|I appreciated very much the authors’ efforts for this revision. However, I think further improvement is needed.|
The authors argued that the formation of summer haze in the NCP area of China was a result of hygroscopic effects of aerosols for high RH in summertime. This argument is acceptable in principle. However, my major concern is upon the weak evidences for the argument. As shown in Table 2, the correlation coefficient between observation and simulated PM2.5 was 0.43 for summer case in Beijing, which means that the mechanisms considered in the model explained only ~20% of the variance in the observed PM2.5 concentration. Moreover, as shown in Figure 3, the model significantly underestimated summer RH at Miyun (Beijing) station. While the simulation of both key parameters is subject to high uncertainties, it’s hard to draw a solid conclusion. I thought that the apparently fair R of 0.65 for visibility (Table 2) was a result of those extreme events with PM2.5 up to 0.2-0.3 mg/m3, where the impacts of uncertainty could be relatively small. Without a robust simulation and data analysis, the current results could not be extrapolated to cases of low PM2.5 loading as shown in Figure 15.
1. Model evaluation: The authors claimed that “The modeled temperature, relative humidity, and wind speed were in good agreement with the observations at nearly all stations” (Line 173-174). However, there is indeed obvious inconsistency in the results shown in Figure 3. For example, the model underestimated RH for the second half of Feb at Tanggu and Miyun, and for almost the whole July at Miyun. In addition, overestimation of RH is significant for the summer case at Baotou station. The simulation of RH is particularly important to this study as the summer haze is attributed to the hygroscopic effects of aerosols. I’m not meaning that the model should be perfect. Nevertheless, the differences in the simulated and observed RH are so substantial in the Beijing-Tianjin area and thereby could have induced significant uncertainties in the conclusions. I’d like to suggest the authors make an in-depth analysis on the relevant uncertainty that will get their arguments more convincing than the current ones.
2. Model evaluation: The authors indicated that “A persistent underestimation of wind speed by the models was found at the Wutaishan and Taishan sites” (Line 174-175). This statement disagrees with the corresponding plots (Figure 4), which show underestimation for the case of Wutaishan but “overestimation” for Taishan. Thus the relevant analysis and argument for the inconsistency in wind field need a substantial revision.
3. Line 205-206/Model evaluation: The large diurnal variation in NO2 was attributed to the “gas-phase chemical scheme in ISORROPIA”. As the ISORROPIA is dealing with the thermodynamics of aerosols and their precursors, are you meaning that the diurnal variation of NO2 is dominated by heterogeneous reactions? This could be an interesting finding because the diurnal variation of NO2 was usually controlled by deposition flux and gas phase chemistry in urban areas.
4. Sec4.1: The main argument here was that comparable degradation in visibility corresponded to less PM in summertime than in winter case because of high RH in summer. In this context, why not showed the correlation between mass- specific extinction efficiency (cross section) and the RH directly? A plot showing the extinction cross section vs. RH will depict the argument clearly.
5. Sec 4.2: Line267-269 indicated that nitrate was the main particulate pollutant during winter. This was based on simulation I think. In fact, the observation (Figure 10) showed that OC was comparable with nitrate. Thus the total mass of organic matters (OM) would be larger than nitrate.
6. Line 294-296 indicated that except nitrate the contribution of each aerosol component to the extinction “did not significantly change”. However, as shown in Figure 12 the contribution of sulfate exhibited obvious diurnal variation. Moreover, sulfate behaved very much like a counterpart of nitrate and the total contribution of sulfate+nitrate was leveled off, particularly in summertime. As the mechanisms dominating sulfate and nitrate aerosols are very different, I wonder how the two species can compensate each other so closely. It seems to me that the total contribution of S+N could be controlled by something in the model.
7. Sec 4.3: A nonlinear response in visibility to PM concentration was found from model simulation. I suggest make an in-depth analysis on the mechanisms for this result. This is particular for the drastic changes in visibility on July 25 (Figure 13), where the visibility increased from 20 to 70 km and decreased to ~40 km with only very minor changes in PM within 12 hours.
8. Finally, the threshold of PM concentration for haze was described as a function of RH. The information is useful in air quality management. However, this information must be provided with cautions. Assumptions and thereby uncertainties should be addressed very clearly.
9. Furthermore, I compared the two plots in Figure 15 and I think they are same. Actually, according to the parameters in Table 6 and the equation 2, the RH will increase with M and the threshold of PM will be zero with RH of 97% in summer. This is certainly disagreeing with the reality. So PLEASE CAREFULLY check all the calculation and presentation .