Yes, the review you’ve all been waiting for. Before I start, let me point out that this has been discussed by WE at WUWT, who has pointed out the obvious problem. It has also been mentioned by KK, though that appears to be more of a meta-discussion about the paper’s reception rather than the paper itself.
[Note: follow-up here.]
To quote KK:
But back to the show. One commenter at WUWT, noting the negative reaction to Judith, gives her a backhanded compliment when he writes: I have to applaud Judith Curry on having the guts to present her paper in the boxing ring of climate blogs where the wild and ignorant rule. but also these that think unbiased and try to address problems in creative ways. I just hope she was not counting on any mercy here. Here’s Judith’s devastating parry: I don’t want your stinkin’ mercy, I’m just lookin’ for some evidence of sentient thought.
Having read through the comments, I don’t think she got any (in the comments). Which is a good time to remind you of my comment policy, which will be strictly applied if needed. In particular, there are plenty of other venues for you to discuss motives, if you want to. This thread is to discuss the science.
Incidentally, because no-one has heard of Liu, everyone is calling this “Curry’s paper” or whatever. Which is wrong. It is Liu and Curry. I’ll call it LC in the sequel, for short.
I was ready to credit LC with the idea of explaining the recent Antarctic sea ice “increase” (it is in quotes because I haven’t myself verified that there is a (stat. sig.) increase; nor (unless I’ve missed it) does the paper present any evidence for a (stat. sig.) increase. That doesn’t matter too much: the important point is the non-decrease, which is interesting, and which is what I’ll call it from now on) but Hank points out that Bob blogged on this back in March pointing out that Manabe said much the same thing back in 1992. Hank’s attempts to get Curry to answer this are here and there but so far Curry hasn’t really read what Hank is saying so her answer is besides the point.
Looking at the paper – data
The main problem with the paper is the uncritical use of invalid data. WE conveniently provides a picture. Note that the pic is somewhat misleading: remember that we only care about 1950 onwards. But it isobvious from that picture there is a clear discontinuity in 1978/9, when the satellite record starts: before then, there is little variablity. This is true at the very least south of 60S. And arguably you can see it south of 50S. There is nothing new in this – everyone in the field knows there are very few obs before the satellite era. Curry knows it very well. But all the paper says is:
We confine the analysis to the period 1950-1999 due to large uncertainty associated with sparse in situ SST samplings in the Southern Hemisphere for the first half of the 20th century.
This is unacceptable. This should get a review of “major revisions” just for this problem. I’m open to correction on this point but I don’t think PNAS is strongly page-limited (unlike, say, Nature of GRL) so there is no reason why they haven’t got room to investigate their data set and see how suitable it is (err, other than the obvious: if you do the investigation (which they must have) you rapidly discover that it *isn’t* suitable). There is no excuse at all for doing what LC have done: which is to say, just picking up a dataset and using it with no attempt at verification (Curry comments on that but WE doesn’t buy it and just for once I’m with him).
So, all the obs stuff they have done S of 60S, or more likely S of 50S, is meaningless. But it gets worse (I think) because EOF is a field process: the inevitable corruption they’ve mixed in doesn’t stay S, but invalidates the entire EOF. So none of the obs EOFs can be trusted, anywhere. The same applies to the analysis of P-E. It uses ERA from 1958 onwards, but again: there are precious few obs to tie the analysis before the satellite ERA. Maybe you can use the data – but you need to use care, and you need to either do some validation or cite some papers that do it for you. LC do neither.
Which brings in another problem: why are they using EOFs anyway? Had they used simple temperature trends point-by-point, at least the fields where there was vallid data would remain valid (curiously they *do* provide this, or equivalent, in fig 4 – but only for the models.Argh). There is no justification in the paper for using EOFs. Also, they only show the first EOF and they only tell us how much var is in the first EOF. Quite often the trainling EOFs have far less var – but not always. And ~30% is hardly overwhelming. There is clearly more info in the unreported EOFs. Oh, and they don’t say if the EOFs are area-weighted. This begins to look like one of those papers where people pull off an off-the-shelf dataset, and some off-the-shelf stat software, and run them through to see what happens. The results are not usually happy.
Indeed, it gets worse: they only use obs from 1950-1999. This means they are gratuitously throwing away an entire decade of hard-won observations (1999-2009 or thereabouts). Why? Because they want to use the C20C (climate of the 20th century) integrations to compare to the obs. But… this is just laziness. Those integrations continue into the 21st century. It is possible to patch the runs together. But LC couldn’t be bothered. This, too, would be grounds for “major revision” before publication.
All of which means that what I take to be the major interest of the paper, viz:
Over the course of the second half of the 20th century, strong warming in the middle latitudes of the Southern Ocean with weak cooling in the high latitudes is hypothesized here to lead to an enhanced hydrological cycle in the Southern Ocean, whereby evaporation is enhanced and the moisture content of the lower troposphere increases in the middle latitudes of the Southern Ocean. This additional moisture is transported poleward by the meridional circulation (Ferrel Cell), resulting in an increased precipitation, and a freshening of surface water in the high latitudes of the Southern Ocean.
remains an interesting hypothesis, but the observational testing done in the paper is badly flawed, probably to the point of uselessness. Disclaimer: it is more than two years since I last played with these datasets. I’ve probably forgotten much.
[Update - oh yes, the other thing to think about might be the De la Mare whale-catch data, though the interpretation of that was always controversial. And see-also the TAR fig 2.16 which (if you believe it) would call into doubt LC. However, marrying ESMR into SSMI/R isn't well accepted]
Looking at the paper – models
As to the modelling, I’m dubious about:
To increase confidence in the interpretation of simulated SST variability, we restrict our analysis to models that perform well in simulating the Southern Ocean climate: 1) the NCAR Community Climate System Model 3.0 (CCSM3), since the observed Antarctic Oscillation and sea ice variability is well represented in CCSM3 (10, 11), and 2) the Geophysical Fluid Dynamics Laboratory Coupled Climate Model (GFDL-CM2.1), which has peak winds close to the observed latitude and a reasonable wind stress over the Southern Ocean, fed with the right amount and properties of the North Atlantic Deep Water, resulting in near-observed ACC transport.
I think they’ve just picked a couple of their favourite models. See-also An Antarctic assessment of IPCC AR4 coupled models for a more objective assessment (by me, of course).
Having just read through the paper, I was looking for what they would then actually do with the modelling results. With the 21stC runs, they don’t do much more than say “it got warmer”, which isn’t news. What they do with the C20C runs sort of slipped by me. Looking back, there doesn’t seem to be much there. What they *could* have done, since with the model you really do have all the info, would be a careful analysis of the hydrological cycle, P-E, and the ocean salinity, and the ocean vertical heat transport. OTOH, there is a big problem with doing that, because (as they admit) none of their models show EOFs that look anything like the obs; so by examining the models, whilst you could look at the mechanisms that affect the ocean vertical heat transport, you wouldn’t overall be looking at what had happened in the real world.
Looking at the comments – Dumb America (again)
OK, that is enough about the paper for now at least. Time for some fun.
Since I ref’d WUWT, I should point out that there is a true classic of the Dumb America problem. At first it was just funny, but then someone managed a second layer of ignorance on top of the first that made it delicious. And that all of this is in the context of the poor septics unsure whether to bury Curry or to praise her makes it all the better.
So, the problem. The paper includes
This increased precipitation, mostly in the form of snow, stabilized the upper ocean and insulated it from the ocean heat below. This insulating effect reduced the amount of melting occurring below the sea ice. In addition, snow has a tendency to reflect atmospheric heat away from the sea ice, which reduced melting from above
So, all reasonable enough, to anyone who knows about this stuff: this is a scientific paper after all, we don’t need to draw with crayons. Snow falls on the ocean (and melts – this was so obvious that the paper didn’t say this, but alas it has confused many – see later) and therefore reduces the salinity of the upper layers. Less saline water is less dense, and therefore the tendency of the upper layer to turbulent motion, leading to heat gain from the relatively warmer lower layers, is reduced. Indeed, this is explicitly spelt out in the paper:
The increased freshwater input in the high latitudes of the Southern Ocean would decrease the upper ocean salinity (density), leading to a more stable thermohaline stratification and weakened convective overturning. This reduces the upward ocean heat flux available to melt sea ice.
But, maybe you don’t know this stuff – that isn’t unreasonable either. I’m sure there are papers about QM or GR that I don’t understand. But the difference is that with the septics, anything not understood must be wrong, and must be mocked. So we have:
So I read it; ten times actually; and it forms a picture in my brain. In this picture I have the Southern Ocean which is a favorite place for round the world sailors; and it consists of salty water and other stuff. So there are clouds above this salty water; and if it is cold at times it snows over the Southern Ocean and the snow falls in the salty water; and being fresh water the snow floats on the salty water. Funny thing is I have never heard of any round the world sailor who ever mentioned finding snow on the water while down there in the Southern Ocean; but I’ll Take Dr Curry’s word for it if she says there is snow on the water.
and there is more (someone provides the right answer here). But even that doesn’t cover all the misunderstanding, because someone then manages:
I can understand the effect of snow on the albedo of ice, but that is not the issue. It is the effect of warmer water sea ice is floating on. I have looked up the thermal conductivity of pure water (0.561 W/mK) and of seawater (0.563 W/mK). This small difference in no way makes fresh water an “insulator” from sea water or makes it easier to cool, in fact it appears to be the opposite.
The error there being the assumption that thermal conductivity matters much, when the heat is actually transported by bulk motion, which is what is being suppressed by the less-saline surface water. Err, as the paper itself says (Curry explains that, but perhaps not as clearly as they need; rt explains it more clearly; but not clearly enough it seems).