|I find that the authors have done a reasonably good job of addressing the reviewer concerns. However, I believe that a few issues must be fully addressed prior to publication. |
P16, L30: The authors should change this such that it does not say that the n and SSA values were larger than for period 4. I see that they say now “within uncertainties,” but this is not sufficient here given that (i) the SSA values are different outside of the standard deviations (not uncertainties, technically), (ii) the n values are identical within the standard deviations and (iii) the k values are actually smaller than those reported for Period 4. Thus, this remains incorrect. The authors must be more specific here regarding each term (SSA, n and k) that is being compared.
P14, L3: The authors have revised their statements regarding the increasing importance of “inorganic species” during haze periods. However, I still find this to be inconsistent with their measurements. They state “The significant increase in the concentration and fractional contribution to PM1.0 extinction coefficient of inorganic species, particularly nitrates (discussed in Section 4.3), indicate that inorganic species become more important during haze...” However, they observe that the fraction of PM1 that is sulfate actually decreases. Thus, this statement is too general. It is not “particularly nitrates”, but “specifically nitrates.” The authors should make the statement specific (and avoid using terms such as “significant” unless they mean it in a statistically significant manner): “The notable increase in the concentration and fractional contribution to PM1.0 extinction coefficient of nitrate (discussed in Section 4.3), indicate that nitrate becomes more important during haze…”
P21, L2: It is not clear what is meant by “improved mass extinction efficiencies.” Improved relative to what? Do they mean their “modified” (i.e. wavelength-adjusted) values?
P23, L15: The measurements do not support the notion that the “EC fraction decreased with increasing pollution level.” The values given are 4.3 +/- 3.6%, 4.0 +/- 2.0% and 3.6 +/- 1.1%. These are all identical within the large uncertainties (i.e. standard deviations). This should be revised, along with the relationship with the previous sentence.
P23, L20 onwards: I find this discussion to be unclear. The authors are comparing their measured MSE and MAE values with calculated values from the IMPROVE algorithm. This is fine, but also potentially not useful. In particular, the IMPROVE algorithm is a crude approximation to actual behavior (it assumes some aspect of the size distribution and treats size distributions in a binary manner of “small” and “large”). It can give an idea of what is going on, but quantitative comparison is not warranted unless the goal is to develop regionally-specific coefficients for use in the algorithm based on the observations (which is where the numbers in the algorithm come from in the first place). The authors don’t really do anything with the MSE comparison, besides note that a disparity exists. I find that discussion here is still lacking. Regarding the MAEs and the conclusions regarding “brown carbon,” the authors’ logic is now easier to follow. However, they should also consider other possibilities, such as that the absorption by EC might itself be “enhanced” due to coatings, especially since the wavelength considered here is not especially far into the blue. Regarding the Wang et al. reference, the authors should indicate the wavelength in that study.
Related to the above comment, the authors did not address the issue of what the MAE values are relative to EC specifically, which is a much better metric of comparison than the MAE for PM1. From the authors’ assumption, the MAE for EC in the IMPROVE algorithm equals the MEE for EC, and thus is equal to 10 m2/g. First, this assumption is not really correct as the SSA of EC is not zero (more commonly, values of ~0.2 are found, indicating that the MAE values are actually about only 80% of the MEE values. This uncertainty is not accounted for in Fig. 11 or the associated discussion.) I calculate from Table 1 EC-specific MAE values of 24.2, 32.0, 33.4 and 27.4 m2/g for the four periods (clear, slightly polluted, polluted and all). From the arguments presented by the authors, I would compare these values to 10 m2/g, meaning that the observed MAE values are 2.4, 3.2, 3.3 and 2.7 times as large as the IMPROVE value. This would imply then that if BrC were the reason for this disparity that it contributes about 60-70% of the total absorption. These fractions would be even larger if I assumed that the MAE for EC from IMPROVE should only be about 8 m2/g (instead of 10). This seems very large to me given the measurement wavelength (467 nm). But perhaps it is correct. Nonetheless, I strongly suggest that the authors present a more direct discussion of the EC-specific values (which I implied in my prior review, but perhaps I was not explicit enough). However, associated with this, I think the authors should provide some discussion as to the reasonableness of their observed EC-specific MAE values (which, again, to me seem very large but not out of the realm of possible). I think that a presentation of the EC-specific values would be much clearer than the discussion of the PM1 values as it focuses the discussion on the key measured absorbing component.