Review of the paper “Where does the dust deposited over the Sierra Nevada snow come from?” by Huang et al., submitted for publication in ACP.

General comments

The paper aim at characterising the circulation types associated with dust deposition over the Sierra Nevada snow cover, by using SOM based clustering. The paper is well written, in a clear and concise manner. Motivations and objectives are clearly declared and put in the context of the existing knowledge. Methods are appropriated and clearly described. Conclusions follow from evidence. The findings discussed in the paper are relevant, and worth of publication. However, I believe that the paper could be further improved with a few modifications.

My first and main suggestion is motivated by the title of the paper, which questions the origin of dust deposition over the Sierra Nevada. I wonder why the authors didn’t corroborate their findings by adding a back trajectories analysis. I really believe that the robustness of the results would significantly benefit from such an analysis, especially the part concerning the long-range transport.

I also wonder why the authors only analysed one year. Possibly because of limited computational resources. However, reliable reanalysis products are available (MERRA, CAMS) for analysis on the long term (even longer than the 11-year climatology presented in the paper), which could then be used to select one or more interesting years for more WRF simulations.

Specific comments

L36: what is “the dust prone area”? Can you be more specific?

L39: this sentence is not clear, premature mortality is rather a consequence of air-quality associated diseases (which include cardiovascular illnesses) than a direct impact. Please rephrase to highlight direct and indirect impacts of dust on health.

L183: too generic, please clarify what you mean by “projecting high-dimensional data into a two-dimensional grid”.

L198: are these model levels?

L206: did you select the number of modes subjectively by testing the method, or did you use any objective methodology to assess distinctiveness and robustness? The approach used should be clarified. Also, how many EOFs have been retained before clustering? On which spatial domain the SOM method has been applied?

L210: remapping to a much finer resolution is not recommended, one cannot “create” the physical information not included in the coarser data. I suggest to remap WRF data to MERRA, and then validate.

L222: this is not evident from Fig. 2, can you provide a ref?

Fig. 2: please provide uniform colorbars, for better comparison.

Sec. 3.2: it would also be interesting to analyse the preferred transitions of circulation types along with persistence.

L274: how can we appreciate the dust deposition in Fig. 3?

L276: Fig. 6 is discussed before Fig. 5. Please change the order of figures (it also makes more sense to discuss local scale before teleconnections).

Fig. 5: could you add some reanalysis aerosol product to display possible transport paths across the Pacific?

L312: please provide a ref.

L314: how can we appreciate the dust deposition in Fig. 3c? (see also the comment to L274)

L342: showing some reanalysis aerosol product could also be helpful in visualising the transport pattern.

Fig. 11 and 12: please use the same colorbar as Fig. 3, for better comparison. In Fig. 11, the relative importance of the Mojave desert looks reduced in MERRA, any comment?

L465: can you please show and comment the interannual variability of the occurrence of the SOM clusters?

Technical corrections

L24: “We find that dust…”

L26: “mostly in winter and…”

L28: “in the presence of the NPH”.

This study integrates the WRF-Chem simulations with satellite data and reanalysis products in sppring to understand: (a) where does the dust deposition on the Sierra Nevada come from? (b) how is the dust transported to the mountainous regions? and (c) how is the dust deposited on the Sierra Nevada? The self-organizing map (SOM) analysis is applied to identify four major clusters of dust sources. Dust deposition onto the Sierra Nevada is believed to be an important factor that will affect the snow melt in the region and influence the water resource in the American West. Results from this study are an important contribution to the study of dust-cryosphere interactions. The paper is generally well written. I recommend the paper be published in ACP after addressing some minor issues below. 

1. in the abstract, it is necessary to say which data have been used in this study.

2. lines 116-118: please explain why you didn't use outputs from full simulations (September 20, 2018 to August 2021)? 

3. lines 137-140: How does WRF-Chem define PM10? Is it defined with geometric size or aerodynamic size? 

4. lines 230-232 (and throughout the paper): please consider to use alternative words for "underestimates" and "underestimation".  It is kind of weired when one says that observations "underestimate" something with a reference to model simulations.

5. Figure 2: Given that (c) and (d) are for dust optical depth, I would suggest that they  also show dust optical depth ion (a) and (b). Also there is clear difference between MERRA-2 and MIDAS in terms of spatial pattern. Please explain the difference. 

6. Figure 3 caption: "Figure 5" should be "Figure 6", I guess.

7. Figure 10: These violin plots are not straightforward or even confusing. It was very hard for me to figure it out and understand. Is it possible to change it to a more conventional PDF line plots?