By and large, the authors have addressed my concerns except for several places that need further clarification or revision. I appreciated their efforts and hard work. Prior to be accepted for publication in ACP, I suggest the authors revise their manuscript by considering again the following specific comments:

1. To be honest, i do not like the reply to my comment #8, most of which are flawed: For instance:
a) Data quality issue of radiosonde measurements: "Operational specifications for conventional upper-air meteorological observations" issued by CMA is way too outdated, the authors are strongly recommended to refer to Zhang et al., 2018 (doi:10.1175/JCLI-D-17-0231.1) and Guo et al. 2019 (doi:10.1029/2019GL082666) and the references therein for more details.
b) There is grammatical errors in "The sounding balloons incorporate radiosondes were regularly..", which can be considered revised to "The sounding balloons were operationally.."
c) "60 layers" is misleading. Actually, there is typically about 3000 layers given the sampling frequency of 1 second for the balloons (c.f. 5 - 8 m resolution in the vertical)
d) "35 km": for most of the soundings, it is hard for them to reach such altitudes. the authors can revise it to "Normally there were 3000 measurements, more or less, recorded during the flight path of balloon (Guo et al., 2016, doi:10.5194/acp-16-13309-2016)."

2. Page 1 L25: "This implies that" can be added before "Data assimilations can partly" .

3. Page 2 L32-34：The authors only discussed the direct way that dust affects the radiation budget. The indirect way is ignored. Basically, dust aerosols, particularly those coated with sulfur and other soluble materials, are found to alter the cloud properties by serving CCN), giant CCN (GCCN) and ice nuclei (IN) (Yin et al., 2002, doi:10.1029/2001JD001544; DeMott et al., 2003, doi:10.1029/2003GL017410; van den Heever et al., 2006, doi:10.1175/JAS3713.1).


4. Page 2 L40: "also" is redundant and can be removed.