Lin et al. investigated the growing role of secondary organic aerosol in PM2.5 pollution in Shijiazhuang by both field observations and modelling. They found that the contribution of SOA to the total OA exceeded that of POA during the winter haze in 2024 compared to 2014, and aqueous oxidation of POA might explain the increased SOA. The response of OA to emission reduction was also discussed. The results are interesting, and the manuscript is well-represented and organized. It is publishable after the following questions have been well addressed.

1. In lines 105-106, the 2024 emission data is constructed based on the MECI 2020 and the 2024 measurement, but it is not clear for this scale.
2. In line 160, the PM1 components in 2014, but PM2.5 in 2024, will cause uncertainties when comparing the growth role of SOA. A reasonable discussion should be added.
3. In Figures 1a and b, the composition of PM2.5 has changed greatly in 2024 compared to 2014. In particular, the fraction of nitrate exceeded that of sulfate. This means aerosol acidity and aerosol liquid water content (ALWC) might have greatly changed at different pollution levels, which should influence aerosol formation, including both inorganic and organic aerosols. Those factors should be accounted for when you discuss SOA formation.
4. In Figure 2, although the correlation between SOA and RH is meaningful to understand the role of aqueous reaction in SOA formation, it should be more reasonable to discuss it using the fraction of ALWC in PM2.5 because ALWC should be related to RH, temperature, and chemical composition of aerosol.
5. OA was reduced by 79% in 2024 compared to 2014, with a 78% reduction during the most polluted period. It is meaningful to separate the reduction of POA and SOA.
6. In lines 185-187, aqueous oxidation might promote SOA formation, while oxidation of SVOC and IVOCs might also be enhanced in 2024 because of increased oxidation capacity.
7. In Figure 3, the mean values of observed ncPOA and SOA concentrations should be shown for a better understanding of the performance of the model.
8. In lines 205-206, the data in Figure 3b can not support this contribution (80) of SOA to OA. Please clarify. And it can give the detailed data (such as a table) for the cities in Figures 3c and 3d to make it easier to understand.
9. In lines 206-210 and Figure 3, another concern is that the contributions of biogenic sources (such as isoprene and pinenes) and aliphatics to SOA are not considered, which will introduce some uncertainties to the aging of POA.
10. Why does the contribution of local emissions of ncPOA in SJZ increase greatly in 2024 compared to 2014?
11. In lines 227-235, Figure S4 can not support this expression and conclusion. And the scenario setting in Section 3.3 is confusing, and it should be clarified. And the legend in Figure 3c should be checked. I think the difference shows the emission rather than meteorology.
12. In Figures S3 and S4, the performance of SOA modelling is terrible, particularly in 2014. The diurnal curve can not be correctly captured by the model. So, how are you sure about the SOA formation from aqueous oxidation of POA?
13. Figure S10 is not cited.
14. The formatting problems should be checked and revised, such as the upper and lower marks (lines 84, 130-134, etc.), the fonts (lines 37-45, 62-70), and reference information, like journal, volume, pages, and doi (lines 343-345, 366-369, etc).

This manuscript investigated the differences in aerosol mass concentration and chemical composition between 2014 and 2024 for a large city in the North China Plain. The two campaigns were performed using a Q-ACSM and SP-AMS, respectively. The major finding was that with large reductions in primary emissions, the contribution of SOA to OA increased significantly, from 27% in 2014 to 87% in 2024. The results contribute to the understanding of long-term variations in ambient aerosols in China. I think this manuscript could be considered for publication if the following concerns could be properly addressed.

First, the 2014 and 2024 measurements were conducted at different sites (i.e., urban and suburban, respectively), indicating that results from the two campaigns were not directly comparable. For example, it is not surprising that the SOA contribution was higher for the suburban site (i.e., the 2024 campaign), even if the anthropogenic emissions were unchanged. To explain the observed variations of SOA, the influences of different factors (e.g., primary emissions and the types of measurement site) should be properly distinguished. This is essential for an ACP paper.

Second, the CMAQ performance was questionable. As indicated by Figure S3, the observed and simulated SOA results in fact had little correlation. I don’t think this level of agreement could properly support the related discussions in the main manuscript.

Third, the comparison between CMAQ and AMS (or ACSM) results must also be performed for POA. A reasonable agreement is the precondition for the related discussions in section 3.2.