Summary

This is an excellent paper that will contribute to the reduction of uncertainties in the radiative effects of dust due to insufficient knowledge of dust properties, an in particular the composition of the dust and its effect on optical properties. It highlights the importance of understanding the mineralogical content of dust, particularly the role of iron oxides like hematite, which affect the dust's optical properties. The study uses lidar measurements and atmospheric modeling to explore the relationship between hematite content and dust's optical properties, such as the lidar ratio and Ångström exponent. The findings suggest that while there is a positive correlation between hematite content and certain optical properties, the relationship is complex and influenced by particle size and composition. The study emphasizes the need for further research to better understand these interactions and improve the accuracy of dust's radiative effect estimates in climate models.

Methodology of Data Selection

The methodology for selecting data in the dust study is systematic and well-structured, focusing on ensuring that the data is relevant and reliable. Below we acknowledge some strengths and potential areas for improvement:.

Strengths:

Multi-step Approach: The use of a three-step process (AERONET data, PollyXT measurements, and COSMO-MUSCAT simulations) ensures a comprehensive selection of dust-dominated cases.

Quality Control: The focus on data from specific campaigns (JATAC) with rigorous quality control and cross-validation enhances the reliability of the data.

Seasonal Consideration: Selecting data from summer months when Saharan dust transport is most pronounced helps in capturing significant dust events.

Clear Criteria: The use of specific criteria (AOT and Ångström exponent) for filtering AERONET data ensures that only relevant dust events are considered.

Areas for improvement

Controlling for variations in intensive properties

The correlation between lidar parameters and hematite might be influenced by changes in the size distributions of the aerosol layers or the optical thickness of the layers considered for the study. Controlling for these by using layers of comparable optical depths and size distributions will eliminate any influence of the variations in these properties

Cloud Influence: While cloud screening is mentioned, the methodology could benefit from a more detailed description of how cloud interference is minimized or accounted for in the data analysis. In particular how does cloud contamination in the data manifest itself in the results.

Temporal and Spatial Resolution: The methodology could discuss the temporal and spatial resolution of the PollyXT and COSMO-MUSCAT data to ensure that the selected cases are representative of broader dust transport patterns.

Model Validation: While the COSMO-MUSCAT model is used to confirm dust layers, additional validation against independent datasets could strengthen the reliability of the model's outputs.

Potential Bias: The focus on specific periods and locations might introduce a bias. Expanding the study to include different times and regions could provide a more comprehensive understanding of dust transport and its properties.

Overall, the methodology is robust, but incorporating additional validation steps or at least discussing and acknowledging the potential bias and expanding the scope could enhance the study's comprehensiveness and applicability.

This paper is an attempt to relate several dust optical properties (lidar ratios, particle depolarization ratios, backscatter Angstrom Exponents, and extinction Angstrom Exponents) to the fraction of hematite in mineral aerosols. The authors use lidar measurements for the optical properties and the COSMO-MUSCAT model for the hematite fractions. Since the hematite fractions are not measured and modeling the mineralogy of aeolian dust is presently not robust, it is unsurprising that the correlations between the measurements and the model are rather poor. 

I am not sure why the authors think that the mass fraction of hematite has an effect on the scattering field of atmospheric dust, even if they had measured the mineralogy (instead of modeled it). Hematite fractions are quite low in mineral aerosols,  generally less than ~5% by mass. Thus, varying hematite from its minimal mass fraction (0%) to its maximal mass fraction (~5%) does not alter the real refractive index or the scattering field significantly. And indeed, that is how it turned out -- the correlations presented here are quite poor.

 

On the other hand, hematite and the other iron oxides are responsible for nearly all of the absorption in mineral dust, so one would expect some significant variability in the single-scatter albedo (SSA) or mass absorption efficiency (MAE) associated with the mass fraction of hematite. Unfortunately, the SSA and MAE are not discussed in this paper (probably because they are not available from lidar measurements). 

The POLIPHON method is layed out in 3 steps, but it lacks details and requires much hand waving. A brief review needs to be provided with enough details for the reader to understand what the authors are doing and the associated uncertainties. For instance, in Step 1 (line 254) the authors say "The initial step involves separating the particle backscatter coefficient based on the particle linear depolarization ratio." How? What depolarization thresholds are you using to separate fine and coarse (if that is how you are doing it), and how do you separate fine dust from fine non-dust? In step 2, lidar ratios are estimated based upon their probable origins (how determined? a model?). Step 3 is an extinction to mass conversion that is based upon AERONET climatologies, but again, no reported conversion factors nor any details of how they were obtained. Then the authors state that an "advantages of this method is that it does not require a dust particle shape model in the data analysis since it relies solely on the measured optical properties." This is not true, though, since AERONET needs to use the spheroid optical model and a density assumption to relate aerosol mass to extinction.

There are way too many references to figures in other work in this paper (twice on line 52, lines 125, 272, twice on line 281, lines 423, 426). This is laziness, in my opinion, and as a reader I do not want to search through a bunch of extra journal articles to read a paper -- a paper needs to stand on its own merits. 

To sum up, the authors don't provide any physical basis to explain why the hematite mass fraction should be related to the extinction, linear depolarization ratio, or Angstrom exponents. So this paper is really an epidemiology study. However, the optical parameters that the authors chose are not physically related to iron oxide content (and the authors have not tried to demonstrate this), so the epidemiology study did not show any skill between the parameters  (as expected). Thus, this paper is not suitable for publication, in my opinion.

If the authors truly want to demonstrate a relationship between iron oxide fractions and aerosol optical properties, they need to

1.) replace the model in this paper with measurements, and

2.) choose dust optical properties that are sensitive to absorption (e.g., SSA).  

Line items:

Line 69, Authors state:

"Given that previous studies demonstrated the impact of iron oxides on the extinction (absorption plus scattering) properties, particularly in the UV-VIS specturm, this project hypothesizes that this effect will manifest in the lidar measured intensive optical properties at 355 nm and 532 nm wavelengths."

What previous studies? If true, these need to be cited. Personally, I have never seen papers that link hematite content to extinction. Hematite is typically less than 5% of the total dust mass, so I would be surprised if the variation of hematite from 0-5% has a significant effect on the extinction.  

Line 83, Authors state

"Gonçalves Ageitos et al. (2023) found that GMINER dataset fairly reproduces iron oxide content for the Sahara Desert." Looking at Fig 11 in Goncalves (2023), I have to disagree. 

Lines 275-280:

A rather strange partitioning of fine-to-coarse dust mass concentrations is described:

fine/coarse > 11%

9% < fine/coarse < 11%

fine/coarse < 9%

Can the authors demonstrate that their fine/coarse partitioning is accurate enough that it makes sense to define a range of only 2% for the 'intermediate' fine/coarse dust mass concentrations?

Figure 7:

Figure 7 showes the correlation coefficient (r) and presumably the coefficient of determination (R^2) for each panel. How come r^2 is not equal to R^2?

Line 396-398, Authors state: 

"Despite thesmall sample size, particularly in cases with larger proportion of coarse particles (just six samples),  the strong correlation coefficient suggests a meaningful relationship between the hematite fraction and lidar ratio, at least so for the VIS portion of the spectrum at 532 nm."

A strong correlation coefficent does NOT necessarily suggest a meaningful relationship when the sample size is small.