General comments:

The manuscript titled "The role of dust mineral composition in atmospheric radiation and pollution in North China: new insights from EMIT and two-way coupled modeling" presents a novel and comprehensive investigation of mineral dust impacts using multiple dust atlases and a two-way coupled WRF-CHIMERE model. The integration of EMIT satellite-derived data is particularly innovative and demonstrates significant potential for improving model accuracy in historical dust storm simulations and future forecasting works. Overall, the paper is clearly written and methodologically sound. However, if the following comments are thoroughly addressed within this review process I would suggest publishing this paper in ACP.

Major comments:

While the paper demonstrates the benefit of using EMIT data in methodology, it would be helpful to provide a quantitative assessment of uncertainties introduced by the interpolation and assumptions in EMIT data processing (e.g., feldspar/quartz filling).

The manuscript often mentions ACI (aerosol-cloud interaction), yet the modeling focuses on ARI only. Please clarify this distinction earlier in the Introduction and reduce any ambiguity about what has or has not been included.

The SSR and PM10 comparisons are robust, but more details on the performance metrics (bias, RMSE, etc.) across multiple sites and time periods would strengthen the validation claims.

The influence of mineralogy on PM10 and O3 is clearly demonstrated, but more discussion of the physical mechanisms (e.g., specific reactions, photolysis suppression) would help interpret the observed changes.

The results show that quartz and feldspar dominate dust mass, while hematite dominates radiative effects. This contrast deserves more discussion in both the Results and Conclusion sections.

The model bias discussion (Section 3.1) is helpful but could be deepened by exploring possible reasons for the underestimation of PM10 at high dust sites.

Minor comments:

Line 137: Please specify how missing EMIT data (quartz/feldspar) are estimated — a numeric assumption or spatial filling?

Line 187–198: The bias in SSR is discussed, but no mention is made of possible causes (e.g., aerosol loading or model radiation scheme limitations).

Line 194: The overestimation of SSR and WS10 could be more quantitatively discussed. Is this bias consistent with other dust studies in this region?

Line 213–214: “minimizing the negative biases in T2” — perhaps “reducing the magnitude of negative biases” is clearer.

Line 250: “Positive O3 biases increased” is unclear — do you mean O3 concentrations were overestimated?

Line 305: “−900 W m−2” seems unusually large for surface shortwave cooling. Please double-check this value.

Line 584: Suggest shortening this part of the conclusion and moving satellite technical details into Data/Methods.

Figure 1: Please include a scale bar and clear region names to help interpret mineral distributions.

Figure 2: Consider including error bars or confidence intervals for observed values, “Statatiscal metrices” → should be “Statistical metrics” in its caption.

Figure quality could be improved — e.g., Figures 2 and 7 would benefit from enhanced color contrast and labeled axes for clarity.

Reference format is mostly consistent, but some recent references (e.g., Panta et al., 2023) are missing DOIs.

This manuscript used a 2-way coupled WRF-CHIMERE model to investigate how different mineralogical compositions of dust affect aerosol-radiation and aerosol-cloud interactions (ARI/ACI) and their subsequent air quality outcomes. The model was run on a synoptic scale over North China during a major dust storm in March 2021. The authors observed that using EMIT to enhance the mineralogical details has improved the model predictions of PM₁₀, by revealing significant spatial differences in radioactive forcing and increased PM₁₀ levels in source regions. These results are critical for a better prediction during dust storm periods when the level of PM₁₀ is readily underestimated due to the lack of ARI/ACI consideration. Nevertheless, the manuscript could be improved by addressing the following concerns. I would suggest that paper be reconsidered after a major revision.

Major comments:

1. In multiple positions (lines 37-38, 44-45, 72-73), the authors declared the importance of ACI effects on the Earth’s energy balance and can also be altered by the difference in mineral compositions. However, this research also stated that they did not consider ACI effects in their 2-way coupled model but did not clearly explain why this is not included and how it will affect the final predictions. This affected the rationale of adopting this 2-way coupled model and a justification should be better provided.
2. In Section 2.1, The authors collected environmental data from various sources: meteorology data from CMA, PM concentrations from an online blog, and SSR data from a peer-reviewed paper. How do the authors ensure the fidelity of the data they obtained, and how do they maintain the integrity of them?
3. Lines 134-145, the authors mentioned a lack of feldspar and quartz and the combination of illite and muscovite in EMIT. Their proportions were all estimated based on N2012 or J2014 data. This is ambiguous since there are no details and rationale about which database was chosen. A sensitivity test is suggested to show how different methods of filling and splitting cause the change of results and how the actual methods are selected for each mineral component.
4. Lines 207-208, the authors mentioned a huge overestimation of SSR (>60%) from the model. This is an interesting finding, since this overestimation may lead to large errors in dust dispersion and hence change the PM prediction. An attribution of meteorological biases vs. mineralogical composition to the PM₁₀ prediction would help clarify the conclusions of mineralogical effects. Comparing the simulation with the bias and after correcting the bias may provide insights into how much the actual ARI effect accounts for.
5. Lines 446-453, this paragraph is not well discussed. By comparing the subfigures in Fig 8, we can see that the PM₁₀ levels predicted using different database show substantial disparities. Suggest reducing the tone of the limited effect of mineral composition to PM₁₀ concentration.
6. Overall, this research did not include uncertainties in many of their reported values, such as predicted PM₁₀ levels, changes in PM₁₀ and ozone by including ARI effects, and different PM₁₀ concentrations considering dust mineralogy atlases. It is important to quantify these uncertainties for a study with an improved modeling design, thus statistical measures are suggested to include.

Minor comments:

1. Table 1, the last mineral was written as mica while the note mentioned it as muscovite. Although it is known that muscovite is a mica, it is suggested to make their name consistent to avoid confusion.
2. Table 3, the digits of values should be uniform: suggest to retain a digit of 2.
3. Figure 7, the picture for ozone is dazzling. The max/min/mean values (top left corner) cannot be seen clearly because of the color. Suggest reducing the hue of palette and move the text or change the color of the text.
4. Figure 8, N2012_EMIT-bulk dust is exactly the same with J2014_default-Bulk dust. Suggest recheck the pictures.
5. Figure 9, the texts in the violet region of quartz are not visible. Suggest changing its color to white. Also, the bars outside the pie look protruding; suggest using arrows to denote minimal values.