Dear Authors,

Thank you for your well-written and interesting manuscript. This topic of aerosol-aware DA is a frontier of development in data assimilation and this paper represents an important milestone in that development.

I have minor recommendations for clarity and completeness to make this manuscript more helpful to readers and eliminate some ambiguity in your descriptions. I did not identify any technical changes needed in the work you describe.

• I recognize that CRTM is well-described in other publications, but I think it is still appropriate to add some more text describing how the aerosol and RT modules interact. CRTM is a “fast radiative transfer model,” which means that certain computations are done ahead using a more complete analytical model, with solutions stored in lookup tables used by CRTM. What aspects of the RT are solved via lookup tables and which ones are solved analytically is important information for understanding how this system reacts to the additional information about scattering and absorbing particles.
• Line 58-60: aerosol-aware DA is complicated to describe, because the model prior has high and low biases without aerosol information and different high and low biases when aerosol information is included. This part needs to be looked over carefully, especially the part about “leads to a warmer atmospheric analysis” needs a more complete description that reduces ambiguity about the referenced experiment and results.
• Figure 2 and discussion in Lines 217-222: This part of the description seemed rushed and incomplete. There is plenty of room to expand this figure to show, for instance, the relationship between dust loading and differences in innovations.
• Line 70: “In section 2.3,… given here.” => “In section 2.3,… is given.”
• Line 96: “may degrade the data usage” It is not clear what this means.
• Line 140: “optical properties… are”
• Line 292: “include, but not” => “include, but are not”

This is a well-written manuscript on the important topic of assimilation of aerosol-affected infrared radiances.  The manuscript describes key model developments and highlights the ability of aerosol-aware radiance assimilation to improve forecast performance.  I recommend minor revisions to improve some of the analysis and discussion points in the manuscript. 

Science questions/issues

Line 112: “The optical tables from other aerosol models are not finalized yet”; It would benefit readers to know what other optical tables are in development. Can you mention at least one of them here?

Line 195-198:  According to Gelaro et al. (2017), MERRA-2 includes infrared radiance assimilation of IASI in an aerosol-blind configuration.  As a result, should we expect any significant differences in the meteorological fields (e.g., temperature) between MERRA-2 and the baseline GSI experiment?  Can you provide information on the possible magnitude of systematic biases in dusty regions of the MERRA-2 reanalyses?

Lines 199 – 206: More explanation on Fig. 1b is recommended. This map shows dust dominating in almost all areas of IASI coverage.  I understand this is the active dust season with long-range transport, but could MERRA-2 be overestimating the global dust coverage, which would impact the results in Fig. 1a?  For instance, there are some areas of carbon aerosols mostly over eastern Siberia, but I would expect a more extensive area of carbon dominated aerosols across Siberia. 

Lines 228 – 232: A more detailed explanation of Fig. 3 is strongly recommended.  Additional panels showing dust column mass density (in addition to total aerosol column density) and dust effective radius would be helpful to the analysis and readers.  How are dust sizes varying in the dust contaminated areas and influencing the AER-CTL TEMP differences?  Dust size should be analyzed and discussed here, as it is an important factor when accounting for aerosol-affected satellite radiances.  Also, why are the much cooler BTs in the AER experiment over western Africa (Fig. 2a) leading to only a minimal warming signal at 900 hPa? I was expecting to see significant warmer temperatures in AER compared to CTL at 900 hPa in this area.  Less important, what is causing the significant AER-CTL temperature differences over the Southern Ocean and Antarctica?

Line 242: What are the specifics of this fully cycled experiment (e.g., cycling frequency and assimilation window)?

Line 243: “aerosol-affected satellite radiances are taken into account”; Were infrared radiances from all satellites in GDAS considered?  A short list of some key satellites considered in these experiments would be helpful.

Line 263: Can you explain the poorer results over the Southern Hemisphere? Simply due to less aerosol loading?

Lines 263 – 265: A map of mean total aerosol column mass (or for the different aerosol species) for the period of interest or perhaps a table of aerosol column masses for the different regions would be helpful here.

Technical corrections

Line 232: missing “K” after “0.5° to 1°”

Line 262 – 263: “The RMSE scorecards … while neutral or degradation over the Southern Hemisphere (20° S – 80° S)”, sentence structure needs improvement

Line 270, Figure 5 caption: replace “means” with “mean”