## Watt about a cessation of global warming?

I know that Sou has already discussed this, but I thought I would add a comment of my own. Bob Tisdale has a new post on Watts Up With That (WUWT) called another peer-reviewed paper predicting the cessation of global warming will last at least another decade. The post is about a new paper by Li et al. (2013) which suggests that the current “hiatus” may last another decade or so.

Basically the paper decomposes the northern hemisphere temperature record into two parts : a long-term trend (which is rising) and an oscillation. They determine that the oscillatory part correlates well with the North Atlantic Oscillation (NAO – the dominant mode of atmospheric variability over the North Atlantic region), with a lag of 16 years. Given that there is a lag of 16 years, they assume that they can predict the future northern hemisphere temperature using the known NAO values. However, they do have to make an assumption with respect to the long-term trend, which they assume is linear. The function they use to predict future northern hemisphere temperatures is below.

So, they use the NAO values (which is simply an oscillatory function) and assume that the long-term trend is linear in time. The coefficients a, b, and c are determined via linear regression and are a = 0.38, b = 0.08, and c = 16 (they are determined using two different past time intervals, but these are approximately the values). This means that this model assumes an underlying warming trend of 0.08oC per decade.

Between 1990 and 2011, the NAO decreased by 0.5, so the first term in the above equation (given a = 0.38) means the the NAO would contribute to a drop in NH temperatures of 0.19oC between 2011 and 2027. The linear trend part of the above equation would imply that between 2007 and 2027, temperatures should have risen by 0.16oC. Hence, the Li et al. (2013) model suggests a small drop in NH temperature between now and 2027. However, what Bob Tisdale fails to acknowledge, is that this still includes an underlying warming trend of 0.08oC per decade. So, this does not – in any way – contradict mainstream climate science. Furthermore, the NAO appears to vary by about 0.8-1 over a period of 40-60 years. After 2027, when the NAO is likely to start producing a rise in temperatures, the equation above would indicate that NH temperatures would likely rise at about 0.2-0.3oC per decade (assuming that the model has any merit).

So, a few closing remarks.

1. To a certain extent, this is just a curve fitting exercise. Having said that, it doesn’t seem unreasonable that oscillations in the temperature record correlate with things like the NAO.
2. It only applies to the Northern Hemisphere, so am not sure what it would predict for global temperatures.
3. The model fundamentally relies on an assumption that there is a long-term, anthropogenic warming trend. Therefore, what’s presented here in no way contradicts the underlying theory of anthropogenic global warming.
4. Although the model “knows” the NAO index relevant for the period up until 2027, it does not know the underlying warming trend and so has to assume that this will continue with the same trend as it has in the past. Given that anthropogenic forcings are getting ever more significant, this assumption may not be well-founded.
5. If we assume that CO2 concentrations will double – relative to pre-industrial times – by about 2050, the underlying warming trend used here (0.08oC per decade) would imply a transient climate response (TCR) of about 1.4oC, entirely consistent with many other estimates.
6. The NAO appears to be an oscillation with a period of about 60 years and with an amplitude of about 1. Hence a flat trend for the next few decades (because the NAO influence cancels the underlying warming trned) then implies a period after that where the temperature will rise at least twice as fast as the underlying warming trend.
7. There is, though, a fundamental problem with the temperature trend remaining flat or falling slightly between now and the mid 2020s. Today, we’re accruing energy at about 1022J per year, most of which is going into the oceans and a small fraction is contributing to a slow (0.05oC per decade) rise in surface temperatures. In 20 years time, if the temperatures remain flat, the added forcings due to increased atmospheric CO2 concentrations would have at least doubled this energy imbalance. For temperatures to remain flat, an ever increasing fraction of this would have to go into the oceans. This may be possible, but would seem unlikely.
8. So, an interesting paper but one that doesn’t contradict any of the fundamentals of AGW and that probably suffers from being rather simplistic when it comes to making predictions about the next 16 years. It’s possible that trends could remain flat into the 2020s, but I suspect that this is rather unlikely.
This entry was posted in Anthony Watts, Bob Tisdale, Climate change, Climate sensitivity, ENSO, Global warming, IPCC, Watts Up With That and tagged , , , , , , , , . Bookmark the permalink.

### 27 Responses to Watt about a cessation of global warming?

1. Sou says:

I’m not entirely clear about what is new in this paper. The AMO obviously affects NH surface temperature because it’s embodied in it. I expect it’s the link between the North Atlantic Oscillation and the AMO. Has this not been explored previously I wonder?

2. I’m not sure myself. It seems that you’re more familiar with some of this than I am, so if you’re unsure, I’m even more so 🙂

3. chris says:

From a pure “sciency” point of view, this and similar papers are interesting because they provide what are essentially hypotheses about second order elements of global warming in response to enhanced greenhouse forcing. Since hypotheses by their very nature exist in a state where definitive observational data is lacking one can choose to support or ignore them pretty much on whim. They can be useful in support of agendas, especially if their contents and consequences are ignored or misrepresented :).

(apols if that’s a little philosophy of sciency, but I’ve been reading Lee Smolin’s “The Trouble With Physics” in bed this morning which induces that sort of perspective!)

As you indicate very clearly in your post Tisdale seems to be ignoring important content (i.e. that the underlying trend accounts for essentially all of the surface warming of the last 100 years, and so the ocean oscillations have little to say about the net warming, its underlying attribution and its likely future extent)….and consequences (that any suppression of surface warming by sequestering of heat in the oceans simply stores up greater surface warming to come, since the radiative imbalance is increasing rather than dissipating).

It is fair to point out ‘though that if these oscillations do make significant contributions to surface temperature variability on multi-decadal timescales, the particularly rapid warming from mid 1970’s through 2000-ish may have had a boost from this contribution. However, I don’t believe the issues are yet sufficiently tied down to make solid conclusions yet…

4. Yes, I agree that it is likely that these oscillations will result in periods when the surface warming trend is faster than the underlying long-term trend and periods when it is slower (as now).

Something I hadn’t appreciated when I wrote the post is that when the paper decomposes the NH temperature into an oscillatory part and a trend, it considers the period from 1920 to now. It shows that the underlying trend is roughly linear in time. However, during the period 1920 – 1970, solar forcing contributed partly to the temperature rise. Solar forcing is now dropping and there is evidence for some cooling through aerosols. That would seem to indicate that the anthropogenic contributionhas been increasing with time and hence that assuming a linear extrapolation likely underestimates the anthropogenic contribution to the future surface warming.

5. Tom Curtis says:

Wotts, some of your points are, IMO, incorrect. Specifically:

3) The model does assume a long term trend, but the assumed trend is just 40% of the CMIP 3 (AR4) expected trend, and 33% of the CMIP 5 AR5 trend. That would suggest climate sensitivity to be 40% and 33% respectively of that predicted in those reports – or around 1-1.2 C per doubling of CO2. That is not in agreement with the underlying theory of AGW; or perhaps, it is in agreement on the underlying physics of the greenhouse effect, while suggesting a near neutral net feedback response.

5) Assuming an anthropogenic forcing induced trend of 0.08 per decade, and a current temperature increase of 0.8 C relative to the preindustrial era, that shows a net anthropogenic increase of less than 1.1 C by 2050, and a TCR of 1.1 C per doubling of CO2 (using your assumptions about forcings). The Lee et al model may predict a higher temperature by 2050, but only because it will be in a projected positive phase of the lagged NAO.

Personally, I think these incoherencies with AGW tell against the theory in Lee et al. That is particularly the case as a 16 year lag is going to require some very interesting (ie, complicated) physics to explain, and appears simply as a deus ex machina for the theory. In fact, the lag is sufficiently large that 53% of possible sixty year cycles can be “explained” as either a lagged response to the NAO, or a lagged response to the inverted NAO with a lag of 16 years or less. That being the case, Lee et al’s evidential constraints are so lax as to provide almost no justification for their claims.

6. Okay, yes you’re right that their long-term trend is below the CMIP3 and CMIP5 trends but that’s kind of what I was implying in my point 4 above (that their linear assumption may not be well founded). In my response to Chris above I also point out that their linear assumption is likely an underestimate because it ignores the other known forcings.

My TCR estimate was simply an extrapolation from 1880 to 2050 assuming 0.08oC per decade giving a change (since 1880) of 1.4oC by 2050. However, I see what you mean. From now to 2050 would indeed only give a total change of 1.1oC so, yes, I agree that would be quite a bit lower than other assumptions.

So ultimately I agree that Li et al. is probably way too simple and unlikely to be correct. However, even given this, Bob Tisdale using this to refute the IPCC conclusions is a mis-representation of the paper. Bob has left a comment on the post that – in my view – illustrates his confusion

I forgot to call attention to the final few words of the Li et al. (2013) abstract, “that temporarily offsets the anthropogenically induced warming,” and the fact that ocean heat content data and satellite-era sea surface temperature data both indicate the oceans warmed via natural processes, not from manmade greenhouse gases.

Really, it’s a fact that natural processes can cause energy to accrue in the climate system at a rate of 1022J per year? Sounds like a violation of basic conservation of energy to me.

7. Note also the sleight of hand, AKA lying. Although the WUWT headline is “Another Peer-Reviewed Paper Predicting the Cessation of Global Warming Will Last At Least Another Decade” and that’s the myth they’ll run with in future, the text says “A few days ago, the Georgia Tech press release for Wyatt and Curry (2013) included a quote from Marcia Wyatt, who said…” and that’s the truth. But “another press release predicting cessation” doesn’t make a good headline.

8. Sou says:

The corollary to what you’ve said, Chris, is that there will be another “boost” in the next upsurge.

What I read is that the period of the AMO is between 60 and 80 years in modern times, and 60 to 100 years going back further in time (centuries). I expect it will take a few more decades to centuries to see if / how it’s affected by AGW.

From Li et al and elsewhere, the AMO affects NH surface temperature by around +/- 0.2 degrees (upswing and downswing over the 60 to 80 years), which isn’t all that much when compared to AGW. Less than that at the global level. If we don’t curb CO2 emissions, it’s likely the effect would be barely noticed compared to the rise in global surface temperature, though it should still be detectable.

9. I think that’s a good point. The influence of anthropogenic forcings has changed considerably over the period used by Li et al. and so assuming that the influence of the AMO will be the same in the future as it has been in the 20th century would seem to be a rather simplistic assumption. In fact, one could imagine ocean heating changing the behaviour of these oscillations.

10. But dropping the snarking and looking at the paper: it does indeed seem to be mostly curve fitting. Scafetta comments (http://wattsupwiththat.com/2013/10/12/another-peer-reviewed-paper-predicting-the-cessation-of-global-warming-will-last-at-least-another-decade/#comment-1446101) (a) that he’s already done all this, so the paper isn’t new, but more interestingly (b) the Li et al. assumption that its the NAO leading SST is dubious; it could just as easily be the other way round.

For myself, I’d go for it just being a co-incidence. Furthermore, it looks to fit when you de-trend SST, but the NAO, so to speak, doesn’t see the detrended temperatures; it sees the whole.

11. That’s an interesting point. I hadn’t considered the possibility that it could be the other way around and that SST are driving the NAO. That may actually make more sense given that the lag is likely because of how long it takes the oceans to respond to changes in forcing.

12. jyyh says:

this lag of 16 years reminds me of the try I made to get the solar influence maximized, eventually abandoned it since the lag was somewhere around 84 months. while this could be tied to the lemming cycle in the arctic, I didn’t dare to suggest it anywhere. later when Tamino did a proper analysis the lag was found to be 2 months, which could be an appropriate time for ocean to lose heat acquired during summer.

13. O. Bothe says:

As I said on twitter: NAO and AMO are like chicken and egg. Timeseries too short. Simulations in some sense inconclusive. Good arguments for AMO forces NAO and for NAO forces AMO. To complicate the issue: good arguments for Pacific sector influence on NAO. Air-sea-interaction discussed since Bjerknes. Then came Hasselmann (as quoted by Li) and many more. Still topic not “settled”.

(OK, that wasn’t really on Sou’s point)

14. Interesting, thanks. That sounds consistent with William’s comment below. We don’t actually know if the NAO actually leads temperature.

15. I’m sick and tired of studies which omit the aerosol forcing when they are aiming at understanding the air-sea-interaction better. It’s outright laughable if you do that in the northern hemisphere where we know there has been a substantial dimming in the order of several W/m2. Dicussing the literature about this very forcing issue in two brief sentences – only to dismiss it entirely – in order to satisfy the reviewer (certainly no aerosol guy amongst them) is not enough I’m afraid. While their established lead-lag relationship might survive in a proper analysis (regardless of how plausible it seems to be), the results are probably not robust and hence not credible. Tom Curtis’ comment emphasizes that there is a substantial problem with this paper.

Same goes for Wyatts and Curry. Interesting analysis, but essentially useless owing to the wrong external forcing assumption. Of course, the usual suspects jump on these papers as if they provide ground truth. They are just fooling themselves! Seems that many colleagues are good in fooling themselves either. An atmospheric physicist should know better! People seem to forget about the physics in their unwarranted panic to explain the “Hiatus”.

Finally, it strikes me as extremely odd, that the most important metric of internal variability, namely ENSO, doesn’t show up half as distinct globally as the alleged NAO/AMO signal in the NH. How likely and how plauible is that, given that we haven’t even found a plausible mechanism other then AMOC? Could it be sulfate aerosols who pull the trigger perhaps? Hemispheric T2m vs Sulfur emissions

16. Karsten, it seems to me that this study has essentially ignored all forcings. As far as I can tell it simply decomposes the NH surface temperature into what is essentially a linear trend and an oscillation and then uses that to fit to the coefficients of the function used to estimate the future change in NH surface. So, in a sense they’ve assumed that the underlying trend is a fair representation of the anthropogenic trend when, presumably, it overestimates the trend prior to the mid-1950s and (as I think you’re indicating) underestimates the anthropogenic component during the second half of the 1900s. This is probably similar to the point Tom was making. Their estimate for the future underlying trend is almost certainly wrong for many reasons.

17. dana1981 says:

I was going to make a comment similar to Karsten. I think these sort of curve fitting papers are basically worthless. You can fit whatever you want to the data – might as well be pirate underwear size and number of sheep in New Zealand. If you don’t have a solid physical basis, then it just doesn’t tell you anything interesting. Scafetta’s got his astronomical cycles, Curry’s got her ‘stadium waves’, and a few others have various ocean cycles. As Karsten says, if you don’t at least account for changes in forcing, it doesn’t tell you anything of value.

In particular the aerosol forcing increased mid-century then flattened out after 1970. If you don’t account for that, your fit is going to mis-attribute that change to whatever variables are in your model (Jupiter’s orbit, NAO, stadium waves, pirate underwear, whatever).

But yes, you’re correct that this paper doesn’t say anything about the cessation of the underlying long-term warming trend. Neither does Wyatt and Curry, by the way. These papers are only trying to explain short-term internal variability. And yes, that’s probably mostly due to ocean cycles. But Kosaka & Xie did a much better job modeling these than Li or Wyatt & Curry, because their model included physics.

18. Yes, I would largely agree. These curve fitting exercises have obvious problems that yourself and Karsten have illustrated well. I can understand using techniques like this if you already understand the underlying physics and simply want to present a nice simple parametrisation that others can use. Actually doing this to try and explain what might happen doesn’t really explain anything, and if it does turn out to actually make a reasonable prediction that’s more likely luck than design.

19. Seems they detrended linearly and also tested regression against GHG forcing. But they didn’t regress against the TOTAL forcing (anthropogenic and natural likewise)! Why??? I just don’t get it! Much more sophisticated studies have already been published earlier (e.g. Terray 2012 to name but one). It simply isn’t controversial anymore that the forcing is highly non-linear! So is the fact that we’ve seen substantial dimming and brightening. NAO is sensitive to volcanoes and undoubtedly to changes in forcing over short periods of time. Why am I supposed to take a paper which does not even discuss this issue seriously? I have no problem to admit that I had rejected the paper right away in the current form. Think about it: Isn’t it strange that the most active NAO phase coincides with the strongest (northern hemispheric) radiative forcing pulse we’ve ever witnessed? Yet, they remain silent about all that. But then … science progresses slow. And being wrong in interesting ways isn’t a bad after all (credit to Ray Pierrehumbert) … though a bit tiresome if repeated too often. I finally got to write a good proposal myself in order to devote my entire time to the subject (current project keeps me from doing it) 😉

20. Pingback: Comments elsewhere, part III – Stoat

21. Bobby says:

Human’s have the ability to see patterns where they don’t exist. Curve fitting exercises that rely on “cycles” and “lags” with non-physical parameters just deserve a roll-the-eyes IMO.

As for Tisdale, nobody ever taught him about conservation of energy.

22. Tom Curtis says:

If repeated continuously, the ways in which people are wrong are no longer interesting. That is because the only interesting way to be wrong is to be wrong in a way which requires us to learn something new in discovering the error. Once we have learnt, the old way of being wrong is no longer interesting except as part of the history of science. I doubt very much that Li et al fall into that category.

Thank you, by the way, Karsten, for the plot of hemispheric temperatures vs sulfate concentrations. It is probably also worthwhile pointing out that nearly all of that variation in sulfate concentration is from either side of the North Atlantic; and that sulfate forcing is not global or hemispheric (unlike CO2), but regional. That is, the aerosol forcing is centered on the North Atlantic and is likely to show up strongest in the North Atlantic and adjacent lands. (That is why the recent uptick in aerosol forcing (from Brazil, India and China) is unlikely to have the same effect on hemispheric or global temperatures as the prior NA centered emissions.)

23. BBD says:

An interesting point emphasised by Tom Curtis there about the importance of regional variability in aerosol forcing (especially over the NA).

24. As there is an alphabet soup of climate oscillations – AMO, NAO, PDO, ENSO, AO, AAO, QBO, IOD, etc – finding one that is somehow correlated with the global temperature record does not seem to be particularly unlikely.
The credibility of any article that arbitrarily sets aerosol forcing to exactly zero should be exactly zero.

25. That would seem to be consistent with most of the other comments with regards to this article 🙂

26. Tom, very well put. I agree. Same is true for Richard Telford’s final sentence 🙂

Re spatial aerosol distribution: That’s exactly the point, it’s North American and European aerosols with the North Atlantic Ocean right in the middle, nicely “equipped” with a low albedo which makes sulphate aerosols a pretty effective short term forcer over sea. Sustain the forcing for years, and you get the typical circulation shifts accompanied with any other forcing, despite the rather regional aerosol origin. My current thinking is that the instantaneous aerosol effect projects primarily on the AMO, while the longer term (circulation) effects might project more on the NAO. Of course, they then mutually influence each other alike. The NAO then drives AMOC … and so on and so forth. It’s true that the AMO temperature signal is stronger than that over the adjacent land, leaving plenty of space for amplifying effects (probably due to changing AMOC strength), but I am convinced that volcanoes and anthropogenic aerosols are the main cause for the pseudo-oscillatory behaviour over the past 160 years or so. Unfortunately, you need GCMs to demonstrate that this is indeed the case.

27. Such statistical analyses can be a first step towards finding physical mechanisms. It is just nothing for press releases and blog posts.