Second review of “Contribution of local and remote anthropogenic aerosols to a record-breaking torrential rainfall event in Guangdong province, China” by Liu et al.
1. The authors missed the point of my first major comment. My comment is about the regions with opposite precipitation response, which could be corresponding to the cold and warm sectors of the convective system, respectively. In their response, the authors chose two narrow areas (Box_N and Box_S), both of which have increased precipitation, to show the consistent features between those two boxes. Both areas are in the convergence zone based on Figure R1 and their cloud properties are of course similar. The cold section is probably northwest or northeast where decreased precipitation is seen (can be identified based on temperature field). Clouds at the cold sector would not be invigorated by aerosols so decreased precipitation can be seen as a result of suppressed conversion into rain or snow. Again, the point is that the authors need to explain the opposite precipitation response for different sections of the system, particularly for the 10Xrun, the decrease of precipitation is in a similar magnitude with the increase and occupies half of the simulation domain (Figure 19b). Based on Figure 19b, there is really no justification of only picking up the red box region to study.
2. The authors did not do a neat job in responses. Many responses have wrong line numbers and they also did not describe what changes they made (also did not copy the revised text to the responses), which made me have a hard time to check their changes.
3. There are quite a bit misunderstandings of cloud microphysical processes by reading the responses only (since I had a difficulty to find the changes in the manuscript due to incorrect line numbers). Here are examples, (1) a mistake in calculating cloud droplet number concentrations. They got unreasonably high (8e4 cm-3) cloud droplets (particularly for area mean, not a maximum value at gird-level) by using water density instead of air density to convert to number concentrations. What’s surprising me is that they still argue the reasonability of it. Such a high number concentration is only possible for aerosols (not droplets) in a very polluted condition. (2) the misunderstandings of BF process, latent heat, and precipitating particles (see my comments on #14 response below). (3) the primary driver of convergence (my comments on #16 response). All these aspects that they misunderstood are the key aspects for analyzing and interpreting the model results this study.
4. For many comments on clarifications, the authors responded but did not clarify in the manuscript, such as comment #4,
5. The writing is a little sloppy. There are typos and many statements are confusing. Here are a few just in a short abstract:
Abstract:
Line 28, “cloud property changes also resembled that in the control run” does not make sense. Changes means the differences between the 10xrun and control run, how can the changes resemble control run?
Line 29, “The precipitation average over Guangdong province decreased by 1.0 mm but increased by 1.4 mm in the control run” does not make sense either. Looks like you are describing an increase or decrease in the control run. Then what are you comparing with? Generally, the description should be the increase or decrease by comparing with the control run.
Line 30, “reinforced” should be removed. Also, downsteam of what? Urban city or aerosol source?
Last sentence in Abstract: Be specific about “the cloud invigoration effect”, which is different from convective invigoration. Cloud invigoration refers to larger and/or taller clouds. Convective invigoration refers to stronger storm intensity which usually leads to more extreme rain, more lightning, etc.
6. Detailed comments on responses
(1) #4 response: the description in the manuscript is still confusing. In the manuscript, you said BC is also scaled by a factor of 0.1 for domain 2. Since BC for domain 2 should be from domain 1 simulation, how can you scale it? About “In D2 experiment, the IC, BC, and emissions were scaled by 0.1 for domain 1. The IC and emissions were kept as same with the control run at the same time”, Isn't the second sentence contradicted with the first one? I am still confused about what you wanted to say in the second sentence.
(2) #8 response: Need to clarify in the manuscript (such as in the figure caption).
(3) #10 response: Line number is not correct so it is difficult to identify the text you revised for this comment. But I found there is a mistake in P8 Line 24, how can the cloud top for deep convection only extends up to 1 km?
(4) #11 response: Need to clarify in the figure caption that only cloud ice is considered.
(5) #12 response: The authors made a mistake in calculating the droplet number concentrations. They used water density (1 g cm-3) instead of air density (~1e-3 g cm-3 at low levels) for the calculation. The area mean value should ~ 80 cm-3 as I mentioned in the previous round, not 8e4 cm-3 that is not totally reasonable.
(6) #13 response: I do not understand how more cloud droplets are lifted to freeze can be named as "interim processes". Why not directly describe the process instead of using a term that is not known?
(7) #14 response: there are a few fundamental misunderstandings about cloud microphysical processes: (a) BF process. This process only occurs in the limited regime where Sw<0 but Si>0. In deep convection, most of updrafts are strong enough to make Sw>0. In that situation, both droplet and ice crystal will grow. In addition, this process only increases ice crystal mass, not ice crystals as authors claimed. (b) latent heat. The statements “the magnitude of snow and graupel mass is ten times of that of rain water. The latent heat release due to deposition in cold cloud is stronger than that due to condensation in warm cloud even though the latter is also important” have problems. It is conceptually wrong to discuss latent heat magnitude based on the mass for different phase of hygrometers. snow and graupel are not mainly formed from deposition. Riming is the process for graupel forming which converts a lot of liquid mass to solid phase. The latent heat release from riming may be small only because the latent heat release for converting per unit liquid to ice is only about 1/8 of that converting per unit of water vapor to liquid. I'd want to know in detail how you calculate latent heat for each process in the model. Currently it is just said “diagnosed”. If it is diagnosed from the mass like described here. Then it is not correct. (c) It is also not correct to say “most of the ice crystals fall as precipitation”. Ice crystals would not fall as precipitation. Snow and graupel are the precipitating particles. (d) The figure R10 is confusing. How can warm cloud have deposition and freezing? How do you define warm clouds? Also, why not show the values below 3 Kd-1, which is significant in differences? Please clarify "anomalies that exceed 90% significance level". First, there is no observations so please define anormaly here. Second, how the significance test is done since data between two simulations can not be compared in pairs in grid level because very different clouds could form. If the test is conducted based on mean values, are there enough data for such a test?
(8) #16 response: The convergence should be primarily because the dry cold air meet with warm humid air as a result of large-scale dynamics. Microphysics might enhance the convergence, but it is not the cause of the convergence over the large region. In addition, moisture is increased in the red box domain, which need to discuss where the source is.
(9) #18 response: (a) I do not understand “the persistent convective system makes the impact last for longer time”. (b) The authors missed the point about my question “how the changes in domain 1 impact the results over domain 2”. When emissions and aerosols are changed in Domain 1, the methodological field including temperature and moisture would be changed too. Those changes would impact domain 2 simulation since BC is from domain 1. |