lidar-photometer lidar ratio retrievals to the AERONET lidar ratio retrieval products. The authors did a great job of laying out the ‘nuts and bolts’ of the retrievals and data descriptions in Sections 1 & 2. I think that the authors pulled together a very solid paper and I find it suitable for publication.

However, I struggled with many of the scatter plots in Sections 4 & 5… the correlations were poor, and it is rather depressing to look at multiple shotgun patterns with lines drawn through them. None of the datasets can be considered as a benchmark for the truth, but it would be nice to know whether there are statistically robust differences between the different lidar ratios presented. The scatter plots only seem to indicate that both approaches are distributed about a median value. I realize that the authors show histograms below the scatterplots, but are the lidar ratio differences between the various techniques statistically significant?

Thus, I would like to suggest that the authors replace some of the scatterplots with boxplots (or mean & standard error plots), and re-tool the analysis towards a student’s T-test or similar. For instance, it would be rather easy to turn Figures 3&4 into a single figure with 7 notched box-and-whisker plots (one box each for MPL_LR, S1_LR, S2_LR, MPL_LR(AOD>0.2), S1_LR(AOD>0.2), S2_LR(AOD>0.2), and AERONET_LR. Each box is essentially a mini-histogram, so you’ll have a nice visual of 7 histograms right next to one another for easy comparison (or add Fig 11 to obtain 10 boxplots). Do the notches overlap, indicating statistical agreement? Or do the medians have large separations? Are all of the boxes about the same size (indicating similar spreads), or are some larger than others? I think that this would be a much more enjoyable and useful way to look at the data than scatter plots that are nearly spherical. This is not a requirement, but I think that you’ll retain more reader interest if you make this change. 

The authors do a nice job of showing day/night and seasonal variations of the complex refractive index in Fig 16, but why not do the same thing with lidar ratio? Some of their LR discussion already suggested that seasonal variation in the sea salt / dust partitioning was causing differences in the lidar ratio (e.g., line 389), so why not partition the data in that way?  That would strengthen your hypothesis. One could even repeat the boxplots that I describe above for different seasons to see if the boxes actually do move up in the dust season and down when marine aerosols have a stronger presence.

I am not a big fan of the Scenario 1 & Scenario 2 nomenclature, as it replaces something that has meaning (excluding and including volume depolarization ratio) with something else. At least consider labeling such as Scenario E and Scenario I, as that would be easier for the reader to track.

I don’t believe that I have ever seen steradians abbreviated as Sr… I’ve always seen sr.

Line 159:

This is a 2nd description of the KAUST site, similar to the paragraph on line 136. There is good info in both of these paragraphs, so they should be merged and located at the beginning of Section 2 (currently line 136).

Line 166:

Passive tense is ambiguous here and in several places in the upcoming paragraphs. Here, the data 'was processed' using GRASP software. WHO processed the data? Consider "we processed almost three consecutive years of data... "

Line 181

diluted?… or dissolved?

Line 205

copped?… do you mean capped?

Line 260

“…allowing the lidar signal to influence the photometric observations and vice versa”.

Do you mean ‘calculations’ or ‘computations’ instead of ‘observations’?…

The lidar signal won’t influence the photometric observations unless you point the MPLNET at the AERONET.

Line 323

I am pretty sure that you do not mean “…nighttime liar retrievals.”

Line 407

I don’t consider Angstrom Exponent as an ‘advanced aerosol product’ (at least for AERONET).

Line 412

Authors discuss potential issues associated with signal attenuation, but shouldn’t that be easy to test?… just filter our high AOD cases.

Line 513 (and elsewhere)

80.60% and 69.31%… that’s a lot of precision.  Why not round off to the nearest percentage?

Line 520

6000, not 6oo0

Figure 9

Light blue is difficult to see on white. Consider a grey background.

Table 3

Would be nice to see the avg values as well.

Fig 15

Some clarification about exactly what these mode fractions mean would be helpful. For example, should the winter nights from the three panels add to 100%? It doesn’t appear that way, so I am not quite sure of what these fractions mean.

Line 683

“It should be noted, that in autumn real part of refractive index has similar values,…”

needs clarification. Similar to what? Also, Autumn has stronger absorption than in Winter and similar to Spring and Summer? Is this the fine mode or coarse mode? Either way, it is difficult to reconcile this sentence with the RRI and IRI of Fig 16.