Explaining too much

A couple of people have recently complained about the rash of stories explaining how snow-in-December (or, your pet weather event at whatever time of year) is compatible with Global Warming. For example: Cold Spells From Climate Change? (DA) or Yes, they have now said it (TW; I’m sure those two will love being associated).

What I think is correct is to explain that Yes, X has occurred, which perhaps you might not expect given GW, but then showing that it is entirely compatible with GW. Not that it is a particularly exciting game to play, because pretty well any form of weather will be; the variation in weather is so large, compared to the slow changes under GW, that it will be a long time before any given weather event we’re used to could possibly be ruled out.

What I think rather less correct is attempting to say Y has occurred because of GW; and that is so for pretty well the same reasons. And of course it doesn’t play terribly well either: “ha, they are saying that both heatwaves and freezes are caused by GW, hence it must all be nonsense”. Of course that isn’t what the science is saying, only that both are compatible.

So before explaining such-and-such an event, the first thing you need to do is to show that there is something in need of an explanation. A cold December in Europe doesn’t fall into that category. So when the NYT (ref’d by DA) says

Over the past few weeks, subzero temperatures in Poland claimed 66 lives; snow arrived in Seattle well before the winter solstice, and fell heavily enough in Minneapolis to make the roof of the Metrodome collapse; and last week blizzards closed Europe’s busiest airports in London and Frankfurt for days, stranding holiday travelers. The snow and record cold have invaded the Eastern United States, with more bad weather predicted. All of this cold was met with perfect comic timing by the release of a World Meteorological Organization report showing that 2010 will probably be among the three warmest years on record, and 2001 through 2010 the warmest decade on record.

How can we reconcile this?

The obvious answer is: Der, this is global warming, not local, and 2010 is more than just a few weeks. Annual cycles like El Niño/Southern Oscillation, solar variability and global ocean currents cannot account for recent winter cooling is probably true, but deliberately leaving out natural variability as an explanation is at worst ignorant. But all this is just a peg to hang some bloke’s forecasts off: Model accuracy demonstrated says “Judah Cohen, AER, Inc.” who looks to have consultancy to sell. Cohen issued a real-time winter forecast for Europe for 2010, which was significantly more accurate than those issued by the European forecast centers it says, but conspicuously fails to link to the forecast. Remind you of anyone called Piers? (that’s unfair, I know). FWIW the overall idea (less sea ice => more moisture => more seasonal snow => changes to the jet stream => pick your own favourite result, it could do anything) is at least plausible; I’ve no idea if it is also true.

Meanwhile, George Monbiot has another explanation which made Timmy sad, also keyed off the sea ice, but using a different mechanism (warmer Arctic => reduced pressure gradient between Iceland and Azores => pick your own favourite result, again, possibly mediated by the jet stream if it pleases you).

Or possibly the two mechanisms are, cunningly, the same thing in disguise. But at the moment, such things are subjects for research papers, not settled science. The correct (but boring) response is to go back to the beginning, which should always be, Is there actually anything in need of explanation?


* Environmental reporters ought to be more responsible too (Eric at RC)
* Snowfalls are now just a thing of the past
* Cold winter in a world of warming? – Rasmus at RC
* Joe Romm reports Munich Re talking out of its arse but unfortunately he doesn’t realise they are doing so.
* Ben Santer on the attribution of extreme weather events to climate change – ht: CP
* NCAR’s Trenberth on the link between global warming and extreme deluges at CP
* Are recent snows from global warming? – DA


  1. #1 BKsea

    Well said. I always say the biggest mistake in trying to convince skeptics of global warming is to try to tie individual weather events to it. For people who point out cold weather to negate global warming, an appropriate response is to ask whether they would consider a low tide to be evidence that sea levels are not rising.

    And, because the NYT article mentions Seattle, I can say that in addition to the unusual November snow, we also saw the highest temperature ever recorded in Seattle for November and December has yet to drop below freezing. Of course, this still says nothing about global warming.

  2. #2 John McManus

    Comments about forcasts and this winter are driving me crazy( a short trip I know).Winter is now but a week old . Things may change.

  3. #3 Adam

    “Winter is now but a week old”

    Winter is 29 days old, meteorologically speaking, and seeing as we’re talking about weather, probably the definition (of the possibly three main ones) to use.

  4. #4 Marco

    BKSea: you are right that the current temperatures in Seattle says little about global warming. However, it does show how easily people associate snow with cold.

  5. #5 Chris R

    Judah Cohen’s forcast for the US is here: http://www.nsf.gov/news/special_reports/autumnwinter/predicts.jsp

    [Yeees. I think that reports claim that A forecast model developed by AER scientist Judah Cohen has consistently achieved on-target forecasts for most major cities in the industrialized countries. is obvious nonsense - this is claiming a JFM forecast based on Oct Snow cover (and other stuff) and this just isn't plausible: the snow cover might explain some part of the variability but certainly not enough to claim consistently on-target -W]

    He examines the prediction in this paper:
    Winter 2009–2010: A case study of an extreme Arctic Oscillation event.
    Available here:

    [Yeees, again. So that contains stuff like One of the earliest studies linking autumn snow cover with winter climate was that of Foster et al. [1983], who found that Eurasian autumn snow cover extent (SCE) explained as much as 52%of the variability of ensuing winter temperatures locally (my bold). That is the sort of effect I could believe; but being able to forecast remote temperatures with precision? No.

    In fact, now I look at the paper, I can’t see even the most basic step: that of determining the degree-of-forcing: ie, doing the forecast runs 10-20-50 times and seeing how reproducible they are -W]

    I got the feeling reading paragraph 14 that there was a bit of salemanship going on. The paper as a whole is interesting. Cohen has done a lot of previous research on snow as a boundry condition affecting the atmosphere. I think he has a point, however…

    1) I didn’t think Cohen extricated the cause/effect between increased snow cover and Arctic sea-ice. Although he’s interesting regards the connections throughout the atmospheric column – the stratosphere appears to be driven from the troposphere, which argues against a solar UV effect.

    2) Modelling studies like Deser “The Transient Atmospheric Circulation Response to North Atlantic SST and Sea Ice Anomalies”, which also references other pertinent studies. Find a -ve NAO type response to low sea ice anomalies. i.e. without recourse to snow anomalies.

    I’ve a couple of interesting looking Cohen papers to read, I do think he’s found a real effect but I’m not sure what part of the NAO response is due to sea ice, what part due to snow cover, and I suspect the latter is arguably a result of the former.

    [I'e seen too many papers say things like The pattern of variability across the NH was well predicted by the model, though it incorrectly predicted above normal temperatures across southeastern Russia and especially in Mongolia - ie, the model was right, except where it was wrong; and near-meaningless claims like "well predicted" -W]

    Last year was the lowest AO in the 61 year series, this year looks to be on track for a repeat – we’ll see. If the Feb 2010 AO index of -4.2 is neared (< -3) or beaten this year it does demand an explanation.

  6. #6 Tim Worstall

    Just to point out…..the major point of my own post on this story was to celebrate someone using a picture of a Zebra in the snow as an illustration of global warming.

    Whatever one’s view of anything else, that is worth at leaswt a cocked eyebrow, if not a giggle, no?

    [Perhaps a little; perhaps that kind of denialist stuff is new and amusing to you; but I've heard it all before. As I said on your blog, your post came across as rather unthinking, which was disappointing, and your subsequent defence got rather worse: you appear to have swallowed some denialist talking points whole. I'm sure you wouldn't hesitate to attack anyone doing the same with economic theory -W]

  7. #7 J Bowers

    I’ve found that pointing to science that might explain how these winter extremes could be tied to global warming only inspires cognitive dissonance, where the plastic sceptic jumps to the conclusion that the claim has been made that this and last year’s extreme winter is because of global warming. It is worth pointing out, though, that the sea around the British isles has yet to freeze over and telephone poles fall down into 6 metre high snowdrifts, as happened in 1963, and that currently we are only talking about a week in February 2009, last winter, and the start of the latest (astronomical) winter so far. That would be less than two complete astronomical winters or two full(ish) meteorological winters, and not the three winters claimed by many of the aforementioned plastic sceptics I’ve been having this ding-dong with. Conversely, they are using these cold winter spells as proof of global cooling. Go figure. The asymmetry burns.

  8. #8 J Bowers

    It might be worth pointing out Stephen Leahy’s article of last June on the International Polar Year Oslo Science Conference, with regards to comments made there by Barber, Overland and Sereze.


  9. #9 JM

    Perhaps this is a bit pedantic, but seems to me that attributing single weather event causation to “global warming” is not only incorrect, it is also entirely conceptually wrong.

    Since global warming is a long term accumulation of local weather anomalies in a given (warming) direction, one can then only say that a change in frequency and severity of an ensemble of weather events is a reflection of an underlying global warming trend. Saying that global warming “causes” a snowstorm is saying that a statistical average “causes” an instance of a single event being averaged — conceptually incorrect as there is no causation arrow to apply.

    [I think that is reasonable -W]

  10. #10 yogi-one

    Granted, we need more years to provide any strong correlations. But in the decade of 2000 – 2010 we already have three good case-studies – the winters of 2005-6, 2009-10, and now 2010-11 that are providing data showing a relation between Arctic warming and colder winters with fiercer storms.

    Also, I think the summers should be studied in this regard. we’re seeing record high temps in the summers as well as record lows in the winters. Has anyone studied/published on this side of the issue?

    Actually, I was happy to see the NYT article, and welcomed the change in tone. You know, they could have printed the headline “scientists say not so fast, global warming cannot be tied to weather events” and trotted out somebody from the Cato Institute to show us all a couple of doctored up powerpoint slides “proving” that AGW is a non-issue.

    In terms of the propaganda war around AGW, I consider a shift in tone on the issue by a major international news source such as the NYT progress, at least on the PR front.

  11. #11 Anna Haynes

    How about Sharon Begley’s take, in Newsweek, on fractional risk attribution (link)?
    (“The idea is to calculate how many times an extreme event should have occurred absent human interference, explains climate scientist Ben Santer of Lawrence Livermore National Lab, and the probability of the same extreme event in today’s greenhouse-forced atmosphere. Result: putting numbers on extreme weather.
    …Peter Stott of the British Met Office…”)

    [Yes, perfectly sane article. But it still won't allow attribution of individual events, for the aforementioned reasons. People will, as it says, still desperately *want* answers to random weather; and the ill-informed will probably give answers; but they won't be right -W]

  12. #12 J Bowers

    William, but what about the “well informed” who have the credentials to make a judgement call, or at least better than most others? There may well be a time when we don’t have the luxury of ~9 month’s peer review to make decisions on. The bummer being that those who call for inaction until more evidence is produced is similar to the many French getting their evidence when it marched into Paris under a German flag. Is it possible that the tragedy of the commons is a similar hindrance to wise advice within the academic community as it is amongst the Tea Party?

    (Yes, I’m at the pub. Feel free to delete. ‘Tis the season for it.)

    [The luxury of waiting for peer review isn't too relevant I think: after all, we already have the science we need. This is far more a matter of feeding the meeja something new to say. If the science-side doesn't, the septics will be very happy to -W]

  13. #13 Anna Haynes

    Fractional risk attribution, Santer via Piltz via Romm (link)

  14. #14 blueshift

    Is it terribly wrong to say that weather now is the always the result of AGW *and* natural variability?

    That’s my sound bite take away from this CP post:


    [My opinion is that while you're OK to take your news from CP, you should be cautious of taking your science from there. In this particular case, the quote from Trenberth is fine (if perhaps a trifle over-enthusiastic) but the over-bolding and over-interpretation at CP is less good.

    Notice how Trenberth echoes my language (or I his) with that’s consistent with the expectations with regards to global warming (my bold) that is the key point: these events are*consistent* with GW, but by no means prove it, nor can they be said to be individually caused by it.

    I'm not sure I'm really happy with weather now is the always the result of AGW *and* natural variability? - it isn't wrong, but it isn't how I'd phrased it. The climate we have now is the result of whatever - 0.7 oC, say - warming over ~1900 (so in a rather imprecise sense could be called preindustrial+AGW) - and that climate is the average (well, the statistics of) the weather.

    Probably that is unclear. Trying another simplification: suppose we regard the "climate" over the last 100 years as a linear trend of increase of T_global at 0.1 oC / decade. And then the "weather" as that average, plus zero-mean normal-noise with a deviation of 10 oC. Then you can see at once that the "weather" in 1900 is very nearly the same as the "weather" in 2000 and it would be impossible to pick one event from either end and say it could not have happened at the other end. I made up the numbers, of course, but they are not wildly unrealistic -W]

  15. #15 Chris R

    Seems I missed off the last part of my comment #5, was going to say it’ll be noteworthy if extreme low AO occurs again this year.

    Thanks for the observations re Cohen’s paper. I agree w.r.t. his ‘predictions’ but snow cover will have an impact on the atmosphere which could teleconnect, as I said above, I don’t think Cohen has demonstrated how much of an effect.

    [Yes, I think we agree there. The effect is entirely likely; that it is a major source of predictability is unclear -W]

    Regards attributing individual events.

    If you’re examining the probabilities of flipping a coin you need long runs to say anything with statistical confidence. If you then apply a blob of solder to the ‘tail’ face of the coin it’s reasonable to expect a change in the statistics for runs after the solder is applied. You still need more runs to establish the new statistics and you still can’t be sure every flip that gives a tail is due to the solder.

    But the action of applying solder can be predicted to change the statistics even before you do the additional runs of coin-flipping. In this case you can be sure that the solder on ‘tails’ will increase the probability of getting a ‘tail’.

    [Agree entirely. In exactly the same way, a small warming affects the statistics of warm and cold years -W]

    I think that Overland is correct and that we’re going to see more instances of low AO values, hence colder winters for swathes of the NH. In effect I think the Arctic has applied solder to one side of the climate coin.

    [I'd be rather more cautious about that bit, if only because I haven't really read the stuff through properly -W]

    I don’t think that this means ‘Climate Change -> More Colder winters’ as that misses a crucial element in the causal linkages – the role of sea-ice. I do think that ‘Climate Change -> Reduced Sea Ice -> More Colder Winters’, because the loss of Arctic sea-ice is the crucial element.

  16. #16 adelady

    “Reduced Sea Ice -> More Colder Winters” may be true, but only as a transitional state.

    Once the Arctic is more or less ice free for most of the year, things will change again. Because the “lid” of ice over the warmer water will not be lifted, lowered, lifted again. It will be absent. Whole new game.

    [The Arctic will not be ice-free in winter this century -W]

    The issue is how long the transition will last. Then the issue will be trying to predict new circulation patterns in the Arctic area itself and its neighbouring areas.

  17. #17 crandles

    Since fractional risk attribution has been mentioned:


    Is it foolish to believe the results if they come from a single climate model run many times? To what extent is use of several different GCMs needed? I.e. Is use of one model a reasonable guess of unknown reliability while adding extra models increases the reliability?

    Or would it remain a hopeless guess if you used say 5 models but all 5 appear to have natural variability of less than the real world? Or can this be adequately accounted for by watering down the results to the extent necessary to get to a 95% confidence level?

    [I'd say you can happily get away with just one model (run many a time) providing it is a good one. Inevitably it will miss some things, but then so will more than one model. JA might well have a more informed perspective on this (mine was always that HadXM3 was "good enough", but then again it was the model I was using). Certainly for things like "do changes to Siberian snow cover reproducibly affect non-local winter climate"?

    I couldn't look at your link - is the site down? -W]

  18. #18 Steve Bloom

    This recent Trenberth presentation should be of interest.

  19. #19 Eli Rabett

    [The Arctic will not be ice-free in winter this century -W]

    Apres moi le deluge:)

    [I doubt I'll last that long -W]

  20. #20 Eli Rabett


  21. #21 Steve Bloom

    IIRC Louis XIV died about 80 years before the French revolution, so it’ll be a close thing.

    Quibbles about dates aside, will NP deep water formation start before or after the Arctic goes ice-free in winter?

  22. #22 eachran

    Whether the weather be fine,
    Or whether the weather be not,
    Whether the weather be cold,
    Or whether the weather be hot,
    We’ll weather the weather
    Whatever the weather,
    Whether we like it or not!

    But I cant say the same about the climate.

    Steve Bloom and Eli Rabett, the Sun King has been reincarnated as Mr S, please keep up. And dont forget it woz the sun wot dun it.

  23. #23 eachran

    Steve Bloom : And wasnt it Louis XV?

  24. #24 Steve Bloom

    You’re right, eachran, which makes it 15 years, so allowing for a normal lifespan for Eli we should be looking for trouble @2050, and for William maybe a decade later, not that the specific span between XV’s expiration and the revolution is much of a guide.

    Actually according to a couple of the sources I saw XV may well have plagiarized the line from one of his girlfriends, the much cleverer Madame du Pompadour.

  25. #25 eachran

    Steve Bloom, the really clever girl in France was Madame Chatelet


    She used to hang around with that geezer Voltaire (or perhaps it was the other way around).

    But knowing now what one does about the lady it looks like Voltaire’s fame was largely her doing. She was clearly something special.

    Sticking to ladies there was that other very impressive one who shacked up with Napoleon, Josephine


    She put it about a bit and also for Napoleon’s benefit.

    Enough of birds.

    But it does work the other way around too. Some famous women would not be where they are today without the masculine equivalent of Madame Chatelet.

  26. #26 blueshift

    Thank you for the reply WMC. I was hoping I could have something a bit easier to get across to a stereotypical guy at the bar- oh well.

    One thing I’m not quite clear on. You say the 10 C is the standard deviation (in this hypothetical formulation). That would be for local temperature?

    [Where "local" would mean, err, I don't know: England, perhaps. Don't take the exact numbers seriously, of course. All I meant is that the weather noise is much larger than the trend -W]

    Oh and don’t worry I get all my blog science from Inferno ;)
    (Actually, here, RC, Tamino and Eli Rabbett.)