Global 3-D distribution of aerosol composition by synergistic use of CALIOP and MODIS observations

This is a very interesting paper utilizing state of the art satellite data from active and passive remote sensing sensors together with models in order to characterize aerosol optical properties globally, by clustering to aerosol types and relevant aerosol properties.

The novelty and the strength of the paper is the use of the less uncertain inputs from both satellite sensors and combine it in a model run, globally.

Introduction

I miss the state of the art on aerosol global climatologies based on Kinne et al and other AEROCOM related publications.Also, Amiridis et al., for calipso.  In addition, publications discussing dust only ODs based on MODIS-CALIPSO-MERRA2 synergies have been presented by Gkikas et al..

In addition, the distinction of the aerosol types with mixing possibilities in the atmospheric is a basic assumption of the paper and it have to be accompanied by studies elaborating on different approaches of  aerosol typing definition efforts.

Schematic of the retrieval.

Looking at the figure 1 scheme. I was wondering how the optimized x step is achieved, only for part of the aerosol properties or satellite based observations used for the matching at the convergence stage. Or some more clarity needed on the paragraph lines 135 to 144.

Section 3.1.3 It is necessary to introduce a number of assumptions here, so the authors to my opinion have done a good work. A discussion on overall uncertainties of the method would be nice for the reader. Realistically these retrievals and assumptions work much better in different parts of the world and worse in others based on the aerosol field complexity. Could the authors comment on such aspects ?

For example standard deviations in figure 6 I presume, is a mix of “easier” retrievals spatially and more difficult ones that cause these standard deviations. Final effect will be lonked with more uncertain retrievals in some areas and less in others.

Figure 8. MODIS standard AOD have been used in various studies and has been extensively validated with AERONET data. (Moreover, MODIS itself use AERONET to retrieve (some kind of) uncertainty estimation over land and ocean). What is the novelty here with the use of Callipso in AOD only ? Is lower than MODIS standard global AOD more realistic ? And what improvements and errors are dealt here with the combined MODIS-Callipso retreival ?

Figure 9 shows a very limited spatial varability of both parameters. First of all should be nice to increase the size and improve the quality of this figure as details can be already there but not visible.

In general it would be nice to comment on difference the authors find compared with the Kinne aerosol climatologies. Discussion could be combined with figure 13 results.

Figure 10 is the paper highlight and it needs technical improvements in order to be able to see spatial details and changes of the AODs.

Figure 14 is a very nice demonstration of aerosol shape effects.

I would see fig 14 to 18 in a supplement. But it is up to the authors to decide.

Fig. 19 should be discussed much more as properties like SSA are ery uncertain based on the figures 9 and 13.

Major comment

How this method improve compared with other existing ones and what are the advantages that lead to it ?

Kinne, S.: The MACv2 Aerosol Climatology, Tellus B, 71, 1–21, 2019. 

Kinne, S., Schulz, M., Textor, C., Guibert, S., Balkanski, Y., Bauer, S. E., Berntsen, T., Berglen, T. F., Boucher, O., Chin, M., Collins, W., Dentener, F., Diehl, T., Easter, R., Feichter, J., Fillmore, D., Ghan, S., Ginoux, P., Gong, S., Grini, A., Hendricks, J., Herzog, M., Horowitz, L., Isaksen, I., Iversen, T., Kirkevåg, A., Kloster, S., Koch, D., Kristjansson, J. E., Krol, M., Lauer, A., Lamarque, J. F., Lesins, G., Liu, X., Lohmann, U., Montanaro, V., Myhre, G., Penner, J., Pitari, G., Reddy, S., Seland, O., Stier, P., Takemura, T., and Tie, X.: An AeroCom initial assessment – optical properties in aerosol component modules of global models, Atmos. Chem. Phys., 6, 1815–1834, https://doi.org/10.5194/acp-6-1815-2006, 2006. 

Kinne, S., O'Donnel, D., Stier, P., Kloster, S., Zhang, Z., Schmidt, H., Rast, S., Giorgetta, M., Eck, T., and Stevens, B.: MAC-v1: A new global aerosol climatology for climate studies, J. Adv. Model. Earth Sy., 5, 704–740, 2013. 

Gkikas, A., Proestakis, E., Amiridis, V., Kazadzis, S., Di Tomaso, E., Tsekeri, A., Marinou, E., Hatzianastassiou, N., and Pérez García-Pando, C.: ModIs Dust AeroSol (MIDAS): a global fine-resolution dust optical depth data set, Atmos. Meas. Tech., 14, 309–334, https://doi.org/10.5194/amt-14-309-2021, 2021.

Amiridis, V., Marinou, E., Tsekeri, A., Wandinger, U., Schwarz, A., Giannakaki, E., Mamouri, R., Kokkalis, P., Binietoglou, I., Solomos, S., Herekakis, T., Kazadzis, S., Gerasopoulos, E., Proestakis, E., Kottas, M., Balis, D., Papayannis, A., Kontoes, C., Kourtidis, K., Papagiannopoulos, N., Mona, L., Pappalardo, G., Le Rille, O., and Ansmann, A.: LIVAS: a 3-D multi-wavelength aerosol/cloud database based on CALIPSO and EARLINET, Atmos. Chem. Phys., 15, 7127–7153, https://doi.org/10.5194/acp-15-7127-2015, 2015.

A nice structured work, developed by the utilization of CALIOP and MODIS retrievals for the establishment of a global aerosol-speciated 3D distribution. Typical aerosol properties are derived and collocated against ground-based stations (AERONET). Finally, SDRF values (under clear sky conditions) are retrieved and compared against results in previous studies for the estimation of aerosol induced perturbations on the Earth-Atmosphere radiation budget.

1 Introduction

I think the revised V4 types of CALIPSO and some weaknesses of CALIOP and MODIS retrievals - not only the limited wavelength information and the strong surface reflectance, respectively - should be mentioned (these preferences would probably have a reasonable contribution to the uncertainty in some CALIOP-MODIS retrievals).

 5 Retrieval results from the CALIOP and MODIS observations in 2010

In Figure 8 the different strong aerosol sources (e.g. dust source in the region of Bodélé) are not visible. For example, a well-known problem of CALIOP-CALIPSO retrievals is the sufficient underestimation of AOD over strong aerosol sources, an inadequacy strongly related to the presence opaque layers completely attenuating the laser beam. Probably a colorbar with a lower AOD limit (less than 0.8) or with modified bins or just a different colorbar could help with the visualization of this result. If a filter is applied for the smoothness of the colors on the map, this filter maybe contaminates the AOD over the sources especially if the surrounding regions have substantially lower AOD.

In Figure 9 an aerosol-speciated distribution is not clear. It’s like having 2 groups of SSA values (land-ocean). A narrower colorbar (starting e.g. from 0.8) could help with the distinguishing of some areas. For example, over the Northern and the Central Africa a lower and a higher SSA value should be visible (dust and more absorbing particles-like smoke from biomass burning- respectively). The same problem is visible for AF.

In Figure 12 it’s not clear for me some hotspots of coarse DS particles over the Norway and Sweden

In Figure 13 AOD shows a good agreement with AERONET, but the other parameters rather deviate. In comparison with Figure 9 maybe the results for the other properties need further investigation, since these parameters are also used for the radiative simulations and furthermore for the heating rate.

General comment

This is an interesting paper which develops an optimal estimation aerosol retrieval using combined CALIOP and MODIS observations.  The retrieval attempts to retrieve effective aerosol size and optical properties for each of four aerosol types.  This goes far beyond the standard retrievals of CALIOP and MODIS, or any other satellite sensor, so it is good that comparisons with Aeronet retrievals are included to evaluate the performance of the retrieval.  The algorithm is described well and the authors do a good job of examining some of the uncertainties (particle model assumptions) but more details on uncertainties (Section 4) and a few other topics would be helpful. 

Specific comments

CALIOP Level 1B data is pre-processed by calculating running means using horizontal averaging over 10 km.  Are retrievals performed only on 10-km averages that do not contain clouds, or are cloudy profiles removed before averaging to 10 km?

The authors mention several times in Section 3 that the DVCs and DMRS are ‘optimized to all CALIOP and MODIS measurements’.  It is not clear to me what this means.  Are the retrieved parameters adjusted to minimize the merit function for each MODIS-CALIOP data pair, or is there some sort of global optimization which is performed?

What is the altitude range of the CALIOP-MODIS retrieval?  It appears to be 0-10 km from Figure 8b.

I’m not sure how to read Table 3.  What is meant by “relative value”?  Relative to what?  In the first row, is the mean difference between retrieved and simulated values an AOD of -0.15 or is it 15% of something?

Lines 284-285: How does it follow that the AOD of WS is greater than the AOD of LA because the SSA of LA is lower?  The scattering extinction coefficient is reduced when SSA decreases, but the total extinction is unchanged.  Please explain.

To understand the realism of the retrieval simulations more detail should be added on how the satellite Level 1 data was simulated in Section 4.  From Line 256, it appears that a single number is used for random error in lidar backscatter (15%).  But the relative random error of CALIOP attenuated backscatter profiles varies with altitude and with the albedo of the underlying surface and can be much worse than 15%, especially at higher altitudes.  Was noise from the solar background simulated and added as a random variable to each sample in the vertical profile?  Were retrieval errors due to systematic MODIS calibration errors or estimates of surface albedo considered?

It is odd that retrieval uncertainties are larger over ocean than over land, while CALIOP and MODIS retrievals are both better over ocean than  land.  Is this really because SS is retrieved over ocean but not land, as the authors say, or could it be because AOD over ocean tends to be  much smaller than over land? Or is it due to uncertainties in the optical model used for SS?  It would be good to discuss reasons for this behavior in more detail.  Marine aerosol is not just ‘sea salt’ and often contains internally mixed biogenic sulfate or biogenic organic compounds.  This might impact the refractive index of the particle model used.

The authors comment that extinction coefficients are unnaturally large at 70N and 70S-80S.  I do not see evidence of this in figure 8a or 8b and am wondering what the authors are referring to.  There appear to be very few retrievals at 70S-80S.  The text says the large EC are due to cloud contamination, but could it be due to ice cover and thus high surface albedo?  Are retrievals attempted over ice or only over ice-free ocean? 

Minor comments

It was not clear to me what is meant by ‘dry volume concentration’ (line 125).  What are the units?

Equations 10 and 13 explain constraints applied to the solution, using somewhat different approaches to notation.  I find the approach used in Eqn 13 to be more clear than Eqn 10.

Lines 422-423: rather than “SSA of the land ..”, I think  “SSA over land ..” is meant, and the same for “of the ocean” and for AF

The authors introduce a large number of non-standard 2- and 3-letter abbreviations for various parameters (ABC, LR, DMR, …) , and then later introduce mathematical symbols for some of these parameters when used in equations.  It would be simpler to define the math symbols and use them throughout the paper.  I found DMR and DVC especially awkward and had a much easier time reading r_e and v_dry . 

Depolarization ratio (DR) and linear depolarization ratio (LDR) are both used.  Aren’t these the same parameter?