|Although the authors make an effort to address many points raised by the referees, they only tried to provide an explanation for the strong atmospheric LW cooling and weak SW warming through indirect means, i.e. through the examination of the day-night differences in CALIOP AOD. Also, I still cannot understand why the authors do not simply switch off the dust cycle in order to investigate its effect and try to do so through unnecessarily complicated land-cover experiments. Until these important issues are resolved, I cannot suggest publication for the manuscript. Please see my justification for this below. I have also a few other criticisms, but they are not as important.|
Section 3.1.4 and Discussion
My additional efforts to identify cases with general dust-induced atmospheric LW cooling stronger than SW warming were fruitless. In contrast, I came up with even more studies where SW atmospheric warming is significantly larger than LW cooling (e.g. doi: 10.1002/2014JD022077, 10.5194/acp-14-3751-2014, 10.1002/qj.771 (indirectly), 10.5194/acp-17-2401-2017). Even the same authors in their earlier paper Sun et al., 2012 (DOI: 10.1029/2011JD017219) show in Fig. 9 a dust radiative effect with significant atmospheric SW warming and less significant LW cooling, as expected. In Fig. 12 of the same paper, they present dust-induced atmospheric warming.How can the radiative cooling be so prominent only for the Tibetan dust?
The day-night differentiation in Fig. 6 and Section 3.1.4 is not convincing. The CALIPSO night and day AOD seem rather comparable, with the exception of missing data over the Qaidam basin for the spring days. Missing AOD data does not mean small AOD. The stronger radiative effect in the LW has not been adequately justified because:
1) the day-night difference in the model AOD has not been shown. It is this difference that might create the stronger LW cooling, while the day-night difference of CALIPSO is not convincing. Even if it were, it would be a indirect explanation, because it is not used in the radiative transfer model. The model AOD is.
2) The reported CALIPSO AOD is given probably at 532 nm and the model AOD wavelength is not reported. The strong LW cooling would be better examined if the AOD in the thermal infrared were reported.
3) The optical properties of dust both in the SW and the LW are not given in the paper, even though they were requested in my previous review. The Tibetan dust might be not as absorbing in the SW as at other areas and therefore explain the weak SW warming. At least the mean area single scattering albedo would help highlight the reason for the dominant LW cooling.
The authors state in their reply that "The warming effects in their study may be caused by black carbon or the dust aerosols from Taklimakan during daytime, but the cooling effects in our study is mainly caused by the dust aerosols emitted from the TP at night." Could this be verified by a two-dimensional map of the effects, similar to Fig. 8a or b?
p. 10, ll. 20-22: "the dust effect reported in our simulation is significant in the heavy dust years, but the cooling over central India in the light dust years may be caused by the model's internal variability". Why? It appears that the magnitudes of cooling in the two cases are 0.6-0.8 and 0.4-0.6, respectively. Both seem quite larger than the standard deviation of Fig. B2a, unless the part of India that is not shown in Fig. B2 is characterized by larger standard deviations.
p. 10, ll. 24-25: "The dust cycle in the two experiments was turned off". It appears that the authors can switch off the dust cycle. This (without changing the land cover) is what I proposed in my initial review, but the authors decided to keep the land cover change. The reasons remain unclear to me. If it is possible to switch the dust cycle off, then why would one stick to the methodology of changing the Tibetan plateau cells land cover? After all, the focus of the paper is exactly the effect of the dust and I presume not the effect of the land cover type. Instead of the comparisons a) normal land cover with dust vs. altered land cover without dust and b) normal land cover without dust vs. altered land cover without dust, the authors could have only performed the more straightforward c) normal land cover with dust - normal land cover without dust. I suspect that unwanted factors other than dust may influence the paper results. I would also (if possible) test the sensitivity by decreasing directly the dust emission in steps.
Related to both this point and the unexplained to me dust-induced atmospheric cooling, it is strange that in Fig. B1 of the authors reply, the enhanced LW cooling is reported as a 32-36 deg N average. I think that relatively little dust exists in these latitudes (Fig. 6 left column). Could it be that some other factor, unrelated to dust but related to the altered land cover, influences this LW cooling?
p. 12, ll. 14-15: The last sentence should be removed after the revision.
p. 1, l.14: "dust coupled" -> "dust-coupled"
p. 6, ll. 20-21: "were relatively low" -> "shows larger values"
p. 8, l. 15: "donwstream" -> "away", unless there is reason to define an atmospheric flow direction, which I may have missed
p. 10, l.7 "Dust direct radiative effects is" -> "Dust direct radiative effects for the atmosphere are"