|This is the revised version of a paper using extensive analysis of satellite data to explore the radiation balance at the surface and top of the atmosphere as it pertains to climate change and the Arctic amplification, and especially the interplay with melting sea ice and changes in clouds. Like I commented in the first review, the study is highly timely and very important and the results should be published. It is very welcome and I do encourage the authors to revise and resubmit this paper. |
While the original paper was rather poorly put together, it pleases me to note that the revised version is much better, and I enjoyed reading it. In think it shows that the reflectivity in the Arctic is not decreasing as fast as it should¨, given the melting sea ice, and snow, that this is due to a concurrent increasing in optical thickness of the clouds and that this is due to more liquid and less ice. But sometimes I do feel confused by the text and, hence, we’re still not at the point where I can recommend publication. The language is sometimes a problem and there are also other issues with the manuscript as such that needs a second revision. I don’t think much more analysis is required, although some statistical measures could be refined, but I will still recommend major revision, just to make sure this revision will happen.
While the clarity of the text is increased substantially, there are still problems with the message. It is unclear to this reviewer if the CRE – which incidentally is the accepted vocabulary and not CRF – is positive or negative at the pan-Arctic scale. On line 632 RTOA is increasing but on line 604 there are “decreasing pan-Arctic trends of reflectance” – unclear if at TOA or surface. And these are not the only times contradictory, or at least seemingly contradictory messages are delivered. Hence the whole text needs a work-over to make sure that the message is clearly conveyed in an understandable manner. I think I get it, but also I am sometimes confused.
The issue of what is measured and what is coming from other sources, be that a priori model data (e.g. reanalysis or operational models) or from radiative transfer modeling (or both) needs to be much clearer. I appreciate the discussion on taking as much as possible from direct observations, but no satellite in the world can observe downwelling radiation at the Earth’s surface; that just has to be a model product, although it may be constrained by observations. And very likely there is a (fallible) model temperature profile in there somewhere, am I right?
The issue of what is and what is not significant is much better handled in this revision than in the original manuscript. But it only goes as far as it does; several of the maps still lack stippling for what is considered statistically significant, and the text sometimes ignore this and discusses significant and insignificant trends in the same sentence. I note that very few of the results over perennial sea ice are ever significant, which is something the authors need to comment on. I also see many references to trends at or near the North Pole, where there are no observations at all. I also see significance coming and going between optical thickness for liquid, ice and total clouds water; presumably if one of these are not significant, none of the others can be and there is very little discussion about how the significance relates to accuracy; a rends can in fact be statistically significant and still meaningless if for example it is so small it doesn’t matter or if the accuracy of the observations is so poor that it can’t possibly be resolved.
Finally, there is a debate in the scientific community as to if aerosol indirect effects are responsible for any of this or if it is all thermodynamics, or maybe even dynamics since clouds tend to form where the dynamics dictates they should form; dominating clouds are different in different climate regimes because of the general circulation more than anything else. The authors discuss changes in optical depth and water paths for quite a while but it isn’t until end of page 24 that aerosol concentrations come in via the effective radius. This is followed by a confusing discussion on trends in this parameter that seems to be there to satisfy someone that wants this to a factor. Statistical significance is not discussed and there’s a lot of handwaving. I suggest that the whole thing about whether the changes in optical thickness are due to changes in water paths or effusive radius is either given its own section and is based on what measurements are available, or left entirely to another paper, where this can be properly addressed.
Page 1: Title is clunky and awkward and reads like a part of the text. I suggest a much short and snappier title would enhance the chances the paper will be read!
Page 1, lines 13: Temperature doesn’t have a “size” as such; you never say “it’s hot today, the temperature must be large”.
P1, l13-20: No need to go back to the “old Greeks” here. Just state that the globe is warming because of anthropogenic emissions of greenhouse gases, that the Arctic warming more then the globe on average and that this set a number of feedbacks into play where radiation is a key to some. Neither Arrhenius nor Keelings work is necessary as abcakground for this paper
P2, l21: A recent paper in Nature (I think) has the Arctic amplification to a factor of four, so it should at least be “larger than twice”.
P2, l37-44: This section is a bit awkward. It starts by telling the reader what the satellites measure, but only for SW radiation (l37). Then at l41, LW slides in but not as something that satellites measure, but as something that “also” modified by clouds. Satellites measure radiation, plain and simple, across the whole spectrum and everything else is inferred from this, quite often using a priori information that is hardly ever discussed.
P2l49-50: Sea ice is also water, so if you mean open (= ice free) ocean you need to say that. Come to think of this is slightly tautological – what elese would and ocean be made of if it wasn’t water?
P2, l49-51: This is confusing; given that the annual cycle of CFC from many studies goes from ~40-70% in winter to ~70-95% in summer across several Arctic stations, it is difficl to see i) how the in summer is “located in the north Atlantic…” and the second – wherever that is – is only 40%.
P2, l53-p3, l58: I think what the author is referring to hear is the fact that the same observations from AVHRR when run through different retrievals give different results. But the question is, if different retrievals are used, rae the results then the “same data sets”?
P3, l64: “by changes in atmospheric”. If the reflection by the atmosphere is not changing it cannot offset another change.
P3, l67-68: You need a reference for this if you are to use this argument here. You cannot start using the results in this paper in the introduction to it.
P3, l70-72: This is trivial and is true for all clouds everywhere on Earth. Without understanding dynamics, understanding the clouds is futile! Cloud formation needs water and temperature and aerosols, but without dynamics (advection, surface cooling, evaporation, lifting, subsidence etc., etc.), it still won’t happen 99% of the time. It is not an accident that we find subtropical stratocumulus in the subtropics and deep convection along the ITCZ!
P3, 73-74: One step to many here; logically, trends in albedo cannot be “compensated” by CFC. This is only one component and your taking to many logical steps at once here.
P3, l87-P4, l92: The Pithan and Mauritsen paper lists cloud feedbacks as a minor third in importance, after albedo feedback and lapse-rate feedback. In fact when they say temperature, the do not mean thermodynamics in general; they do mean “temperature” plain and simple.
P4, l100: What are “sea ice edge shelves”?
P4, l105: “”modulate” is better than “regulate”.
P4, l108-109: First, this is true at the surface as well as at TOA, so drop the last part. Second, here and throughout, the accepted terminology is Cloud Radiative Effect (CRE).
P5, line 142: “Two exceptions to the latter…”?
Figure 1. Is BOA an accepted abbreviation? Else, if you mean the “surface”, then say write “surface”.
P7, l174-P8, l183: Apparently someone criticized how the seasons were divided and the authors feel a need to defend themselves. I would recommend that you don’t, you just looked at the data and use the delineation that made most sense. None of the arguments you give in this paragraph are very good. I don’t understand why meteorological seasons are no good because you are using a long data set, and looking carefully at the figure, the transition from June to July is structurally different comparing the beginning and the end of the time series.
P8, l204: This is not only lower, it is differently much lower in different clouds at different times.
P9. L228: Really!! 369 and not 368 or 370? How accurate is this observation?
P9, l239-241: How large are these biases and why the different signs. Why should surface heterogeneity generate a bias with different signs for small and large values?
P9, l244: Is the sea ice albedo in these calculations always the same? Is that what you say? Have you been to the Arctic? What about snow or bare ice, what about melt ponds. This is gross!
P9, l246: The sentence “cloud radiative … surface” is somehow meaningless or at least trivial. Maybe the authors think about the surface albedo, which often determines the sign of the CRE.
P10, l263: Do not understand the meaning of “seasons of our paper”. Do you refer to the time period or the choice of seasonal boundaries or what?
P12, l301: Why do you expect that a warming Arctic would feature a “statistically significant” decrease in reflectance? Why a decrease and why significant? If you mean that the loss of sea ice should lead to a lower albedo, then say substantial and leave the statistics out of this.
P12, l305-207: Isn’t it also possible that the CFC in summer, when the ice is melting, is so high and the cloud thickness sufficient, that there would not have to be an increase in anything for the surface albedo decrease to go unnoticed at TOA?
P12, l310-311: Drop “local”. Why 75%; SIE is usally defined at 15%...
P12, l316 and onwards: Very few results are statistically significant anywhere over perennial sea ice. That needs to come out in this whole discussion!
P12, l317: In what sense do you mean that a trend in one location can be “compensated” by another trend at another location?
P13, l323: You do not have any observations around the North Pole!
Figure 6: What are the black bars in the figure? They are much larger than anything else...
P15, l348-349: Why mention this at all if it’s not significant? These changes are so small they are well within the measurement uncertainty; CTH cannot even be defined this accurately!
P15, l351: Suggest: “… that the temporal trend over two decades for τ of liquid clouds …”
P17, l368-376: Again, most of the ice area lacks significance and there are still no observations at the North Pole; also at line 376.
P18, l398: So which is it - “and”, “or” or both? Or can’t you tell. Why do you even state this here, when no results have been shown for Reff yet?
P19, l405-407: And yet almost all data from in-situ studies suggest that CRE is positive; that clouds warm the surface almost always, especially over sea ice, except briefly in summer when surface albedo drops enough!
P19, ll417-418: Such low albedos basically mean open water; very fe land surfaces and no see ice has an albedo as low as 10%.
P19, l420: This sentence seems to be contradicting what was stated earlier. You need to be very careful here; are the clouds warming or are they cooling?
P19, l431: Is this result statistically significant?
Figure 10: Statistical significance please!
P20, l445-446: This is the question isn’t it? How do you know this as a fact?
P20, l447: Is it more liquid in the clouds or are there more liquid clouds? Or is it Reff…?
P21, l449: Is this a statically significant result? Just because the trends in both is doesn’t mean the trend in the differences is.
P23, l503: “Exactly what is it that is “the case”?
P23, l513: Do you mean “absorption”? Sounds like a contradiction otherwise…
P23, l522: This has nothing to do with “midsummer”, which here is actually in spring. Rather it is late summer when the surface albedo is at a minimum; mid-September…
P23, l526-528: Awkward and confusing; if the ocean is not the surface of the water how can there be a convergence of it?
Figure 12: Interesting figure but complicated. Try and modify so its easier to understand. What is on the y-axes?
P24, l543-P25, 547: Confusing sentence at the start: what is a "decreasing trend", is that the second derivative or do you mean a “downward trend”? Else, this is the question isn't it, so why wait until here? Either get to the bottom of the Reff problem or leave it open! Referring the reader to an Appendix isn’t good enough; either you do it you don’t!
P25, l547: I don’t see the “mostly decreasing” trends in Reff in Figure C1; the opposite I would say. It seems to be more increasing than decreasing, especially over the high Arctic.
P27, l632: I though RTOA was decreasing but not as much as you expect it to?