Friday, July 31, 2015

Surface temperature is not so different from models

There's a new paper out in Geophysical Research Letters, which looks at global surface temperature. This time the purpose is to compare the observations with climate models. What's interesting is that the authors picked up something that I didn't know about, though scientists probably did. The surface temperature reported from climate models isn't the same as the surface temperature reported as observations. Almost, but not quite.

When an apples to apples comparison is made between climate models and observations, then one third of the discrepancy disappears. Here is a chart from Kevin Cowtan that illustrates this. (Read on below for further explanation).

Figure 4: Global mean temperature calculated from an ensemble of climate models using the simplest method (red), or using the same method we use for the observations (blue), and some intermediate methods. (Note that all but the blue line are global coverage.) Source: Kevin Cowtan

Kevin Cowtan is the lead author of an all star team, many of whom you'll probably recognise: Zeke Hausfather, Ed Hawkins, Peter Jacobs, Michael E. Mann, Sonya K. Miller, Byron A. Steinman, Martin B. Stolpe, and Robert G. Way.

Kevin Cowtan wrote about this new paper on his blog. He explained how the results of models aren't reported the same way as observations. I've put the differences in the table below. In the models, the ocean surface temperature is measured by the air above the ocean surface. In the observations, it's the sea water temperature, not the air temperature, that's recorded: 

 
Land
Oceans
Models
Air above the land surface
Air above the sea surface
Observations
Air above the land surface
Water at the sea surface

As Kevin Cowtan wrote:
While the pattern of warming in the 20th century is similar between the models and the observations, over recent decades the models show a little more warming than the observations. We don't expect individual model runs to match one another, because they have different weather. Similarly, the real world has its own weather, so we don't expect it to match the mean of the models. However it is legitimate to ask if the difference between the models and observations can be explained by weather alone.

Part of the explanation lies with recently addressed problems in the observations (Cowtan & Way 2014, Karl et al 2015) and a temporary slowdown in warming associated with changes in ocean circulation and other natural factors (see this lecture and commentary). However we have identified another factor not included in the IPCC graph and similar comparisons: the graph is not comparing the same things. The results from the models use air temperatures over the whole globe, whereas the observations use a mix of air temperature over land and water temperature for the oceans - Figure 2.

One of the first questions that occurred to me was why this difference is only showing up now. Why were past observations so close to the models? Well, the answer lies in the Arctic, where summer sea ice has been declining rapidly. Kevin put up a diagram to illustrate:

Figure 3: As sea ice retreats, some grid cells change from taking air temperatures to taking water temperatures. If the two are not on the same scale, this introduces a bias. Source: Kevin Cowtan

The bias is introduced because now that the ice is melting, sea surface temperature observations are being recorded in areas that were previously covered in ice. This has the effect of lowering the observations compared to the models. The discrepancy doesn't start to show up until around 2005.

Kevin Cowtan's blog article gives a very clear explanation of the work - you can read it here. Another co-author, Ed Hawkins has an article about it too.


Reaction from conspiracy theorising science deniers


As expected, there is the usual huffery puffery nonsense from science deniers. Anthony Watts must have been in a hurry because he just posted the press release without his usual "claim" headline. He did try to mislead his readers though (without much success), by posting one of John Christy's wrong charts instead of a chart from the paper itself.


From the WUWT comments


Taylor Pohlman doesn't understand that a large part of the discrepancy is because of the recent melting of Arctic sea ice, so there is no major discrepancy in earlier years:
July 30, 2015 at 9:31 am
Because the models are ‘cooked to hind cast this period. The thing is, if the correction he’s talking about gets made to the 1975-1995 peri on, it will then likely be to low to match actual a – like the preverbial thread, this thing just keeps unraveling…

Doug Sorensen cries "squirrel", pointing to the troposphere and away from the surface:
July 30, 2015 at 9:24 am
I think Dr. Cowtan may be juggling his apples a bit. GISS and HADCRUT use “a combination of air and sea surface temperature readings.” However, satellite and balloon data do not. They measure as much of the troposphere as they can, which makes them very much like the model output.

BeefArt asks a useful question, which gets some unhelpful answers:
July 30, 2015 at 9:38 am
Curious why you didn’t use any of the actual graphics from Cowtan et al. They show something dramatically different from Christy’s graphic that you used. Seems a little misleading, don’t you think? 

markstoval wrongly thinks this research was funded by a grant. It wasn't. (Why don't fake sceptics publish their findings if they are so sure of themselves?)
July 30, 2015 at 10:02 am
Dr Cowtan said: “When comparing models with observations, you need to compare apples with apples.”
You do need to compare apples to apples which is why Dr. Cowtan knew to compare the apples to orangutans. (and please send more grant money)

Most of the other comments are equally dismissive and many include conspiratorial accusations of fakery and fudgery. (Anthony Watts blog is a haven for conspiracy theorising science deniers. That's his target audience. Don't expect to ever read anything rational there.)

References and further reading


Kevin Cowtan, Zeke Hausfather, Ed Hawkins, Peter Jacobs, Michael E. Mann, Sonya K. Miller, Byron A. Steinman, Martin B. Stolpe, Robert G. Way (2015), "Robust comparison of climate models with observations using blended land air and ocean sea surface temperatures", Geophys. Res. Lett., 42, doi:10.1002/2015GL064888. (pdf here)

26 comments:

  1. Captain FlashheartJuly 31, 2015 at 3:21 PM

    How come no one thought of this before!?

    Also, it appears Monckers has got himself an iPhone. The horror!

    ReplyDelete
    Replies
    1. Good question, CF. I understand that a couple of different people thought of doing something about this around the same time. Kevin Cowtan has earned a reputation for thoroughness, rigour and attention to detail, so the different people joined forces once they found out Kevin was investigating this, and different authors contributed different things to the research.

      The thing is that while the fact that there was a difference between the way models and obs were recorded was known, the data only started diverging markedly after around 2005, so the impetus to work out the discrepancy wouldn't have been there in the past.

      Also, it's one thing to know it in theory, but quite another to go through all the data and quantify the difference. The size of the discrepancy is probably a surprise to some people The paper reports:

      "Applying the methodology of the HadCRUT4 record to climate model temperature fields accounts for 38% of the discrepancy in trend between models and observations over the period 1975-2014."

      As I understand it, a lot of the work had to be done on a grid by grid basis, working out where there was ice. As well as for a number of different models and model runs. That's a lot of data to work through. And if you read the paper, they examined it in multiple different ways.

      My blog article doesn't do it justice. I'm seriously impressed with the amount of work they did and the very thorough approach. Kevin Cowtan is cementing his reputation as a serious contributor to climate science, which isn't his main field of research. You could call it a hobby - or second vocation.

      AFAIK, this was an unfunded project - though the acknowledgements state that various research institutes allowed use of their computers etc.

      Delete
  2. *cough* Geophysical Research Letters *cough* - in your opening lede

    ReplyDelete
    Replies
    1. Thanks. Fixed. (I'm having an off day today, as you can probably tell.- with apologies to the authors especially.)

      Delete
  3. Tamino also covers the research, with his usual helpful statistical analysis. See here.

    ReplyDelete
    Replies
    1. Already listed in refs above, John. (But thanks anyway :D)

      Delete
  4. Rob Honeycutt posted at Tamino how Watts passes of Christie's UAH graph as Cowtan's | July 30, 2015 at 7:23 pm | Reply


    This one is already causing copious conniptions at Tony’s place.

    Instead of using the actually charts in the Cowtan et al paper, he replaces it with a bizarre John Christy chart that compares tropical (20S-20N) satellite data to an average of 102 CMIP5 models. And he gives that graph a highly misleading title that would lead the unsuspecting to believe it actually came from the paper.

    Instead of apples and oranges, Watts has moncked (my own invented term) this one up to the point that he’s deceptively comparing apples and aspartame.

    ReplyDelete
  5. Two comments in the Watts article from noaaprogrammer that really just contradict each other. Not really joined up thinking! Do you think he is trying to sound knowledgeable and sciencey?

    First:

    noaaprogrammer
    July 30, 2015 at 4:31 pm
    Furthermore, they do err in trying to model chaotic phenomena with traditional, non-chaotic math.


    Followed by:

    noaaprogrammer
    July 30, 2015 at 4:42 pm
    I wonder if a simple neural net program, trained on (proper) historical temperatures would do better than these models? (Of course the interval of time over which it is trained would make a significant difference.)



    ReplyDelete
    Replies
    1. Yep. Typical engineer/programmer thinking that because they work in a technical field, that they know as much/more than the scientists.

      I'm an electrical engineer/programmer as well, but I *think* I know my limitations. For one thing, I know that feedback as applied to electronic circuits does not even remotely work the same way as feedback mechanisms in the climate system. That seems to be one of the engineering types biggest misconceptions.

      I guess (assuming you have the requisite level of acquired knowledge to be able to grasp the science at a fundamental level) how well you comprehend climate science is largely a function of how much you are prone to the D-K effect, and how much you allow your ideology to influence your understanding of the subject.

      Delete
    2. If he is a noaaprogrammer, he probably handles business systems...

      Delete

    3. I like the idea that somewhere there is a "proper" set of historical temperature data.

      Also his caveat about not having too short a record for significance of some sort. That obviously does not matter when talking about a "pause"!

      Delete
    4. Hey, I'm a programmer too. That means I'm able to throw a neural net at the problem and see what comes up:
      http://psychic-vr-lab.com/deepdream/pics/670884.html

      Delete
  6. What are models reporting when they hindcast? They're usually very good at doing that.

    ReplyDelete
    Replies
    1. Never mind, perhaps. If there was a bias that started around a particular year, then the hindcasting prior to that should be good still.

      Delete
    2. And wow, I literally just went through the first "Q&A" as the first WUWT commenter. *sigh*

      Delete
  7. This is the sort of work that really highlights the value of models. Without a model you don't know (i) what you expect within the context of your understanding, (ii) you don't actually really know what you know or what your observations mean and (iii) you don''t really have much of a focus for investigating further.

    The dismal flaws in the (Spencer and Christy) radiosonde temperature measurements were uncovered due to the mismatch with models. Likewise the apparent mismatch between modeled temperature response to greenhouse forcings and measurements in the present case has provided a focus for investigation that highlights (again) some problems with the measurements, or at least in the manner that measurements are compared to models.

    So another hip-hip-hooray for models.

    ReplyDelete
  8. It's been a pretty tough month for Willard. First he loses his beloved pause and now he loses his model/data mismatch. I am crying for him, if I could, but the drought here in the west is too intense to waste the water!

    ReplyDelete
    Replies
    1. I wonder if this will make the cheesy trick of misaligning model/data obvious to even his dorkish followers.

      Delete
  9. Sou, I sent Martin Armstrong links to your most popular (and powerful) stuff. He hasn't been doing anything denier in a long time. Hopefully, he is studying your fine work.
    "There are really no past cycles to look at of polluting our atmosphere and oceans to anywhere near what they are now." I told him.

    Martin has repeatedly blamed Bill Clinton for rolling back Glass-Steagall. I also sent him proof that Glass-Steagall was only on life support after Ronald Raygun. He's been quiet on that one as well. Maybe we're getting noticed...

    ReplyDelete
  10. I'm moved to make a potentially inflammatory observation.

    When it becomes obvious to even the deniers of climate change that humans are changing the climate, we've long passed the point at which we can safely avoid serious disruption of our way of life, and perhaps even avoid the effective destruction of the same.

    That understanding should really inform whatever rear-guard action we take as a species, and the chances of pulling out of the nose-dive are directly proportional to the extemity of the urgency that may (or may not...) infuse those actions.

    Of course, YMMV, but time will tell.

    ReplyDelete
    Replies
    1. YKMMV your Koch money may vary...The GOP leaders signed a pledge to the Koch Brothers to do nothing.

      Delete
  11. Yes, I had wondered about the sea surface temperatures - they will not be the same as the air temperatures above the sea surface.

    I recall the GISTEMP doco said the difference was not all that important as they convert the sea surface temps into an anomaly, and there is reasons to treat the sea surface temperature anomaly as being the same as the anomaly in the air temperatures above the sea surface.

    All the more reason to not get hung up on the absolute average global temperature, but to concentrate on the anomaly instead?

    ReplyDelete
    Replies
    1. I am going to have to think about this some more.

      So they are saying that some air temperature observations have switched over to sea surface temperature observations, as measured at the ship engine intake ports etc?

      Delete
    2. Oh and the John Christy paper - it was written for a congressional hearing wasn't it? Are such papers peer-reviewed? Not to mention starting each series at zero, and not mentioning the baseline used or the confidence interval.

      Delete
  12. Are there CMIP-5 model output datasets that are narrowly targeted at layers of troposphere above “surface,” against which the RSS and UAH variety of indices properly should be compared?

    ReplyDelete

Instead of commenting as "Anonymous", please comment using "Name/URL" and your name, initials or pseudonym or whatever. You can leave the "URL" box blank. This isn't mandatory. You can also sign in using your Google ID, Wordpress ID etc as indicated. NOTE: Some Wordpress users are having trouble signing in. If that's you, try signing in using Name/URL. Details here.

Click here to read the HotWhopper comment policy.