Kong et al. present a refined model for simulating the atmospheric transport of dust. The authors new model is an amalgam of two previous iterations (Kong et al. 2021; Liu et al. 2021), finding that the combination of the previous results produces a superior model compared to either of the previous studies alone. Considering the noted improvements, and the possibility of increased extreme weather events in the future (a point that the authors may consider adding as an implication), the submitted manuscript should eventually be published in ACP. However, I do have some minor comments below that should be addressed first.

General: The figures need to be reordered, as the first figure referred to in the text is Figure 2 (Page 7 Line 172).

General: Figure captions generally lack sufficient detail and definitions.

General: Can the authors comment on potential interfering species to PM10 concentrations, and how that subsequently affects the resultant simulation (e.g., biological particles)?

Page 7 Lines 172 – 182: I’m not convinced that the interpretation of the data from the Wanli station is appropriate given the signal saturation that is clearly observed for the PM10 concentrations. Even with the caveats that the authors provide, providing a numerical evaluation of their model (i.e., percent error) compared to the experimental Wanli data is highly misleading. At the very least, I recommend removing any numerical comparison with this result.  

Figure 3: While providing the differences between ‘Dust_Off’ and other various simulations makes for informative visualizations, it may still be helpful to show the absolute representations (i.e., separate simulations for Dust_Off, Dust_Default, Dust_Refined_1, Dust_Refined_2, and Dust_Refined_3) in the supporting information.

Page 8 Line 196 – 197: The authors have (perhaps unintentionally) opened up Pandora’s box here: if the MERRA-2 reanalysis is subject to error due to a poorly captured wind component (something the authors postulate as a reason for differences in their simulation over East Asia), should it even then be used to validate their more positive results later on? I pose this as a bit of a rhetorical question. However, the very nature of events that the authors are interrogating surround the issue of wind, and if the method used to establish ground truth (MERRA-2) is subject to wind-related errors, then this should be interrogated further (i.e., to what extent is the use of MERRA-2 as a ground truth appropriate?).

Page 8 Line 205: Change ‘Dust_OFF’ to ‘Dust_Off’

Figure 2: Define P1, P2a, and P2b

Figure 4: While the contours may be helpful, the authors more frequently refer to various land masses instead to describe the movement of the dust plume. Therefore, consider replacing the background graphic with one in which the visualization of land masses is more clear.

Figures 4 and 5: Consider combining these figures together, both for the sake of simplified visualization and simplified references within the text. For example, the prose on page 9 seems to present these figures as two distinct talking points despite the high degree of similarity in the provided information.

Figure 5: There are various green and red arrows on the different tiles. The significance of these need to be described via a legend and within the figure caption.

Figure 6: Both axes needs to be defined (i.e., need labels), and the x-axis readability needs to be improved.

General comments:

The authors improved the dust simulation over East Asia based on a few extreme dust events by refining soil moisture factor, dust emission speciation profiles, and bulk soil density. They also examined dust removal by a typhoon. The particulate matter from model simulations were evaluated against observations, which confirmed an improvement. Overall, the paper is well written but require some minor edits before publication in ACP

Specific comments:

Line 349-352: While the study showed an improved NMB with refined dust simulations, the authors should explain the possible causes of the remaining biases and offer suggestions on methods that will further reduce the NMB in future studies. After all, the statistics without refinement still satisfy the benchmark.

Line 589 – 590: How are the thresholds in the benchmark defined? A general statistical benchmark may not be appropriate for dust model evaluations. Even the default simulations satisfy the benchmark, which makes the refined simulation in this study less significant.

Technical comments:

Line 15: Do you mean “dust radiative flux”?

Line 22: It should be either “higher” or “lower in magnitude” since they are negative values

Line 27: “On 15-21 April 2021” => “During 15-21 April 2021”

Line 28: “Typhoon” should be capitalized

Line 586: “bulb” => “bulk”

Line 603: “dot representing” => “dots represent”

Line 645 - 646: The caption needs to mention PM10 concentrations. The same applied to Figure 5.

Supplementary Figure S2: “3-days” => “three-day”. Which figures are based on MODIS? Please be more specific.