|Review of “Observed Trends of Clouds and Precipitation (1983–2009): Implications for Their Cause(s)” by Zhong et al. |
I commend the authors for the significant work in revising their manuscript. Overall, it has been substantially improved. However, there still seems to be a bit of disconnect in some of their statements, particularly those related to what is meant by expansion of the Hadley cell.
To summarize, I feel as if the authors are not using correct terminology, as has been previously established in (many) prior publications. Which adds confusion and makes interpretation of the results more difficult than is necessary. I would recommend changing some of the basic terminology used throughout the paper, such that it is consistent with prior publications.
As mentioned in the first round of reviews, Hadley cell expansion refers to a poleward displacement of the outer edge of the circulation. There are numerous publications that have used this definition. In the context of precipitation, it has been defined as a poleward shift of the subtropical latitude where P-E = 0; the subtropical latitude where P is a minimum could also work. It has also been pointed out that looking at this separately in each hemisphere is important, because the NH and SH show different tropical expansion signals.
The authors choose to construct their own methodology to define “tropical expansion”, based on the rectangular boxes. That’s fine. But they do not really give a simple explanation for what this new methodology is calculating. It appears to me, based on panel a in Figure 3 from the response, that precipitation is not changing near the center of the region (inner rectangles), but it is increasing in the outer regions (outer rectangles). I assume precipitation is estimated in each rectangle independently, and not summed over the inner rectangles? In any case, this would seem to suggest an increase in tropical precipitation starting near layer 6 and extending outwards to layer 15. In other words, tropical precipitation in this region is moving outwards from the center rectangle? Or maybe a better description is that there is just an increase in tropical precipitation moving outwards from the local maximum (assuming the inner rectangle is a local maximum)? This is not “tropical expansion” as used in numerous other publications. To add confusion, the authors also describe this as not only Hadley cell expansion, but also expansion of the Walker circulation.
The center of the rectangles is “in the middle of Kalimantan, Indonesia which is located near the major ascending/wet zone of Hadley cell”. What is meant by “near” the major wet zone? Is the center rectangle chosen so that it represents a local maximum in precipitation?
The authors then go on to say (again, in the response): “In summary, the spatial distributions of the linear trends of total cloud cover and precipitation are characterized primarily by a widening of the center of precipitation (ascending/wet zone of Hadley cells) over the Maritime Continent in all directions”.
Yes, I agree with this description. But this is not what is meant by “tropical expansion”. This is more related to the thickness of the band of intense tropical rain, right? And the authors are showing an increase in this “thickness”? Perhaps this is not related to “tropical expansion”. But it does seem to be interesting, as others have shown that under continued GHG increases, the ITCZ is projected to narrow (or decrease its “thickness”). But the authors are showing the opposite. It would appear that the authors should remove “tropical expansion” type statements and verbiage, and instead replace this with something more similar to what they are quantifying—“thickness” or area, etc. of the intense precipitation over the Maritime continent.
I think the last reviewer summarized this concern well:
Line 69, 131-132, 138-140: See major comment #4. The expansion of the Hadley cells has nothing to do with enhancement of tropical precipitation. It is related to subtropical static stability (Chemke and Polvani 2019: https://doi.org/10.1175/JCLI-D-18-0330.1). If anything, an expansion of tropical precipitation would contradict the literature, which suggests a narrowing of the Intertropical Convergence Zone in a warming climate (Byrne and Schneider 2016: https://doi.org/10.1002/2016GL070396; Su et al. 2017: https://www.nature.com/articles/ncomms15771).
We understand this is a controversial point. Please see our response to your major comment #4.
But as can be seen, the authors have not really addressed this confusion. They continue to refer to their signal as “tropical expansion” (as well as expansion of the Walker circulation). Furthermore, their response again seems to be disconnected from the comment. The comment is pointing out that the authors are not using the term “tropical expansion” properly. And instead of changing it, they simply say that this is controversial? I don’t quite follow.
Regarding this response:
Direct effects of anthropogenic aerosols on clouds and precipitation tend to be regional and/or sub- yearly time scale, which are beyond the scope of discussion in this study
Obviously, this is a small point and not particularly important to the main analysis. But again, there seems to be a disconnect in the response. Why are possible aerosol effects on cloud/precipitation important on only sub-yearly time scales? I do not think this is true. There have been multi-decadal changes in anthropogenic aerosol emissions, which leads to multi-decadal changes in aerosol forcing. So it would stand to reason that such a multi-decadal forcing may in fact lead to long term changes in temperature, clouds, precipitation, etc.
It’s a bit odd (and frustrating) that I’ve pointed out past papers that have addressed causes of tropical expansion (e.g., aerosols, as well as the PDO), and the authors have chosen to disregard any acknowledgement of these prior papers. Why not add a simple sentence in the introduction that points out prior papers that have addressed the causes of tropical expansion? But then again, I do not think what the authors show is really “tropical expansion”, and therefore, the causes of their signal may very well have nothing to do with previously identified causes of tropical expansion. So in a round-about way, lack of addressing this point fine, I suppose.
The authors again use this terminology:
Further analysis of the widening of the Hadley and Walker circulations (Figures 3a-3h)
What is meant by widening of the Walker circulation?
The authors also use this description of their signal, which is more reasonable in my opinion: “widening of the center of precipitation over the Maritime Continent in all directions”. And maybe “broadening” in more appropriate than “widening”, as widening implies an increase in one spatial direction, whereas broadening is more general.
Another example of a disconnect between reviewer comment and author response pertains to significance:
Table 1: How are you evaluating significance? I have a difficult time believing that a correlation of 0.02 is still significant at the 95% confidence level. Are you taking into account autocorrelations among neighboring grid points, which would greatly reduce the number of degrees of freedom in your t-test? Table 2: Similarly, how is significance being evaluated here? A trend of 0% (see T60%) should not be statistically significant at all, especially at the 99% level.
We used the function imbedded in R named corr to do this significance test. The function corr we chose applies Pearson correlation formula…The p-value of the correlation is determined by calculating the t value as follow…then using t distribution table for the degrees of freedom: df = n-2 to get the p-value. We believe even when the correlation coefficient r is very small, due to the big value of n (the number of samples we used in calculation), the t value should remain a very big value, therefore brings a reliable significance.
Yes, the small correlations (e.g., 0.02) are deemed significant in this analysis likely due to the large n. But this comment is suggesting n is not as large as what the authors are using, due to spatial autocorrelation. But the authors do not address this point.