A Few Things Ill Considered

How to talk to crakar – point 2

Crakar said:

We are shown the results of computer model programmes that predict an apocolyptic future, these programs are based on modelling 16 (yes thats right only 16) parameters, many are considered by the IPCC as having a very low and low level of scientific understanding. Do the models incorporate the ocean cycles? or the atmosphere/ocean interactions? No they dont. There are many more parameters that they do not incorporate, but wait thats not all. We are expected to believe in these computer programmes because the IPCC scientists are in general agreement with them even though they predict a hot spot where no such thing exists. Without a hotspot there is no global warming.


I think crakar means 16 forcings are incorporated into model hindcasts and future projections. This is more or less in accord with what I have read. I am curious as to which important forcings have been omitted? I don’t find that a shockingly low number and where is the reason to believe it has not been very carefully and fully considered by climate modelers?

The next part of his point is the appeal to uncertainty. It never fails to amaze me how this argument can be raised in completely obliviousness to the fact that uncertainty cuts both ways. So we don’t have a firm grasp on the complexity of clouds in the climate system, why must that mean that models are over estimating the potential warming? They could as easily be underestimating. In fact, we have a great example of an uncertain set of models drastically underestimating change in the arctic sea ice. Not one credible scientific source (er..that I know of…) from, say, 10 years ago was predicting the minimum extents and volumes of arctic sea ice we are seeing now. Five years ago? I don’t think so. These models failed, but they did not fail on the side that climate ostriches assume they must.

As to the hotspot, we had a thread on this already. This is an example of using an uncertain and developing area of research to draw wildly over confident conclusions. For starters, it is no simple matter measuring temperatures in the troposphere. The WUWT crowd loves to point out how unreliable a good old fashioned thermometer sitting in a box can be, but have not a whit of hesitation in accepting readings from weather ballons that can’t control their altitude or orientation and satellites trying to measure microwave radiation from all levels of the atmosphere as a proxy for temperature. Those are hard problems.

But the key fallacy of the missing hot spot argument is that it is the finger print of CO2 enhanced warming, it is actually an expected signature of warming from many causes, including solar driven warming. Stratospheric cooling on the other hand, well documented by those favoured satellite measurements, is distinctive and related to an enhanced greenhouse effect. It is not related to solar driven warming, warming caused by ocean cycles like the PDO or warming caused by galactic cosmic rays. This cooling is very clear and very pronounced.

Without a hotspot there is no global warming.

And yet…

(image taken from this page)

Comments

  1. #1 koen
    October 22, 2009

    Crakar speaks from utter ignorance, and seems not willing to learn. Why waste time on that?

  2. #2 carrot eater
    October 22, 2009

    One must invite crakar to clarify his paragraph; otherwise you have to guess at what he was trying to say.

    “Do the models incorporate the ocean cycles?….No they dont ”

    They don’t?

    Ocean cycles aren’t hard-coded in there, nor would you want them to be. You’d want ocean dynamics to be a model output, not input, and that output is getting increasingly better – you see ENSO-like dynamics in the model output. Still a lot of room for improvement, but let’s not say there’s no ocean oscillations in the models.

    “or the atmosphere/ocean interactions? No they dont. ”

    You’d need crakar to clarify what he means here. What does he think the ‘coupled’ in coupled model refers to? What specific interaction is missing?

  3. #4 carrot eater
    October 22, 2009

    dhogaza:

    I see we said much the same thing.

    craker’s statement sounds like he’s repeating something incorrect, but doesn’t even quite understand what he’s repeating, anyway.

    Throughout history, there were indeed uncoupled models, so maybe this is a denialist talking point from 1984 that just never went away. Though Manabe’s first attempt at a coupled air-ocean model goes back to the 60s even, I think.

    As for the ocean dynamics, a lot of denialists are a priori convinced that ocean cycles cause global warming, so they disbelieve any model that doesn’t tell them what they want to hear.

  4. #5 dhogaza
    October 22, 2009

    I see we said much the same thing.

    Yeah, I posted to a stale window and didn’t see your post …

    Throughout history, there were indeed uncoupled models, so maybe this is a denialist talking point from 1984 that just never went away.

    Maybe so, or perhaps a misunderstanding due to how HadCrut initializes its model when running it for weather forecasts (AFAICT from reading their documentation, they take a snapshot of ocean conditions near the UK and just run the atmospheric side, apparently because ocean weather changes much slower than atmospheric weather, and they are running the model frequently in order to get near-term forecasts for the UK).

  5. #6 carrot eater
    October 22, 2009

    I didn’t know that, but I’d be amazed if that were the basis of a denier talking point.

    It seems reasonable, though. I doubt a meteorology model used for forecasting weather in, say, New York City, really needs a coupled global ocean that replicates ENSO. On the time scale of 3-5 days, that’s entirely unnecessary.

  6. #7 dhogaza
    October 22, 2009

    I didn’t know that, but I’d be amazed if that were the basis of a denier talking point.

    True, it would require far too much real knowledge, wouldn’t it? :)

  7. #8 Robert Grumbine
    October 22, 2009

    carrot:

    The numerical weather prediction (NWP) models of, say, 10 or more years ago followed your assumption. Any more, however, and there’s more interest in and more involvement of the ocean — even for the 5 day models. Once you get that right, an ENSO model is not far behind (and vice versa). 15 years ago, I surprised the NWP folks by showing that modest changes to sea ice reflectivity actually did matter to the short term forecasts (in fact, even at 24 hours it was noticeable). They figured, quite reasonably, that there wasn’t much ice and the changes I wanted to make weren’t very large, so the small bit of energy redistribution wouldn’t matter. I showed otherwise, published, and the change went in to the model. Turned out that marine forecasters noticed an improvement in storm tracks for systems heading up the North Atlantic and up around Iceland — something I hadn’t thought to verify.
    http://ams.allenpress.com/perlserv/?request=get-abstract&doi=10.1175%2F1520-0434%281994%29009%3C0453%3AASIAEW%3E2.0.CO%3B2

    Nobody argued against modest changes in reflectivity (they were say from 0.5 to 0.55 or vice versa) being important for climate. But such things do turn out to matter for weather.

    This actually looks like something good to take up for a full length post over at my blog. If I don’t sometime reasonably soon, remind me — the ‘question place’ article is the perfect spot to drop a topic suggestion.

  8. #9 carrot eater
    October 23, 2009

    Robert, thank you for that. Lesson learned: don’t underestimate how much pretty much anything could affect a weather model.

  9. #10 dhogaza
    October 23, 2009

    This actually looks like something good to take up for a full length post over at my blog.

    That would be very interesting, please do so.

  10. #11 Robert Grumbine
    October 23, 2009

    Thanks to the thunderous demand (my blog being quiet enough that 1,2 = thunderous), I’m writing something up at my place. It’ll appear Monday morning (26th).

  11. #12 carrot eater
    October 23, 2009

    Robert, I actually look at your blog every day, because I love the way it tells you who else has made a recent post. But your own content is great if sporadic, so I look forward to it.

  12. #13 Marcus
    October 26, 2009

    FYI: some models for shorter term or historical simulations are run with “prescribed” oceans: rather than try to simulate the whole ocean, the model just uses the ocean surface temperatures as an input. For historical runs, this will likely outperform fully-coupled models. And for future runs of short time scale, this will probably do as well as fully coupled models too (since the fully coupled models aren’t going to get ENSO matched up properly with reality, absent a Keenlyside type analysis). Where this fails is in long term simulations where one expects there to be significant atmospheric influence on the oceans, such that the oceans are no longer like their historical state…

  13. #14 dhogaza
    October 26, 2009

    FYI: some models for shorter term or historical simulations are run with “prescribed” oceans: rather than try to simulate the whole ocean, the model just uses the ocean surface temperatures as an input….And for future runs of short time scale, this will probably do as well as fully coupled models too

    From my reading of Hadley documentation, this is exactly what they do for weather forecasting. Same atmospheric model that they couple to the ocean model for climate (long-term) work, but for weather forecasting they essentially set the ocean to an approximation (since observation data is limited) of its current state.