|This is a revised manuscript that provides some improvements and re-focus from the first version. To me, the paper conveys two aspects: 1) assimilation of satellite AOD can bring a better quality of model short-term forecast of AOD, and the better AOD data quality help the better model forecast; 2) on the other hand, assimilation of satellite AOD cannot provide any noticeable improvements of surface PM10, which is a key component of dust related air quality. However, the authors prefer to not discuss much of part 2) but rather to emphasize the positive impact of AOD forecast from data assimilation. Frankly I don’t agree with that approach, because forecast of dust-dominated PM10 is an important focus of SDS, and objectively assessing the value of AOD assimilation for dust forecast in both AOD and PM10 will help the community understand the possibilities and limitations in order to move forward. In the authors’ response, it is said the AOD-PM10 issue “would warrant a more in-depth study” that is beyond the scope of the current paper. I agree that more in-depth study could be deferred to a separate paper, but the current paper should point out the issues and make suggestions on further studies. I strongly encourage the authors to consider such approach to make the paper much stronger, more interesting and worthwhile.|
A few specific comments are listed below, mostly regarding section 4.
Page 3, line 25: Change “assess the performance of the assimilation” to “assess the value of AOD assimilation in surface PM forecast”.
Page 3, line 33: add a sentence on nitrate aerosol not included in the model.
Page 7, Figure 3: explain “FC” and “OB” in the figure caption.
Page 7, equations (1) and (2): what is (are) the reference(s) for these two equations?
Page 8, line 1: delete the comma after “Figure 4”.
Page 9, line 4: change “7” to “Figure 7”.
Page 11-12, section 4: This section needs to be enhanced, see my comments in the first paragraph. More specifically:
Page 12, line 1: “the timing is slightly wrong”: How wrong is slightly wrong? A few hours or a few minutes too early or too late? Needs to be more quantified.
Page 12, line 2: “the values of PM10 in the model are too low”: they are too low for the peak values but are much too high before and after the March 9 front passage.
Page 12, line 6-8: “the skill of the global model for PM10 is not as good as for AOD”: Can you offer some quantitative measure, such as FB, FGE, correlation, etc., that can support the statement?
Page 12, line 8-12: the value of PM10 forecast with and without satellite AOD assimilation should be quantified. “perform marginally better”, “positive impact” are not quantitative measure. The statistical measures, such as bias, root-mean-square-error, correlation coefficients, etc., for model calculation alone, with DT assimilation, and with DTDB assimilation should be provided to indicate the degree of “marginally better” and “positive impact” of including AOD assimilation. From Figure 10, The difference among the three lines seems in orders of magnitudes smaller than the differences between model and observations, so the marginal improvements and positive impact do not help improve the PM10 forecast at all.
Page 12, Figure 10 b): It would be helpful to indicate the latitude and longitude range for readers to grasp the horizontal domain size.
Page 14, line 3: “the spatial patterns are well captured”: Where are the spatial patterns shown?
Page 14, line 6-8: why do "the model biases, coarse resolution, lack of resolved local emissions, and lack of observations of speciation" only affect quality of PM10 but not AOD?