Tuesday, January 24, 2017

How Are Climate Models Doing With Global Temperature?

Quite well, in fact.

Here are surface temperatures compared to CMIP5 models:


"CMIP" stands for Climate Model Intercomparison Project. It's a database containing results from climate models for many different scenarios, that climate modelers use to compare their results to each other and to observations. It's now on its 5th version. I talked to a climate modeler a while back, and she said modelers are usually more interested in how the models compare than how they match reality. The models all make different assumptions, so learning which model best forecast, say, the Atlantic Meridional Overturning Circulation, can give good clues. Or modelers might replace one component of their model to see how it compared to the earlier version.

Clearly, climate models are doing a pretty good job. At least for global mean surface temperature.

The model results for years prior to 2017 use "historical" forcings -- the actual forcings that occurred then, based on the measured values of greenhouse gases, etc. That's only fair -- it would make no sense to compare a 1990 model with a scenario that didn't happen. After all, no climate model can foresee the future of GHG emissions, volcanic eruptions, and changes in solar irradiance.  

And here are the results for HadCRUT alone:


Here are some explanations and caveats on the assumptions and methods, from Peter Stott. Worth reading.


13 comments:

Victor Venema said...

"The model results for years prior to 2017 use "historical" forcings"

As the top figure and Peter Stott's explanation say: the forcing prior to 2005, not 2017 are historical. Not that is matters too much, most of the warming since 2005 will be due to greenhouse gas emissions before 2005 because the climate system only responds slowly to what we do like an oil tanker.

David in Cal said...

It's reassuring that the models have done a fairly good job of predicting temperatures. But, simple extrapolation of the past data would also have done a fairly good job. The models will be more validated when they correctly predict something unexpected.

Layzej said...

In 1981, after four decades of this, Hanson predicted this. People must have thought he was crazy. The linear trend suggested that we'd hit peak temperature back in the 1940s and were headed for further cooling. Yay physics!

PaulS said...

Layzej/David in Cal,

Also, Hansen 1988. Although the paper and presentation happened in 1988 the model was finalised and runs started in 1983, so only with visibility of data up to 1982. I've plotted here observed trends of varying length ending in 1982 (x-axis is trend start date) using the current Cowtan and Way record and current GISS met-station only data, which was what was used at the time. I've gradually adjusted the GISS record down between 1950 and 1980 to better reflect the dataset available in 1982 which showed a greater cooling trend from 1940 to the 1970s.

Also shown as asterisks are trends for 1983-2016 from the projection, against CW and GISS-met records.

The highest 1983-2016 trend a reasonable naive extrapolation would suggest is 0.05K/Decade, whereas Hansen's physical model predicted over 0.2K/Decade. Observations show trends a little over 0.2K/Decade. So, naive extrapolation predicts a trend magnitude 25% of observed. Is that fairly good?

The models will be more validated when they correctly predict something unexpected.

I'm not sure what this can really mean. In terms of the actual science world, what the models predict is almost by definition what's expected. If you mean something "skeptics" don't or didn't expect, I think comparison with naive prediction is validation already. Perhaps you could define what you don't or didn't expect, and reasons for that?

David in Cal said...

What I have in mind would be something analogous to what happened in 1919. "Eddington compared his eclipse photos with images taken when the sun was not present, and announced that the sun had caused a deflection of roughly 1.61 seconds of arc, a result that was in agreement with Einstein’s prediction, thereby validating the theory of general relativity." http://simonsingh.net/media/articles/maths-and-science/1919-eclipse-and-general-relativity/

What an analogue would be for climate change I can't claim to know.

David Appell said...

David in Cal said...
"The models will be more validated when they correctly predict something unexpected."

Models don't and can't predict. In principle. They project.

No model knows the future of human emissions, volcanic events, or solar changes.

But they do a good enough job of projecting into the future, as my post shows.

Models don't have to be perfect to show us that a lot of warming is in our future.

David Appell said...

David in Cal wrote:
"What an analogue would be for climate change I can't claim to know."

Then why bring up Eddington?

David Appell said...

Nice comment, Paul. Thanks.

Ned said...

David In Cal wrote: "What an analogue would be for climate change I can't claim to know."

You should have read the two comments preceding yours; then you would know.

Broecker (1975) and Hansen (1981) did exactly what you asked. Despite 3-4 decades of flat temperatures, they predicted an abrupt change to warming. Obviously they were correct.

Your initial comment about "extrapolation" was bogus, like most of what you seem to post here.

Layzej said...

If you are dismissive of the greenhouse effect then you end up making very wrong projections like these:

Lindzen's 2004 bet "global average temperatures in 20 years will in fact be lower than they are now." - http://www.jamstec.go.jp/frsgc/research/d5/jdannan/betting.html

That hasn't worked out so well: http://woodfortrees.org/plot/gistemp/from:2004/trend/plot/gistemp/from:1970/mean:12


Bastardi in 2011: world will cool 0.1 to 0.2 by 2021 -http://www.nationalreview.com/article/257040/bastardis-wager-matthew-shaffer

What happened instead will surprise you: http://woodfortrees.org/plot/gistemp/from:2011/trend/plot/gistemp/from:1970/mean:12

Then there's the Mills forecast from 2016 that's already failed: http://julesandjames.blogspot.ca/2016/02/no-terence-mills-does-not-believe-his.html

Easterbrooks cooling projections look pretty foolish now: https://skepticalscience.com/lessons-from-past-climate-predictions-don-easterbrook.html

McLean is right out to lunch: https://skepticalscience.com/9-months-after-mclean.html

Etc.

David Appell said...

Nice list, Layzej. Thanks.

David Appell said...

David in Cal said...
"But, simple extrapolation of the past data would also have done a fairly good job."

Not if you don't think GHGs are the cause of the warming.

"The models will be more validated when they correctly predict something unexpected."

To climate deniers, isn't warming "unexpected?" Haven't lots of them been predicting cooling any time now....?

David Appell said...

David in Cal said...
"What I have in mind would be something analogous to what happened in 1919. What an analogue would be for climate change I can't claim to know."

1) the very fact of global warming of 1 C since the pre-industrial era (and even since 1970). Warming has happened as expected.
2) the cooling stratosphere, which is a signature of greenhouse warming (and not solar warming). But you have to account for ozone depletion first, which also causes stratospheric cooling.
3) IR wavelengthg-specific observations like Philipona+ GRL 2004: https://is.gd/ePKTwX, Feldman+ Nature 2015: https://is.gd/vIWMxr

This is also well worth watching:

"Successful predictions of climate science," Ray Pierrehumbert, 2012
https://www.youtube.com/watch?v=RICBu_P8JWI