Here we update the comparison of climate models and observations through October 2023
The way scientists have unravelled the complexity of the climate system over the ages never ceases to amaze me. It shows what humans are capable of doing.
The next project should be a political system that does away with politicians.
I clicked on your hot model paper and saw that Kate Marvel was a coauthor. The name sounded familiar and then I put two and two together and recognized the name from some recent interviews by David Wallace Wells at NYT, e.g.
https://www.nytimes.com/2023/09/05/opinion/ezra-klein-podcast-kate-marvel.html . Very cool.
I guess my take away is that there is still confidence in the models and we shouldn't totally be freaking out.
Would it be correct to say that the models are broadly correct on a global average basis? Most of the outliers we have seen recently are more local (like the North Atlantic) and those outliers really get averaged out when looking at the entire globe.
Zeke, as you point out, for the CMIP6 models they use historical data up to 2015.
As a result, your post-2008 graphs are a joke, half historical, half future. You are NOT looking at how well the models do at predicting the future.
And if you look at that, at how they predict the post-2014 trends, the models are doing HORRIBLY, running almost twice as warm as the real world. So you've clearly cherry-picked your starting point.
Next, I do love the part where you say:
"CMIP6 contains a subset of models that are running quite hot (>5C climate sensitivity per doubling CO2) that generally do a poor job of representing historical global surface temperatures.
For this reason, the most recent IPCC report created Assessed Warming Projections that weight the CMIP6 ensemble by its performance in matching historical temperatures, and tend to show future warming more in-line with the prior generation of climate models (CMIP5)."
Gosh, Zeke, so you're telling us that if you throw away all the models that do a very poor job of modeling, the rest of the models do a wonderful job of modeling?
Really? That passes for a scientific argument on your planet?
Finally, you've put up the historical hindcasts of the models as if they meant something … but you're totally ignoring a curious fact about these '100% physics based' models, a fact pointed out by Dr. Jeffrey Kiehl in 2007 and resolutely ignored by climate scientists since then.
These model have radically different values for climate sensitivity, varying by a factor of four, from about 1.5 °C/2xCO2 to 6°C/2xCO2 … and despite that, they can all do a pretty dang good job of hindcasting the past.
Simply put, that would not be possible if these models were actually "physics-based" as the modaholics claim … but they're not. They are a collection of kludges and tuned parameters that are carefully adjusted, rewritten, and put under pressure until they can hindcast the past.
But as the post 2015 data above shows, they are really bad at predicting the future.
That should come a no surprise. In the US, stock brokerages are required to put a disclaimer on their ads saying something like "Past performance is no guarantee of future success" … climate models desperately need the same warming.
In short, this might make a great high-school paper about models. But it ignores the real failings of the models, and just spreads peanut butter over the glaring gaps in the logic, the use of egregious cherry picking, the discarding of models that don't fit the party line, and the clear lack of any actual physics basis for the models.
Would not recommend.
My best to all, including Zeke. He's a genius meteorologist, but unfortunately, also a craven apologist for the Tinkertoy™ models.
PS—my post on Dr. Kiehl's findings is:
And my post on how the model temperature outputs can be accurately emulated by a very simple lagged, resized version of the forcings is:
I am left curious why CMIP6 performs significantly worse CMIP5. We know sensitivity is not 5C, so why are hot-running, poor postdicting models even included in the ensemble? I would think it better to focus on the top 5 or so most accurate models and vary initiation parameters to fill out the stocastic grey envelope. Over time modeling should get better not worse!
Nitpick re “we would expect roughly one in 20 months to be above the 95th percentile range if models are accurately representing real-world variability.” Shouldn’t that read “… outside the 95th percentile range …”?
Would you please explain "w.r.t. a" indicated at vertical axis in the first 3 figures: warming trend... ?
Hey Zeke, you stated a while back that you did not support the conclusions in Hansen's paper, Global Warming In The Pipeline, but you then went on the New York Times and stated that warming was celebrating here: https://nytimes.com/2023/10/13/opinion/climate-change-excessive-heat-2023.html
Could you clarify your stance? Thanks
Also, what are your thoughts on the documentary by ABC news “Earth 2100”? It depicts a rather scary future where we do indeed collapse by then, but with believable reasons. Take a look, it’s both worrying and interesting!
I Google your name occasionally to catch up with your Twitter/X posts, (I refuse to subscribe to X.) Nice summary on global emissions today at Carbon Brief, https://www.carbonbrief.org/analysis-growth-of-chinese-fossil-co2-emissions-drives-new-global-record-in-2023/ . We will have to wait to see if emissions peak next year.
There is so much misinformation out there and even good people like Andy Revkin get taken in by people like Roger Pielke Jr., https://substack.com/@revkin/note/c-44555537 . My biggest problem with the 'Low Hanging Fruit' post by Roger is that Roger claims that if we replace 5% of the worst coal plants would with nuclear we will decrease emissions by 9 Gt per year. That is so wrong on so many levels especially since global coal power generation emissions are about 10.5 Gt / year. His analysis is predicated on the erroneous paper by Grant et al. I estimate that at best getting rid of the worst 5% would maybe save 8% of the emissions caused by coal generation, .84 Gt / year - Ref. https://www.nature.com/articles/s41893-019-0221-6 .
I refuse to subscribe to THB just to call him on it.
Any chance you or Andrew care to take that on?