AIs Trained on Climate Models Predict Faster Warming

Essay by Eric Worrall

Why did they train the AI initially using climate models? Why not ignore the models, and directly use the observations to directly train the AIs?

AI research predicts planet will warm faster than expected

By CNN 11:27am Jan 31, 2023

The study estimates that the planet could reach 1.5 degrees Celsius of warming above pre-industrial levels in a decade, and found a “substantial possibility” of global temperature rises crossing the 2 degrees threshold by mid-century, even with significant global efforts to bring down planet-warming pollution.

Data shows average global temperature has already climbed risen around 1.1 degrees to 1.2 degrees since industrialisation.

“Our results provide further evidence for high-impact climate change, over the next three decades,” noted the report, published on Monday in the journal the Proceedings of the National Academy of Sciences.

Read more: https://www.9news.com.au/world/climate-change-artificial-intelligence-dire-forecast-for-planet-future/764893d8-2a02-4214-8980-d1e4a4b5ca87

The abstract of the study;

Data-driven predictions of the time remaining until critical global warming thresholds are reached

Noah S. Diffenbaugh and Elizabeth A. Barnes

Edited by Michael Mann, The Pennsylvania State University, University Park, PA; received April 25, 2022; accepted December 14, 2022

January 30, 2023

120 (6) e2207183120

Significance

The United Nations Paris Agreement aims to hold global warming well below 2 °C and pursue 1.5 °C. Given the clear evidence for accelerating climate impacts, the time remaining until these global thresholds are reached is a topic of considerable interest. We use machine learning methods to make truly out-of-sample predictions of that timing, based on the spatial pattern of historical temperature observations. Our results confirm that global warming is already on the verge of crossing the 1.5 °C threshold, even if the climate forcing pathway is substantially reduced in the near-term. Our predictions also suggest that even with substantial greenhouse gas mitigation, there is still a possibility of failing to hold global warming below the 2 °C threshold.

Abstract

Leveraging artificial neural networks (ANNs) trained on climate model output, we use the spatial pattern of historical temperature observations to predict the time until critical global warming thresholds are reached. Although no observations are used during the training, validation, or testing, the ANNs accurately predict the timing of historical global warming from maps of historical annual temperature. The central estimate for the 1.5 °C global warming threshold is between 2033 and 2035, including a ±1σ range of 2028 to 2039 in the Intermediate (SSP2-4.5) climate forcing scenario, consistent with previous assessments. However, our data-driven approach also suggests a substantial probability of exceeding the 2 °C threshold even in the Low (SSP1-2.6) climate forcing scenario. While there are limitations to our approach, our results suggest a higher likelihood of reaching 2 °C in the Low scenario than indicated in some previous assessments—though the possibility that 2 °C could be avoided is not ruled out. Explainable AI methods reveal that the ANNs focus on particular geographic regions to predict the time until the global threshold is reached. Our framework provides a unique, data-driven approach for quantifying the signal of climate change in historical observations and for constraining the uncertainty in climate model projections. Given the substantial existing evidence of accelerating risks to natural and human systems at 1.5 °C and 2 °C, our results provide further evidence for high-impact climate change over the next three decades.

Read more: https://www.pnas.org/doi/full/10.1073/pnas.2207183120

To their credit they have published their code on github. “… Code is available on GitHub at https://github.com/eabarnes1010/target_temp_detection (60) and is archived on Zenodo at the following DOI: https://doi.org/10.5281/zenodo.7510551 (61).”.

My understanding is the researchers are attempting to use the AI to identify climatologically significant geographic regions, or distributions of observations, to try to filter out the noise and reduce the uncertainty of predictions.

My concern with this approach is if the data was sufficient for tuning predictions, the AIs could be trained directly on the data, the AIs could infer climate models directly from the data.

Using a simulation or model allows a large number of training runs to be packed into a short period of time, and constrains the output of the AI. But the AI is then tainted by the model, it effectively becomes an extension of the model.

I guess time will tell whether their approach has yielded increased predictive skill.

Get notified when a new post is published.
Subscribe today!
5 24 votes
Article Rating
97 Comments
Inline Feedbacks
View all comments
n.n
February 1, 2023 8:43 am

Non-linear smoothing functions.

February 1, 2023 8:57 am

Only people who don’t understand what an average is trying to represent would ever even think of this excursion into la-la land. And then they have the cheek to try to tell us their neural network is “explainable”.

This work is a thumping triumph of tools over understanding.

There is some powerful Kool-Aid circulating out there.

Reply to  quelgeek
February 3, 2023 8:07 am

The state of “AI” at this stage is definitely “Automated Idiocy,” not “Artificial Intelligence.”

David Albert
February 1, 2023 9:18 am

“Data-driven predictions of the time remaining until critical global warming thresholds are reached”
Models do not produce data.

ferdberple
February 1, 2023 9:25 am

https://www.youtube.com/watch?v=7LVSrTZDopM

Lindzen / Peterson interview.

The issue is not if CO2 causes warming. Arguing that point lends credibility to warmist arguments.

The only question is how much warming is natural. No one knows. Until then no one knows if modifying CO2 emissions will be good or bad.

From this, WUWT may actually be unknowingly promoting “green” aguments.

Talking about all the different errors keeps the subject alive. Concentrating only on a single overwhelming error kills interest in the topic allowing the mania to die out.

R.Morton
February 1, 2023 9:38 am

Why did they train the AI initially using climate models? Why not ignore the models, and directly use the observations to directly train the AIs?

Ummm…. because that would disprove the AGW hypothesis yet again, perhaps????

strativarius
February 1, 2023 9:39 am

Artificial intelligence?

Glorified pattern matching

Reply to  strativarius
February 3, 2023 8:09 am

I like that one! Not as direct as Automated Idiocy, but more descriptive.

February 1, 2023 10:00 am

From the article: “Data shows average global temperature has already climbed risen around 1.1 degrees to 1.2 degrees since industrialisation.”

The 1.1C or 1.2C temperature mentioned (one estimate by NOAA and one by NASA) pertains to the year 2016, the year that tied for the warmest year with 1998, in the satellite era (1979 to present).

Since 2016, the temperatures have cooled by about 0.75C, according to the latest UAH chart.

Somebody better tell the AI that although CO2 is increasing, the temperatures are cooling. See what the AI makes of that.

The UAH satellite chart:

comment image

Reply to  Tom Abbott
February 3, 2023 3:39 am

All the climate change alarmists are still living in the year 2016.

February 1, 2023 10:14 am

bring down planet-warming pollution.

Or “plant food” as it’s known to anyone with half a brain cell

February 1, 2023 10:19 am

My concern is companies are developing AI to form the basis of search engines.

The output of these AI-based searches will be a page of prose not a series of search results, so it will be “search by consensus.”

What could possibly go wrong?

Reply to  Redge
February 1, 2023 11:13 am

Surely in just the same way as we can exclude certain elements of a search it should become possible to do so with AI.

KevinM
Reply to  It doesnot add up
February 1, 2023 1:03 pm

Who picks?

February 1, 2023 12:01 pm

Artificial intelligence trained on artificial data, what could possibly go wrong?

February 1, 2023 12:50 pm

ai vs nomad…coming to a theater near you.

February 1, 2023 12:51 pm

sorry forgot to attach…

nomad.jpg
largolarry
February 1, 2023 12:52 pm

The original so called Global Warming experts took bad temperature data and used it to do the first models. When that wasn’t good enough they modified the temperature to create “better” models. Now after many iterations they are using the garbage data to generate new models. Garbage in creating more garbage out and that being used again as garbage in.

vinceram
February 1, 2023 1:18 pm

artificial intelligence to project artificial climate outcomes. How lovely.

I wonder if their computations consider the fact that CO2 LOWERS the total emissivity/absorptivity in the atmosphere, thus CO2 has a net COOLING effect.

February 1, 2023 1:28 pm

If they “trained” an AI from Das Kapital by Marx would it tell us that global Communism is coming?

R.Morton
February 1, 2023 1:40 pm

To me, this is the equivalent of trying to ‘train’ an AI program to learn how to care for rabbits by showing it hour upon hour of Bugs Bunny cartoons in lieu of showing it real rabbits in their natural environment.

This is a classic a case of “ask a stupid question, get a stupid answer”.

Bob
February 1, 2023 2:18 pm

This is all you have to know about this study:

“However, our data-driven approach also suggests a substantial probability of exceeding the 2 °C threshold even in the Low (SSP1-2.6) climate forcing scenario. While there are limitations to our approach, our results suggest a higher likelihood of reaching 2 °C in the Low scenario than indicated in some previous assessments—though the possibility that 2 °C could be avoided is not ruled out.”

There it is, these rascals know they are blowing smoke. They know the 1.5C threshold will be crossed no matter what we do to prevent it. We could do everything they demand and still cross their scary thresholds. They are laying the ground work for a dignified(?) backtrack or walk back. I can hear it now – oh no you misunderstood us we didn’t mean what you think, here look at this study.

Bunch of miserable lying creeps.

February 1, 2023 2:35 pm

As I’ve said before.

They have the “Artificial” part down pat.

but are severely lagging on the “Intelligence” part.

It regurgitates like a Norwegian Blue, reporting all the consensus garbage it can find.

Reply to  bnice2000
February 1, 2023 2:53 pm

And with Mickey Mann anywhere near it, any possibility of intelligence becomes infinitely non-existent !

TomInOregonCity
February 1, 2023 2:50 pm

AI? We’ll know we have AI when cars are trained how to drive the same way your mama taught you. This scam chat bot is a parrot for easy stuff coupled with propaganda for controversial stuff: no value discovery by this unthinking shuffled deck of cards. Of course it predicts more warming: it has no way to make value judgments — no nose with which to smell a rat, no ears with which to overhear the conspiracy conversations BS, no eyes to see programmers stuffing pre-determined answers to everything attached to the WOKE narrative.

Are we really better off than with Eliza, almost 60 years ago?

February 1, 2023 8:47 pm

Edited by Michael Mann, The Pennsylvania State University, University Park, PA; received April 25, 2022; accepted December 14, 2022″

Says it all right there.

Reply to  ATheoK
February 2, 2023 8:27 pm

Please, that is Pennsylvania State University Distinguished University Professor Dr Michael E Mann.

Steve Smith
February 2, 2023 1:28 am

“The United Nations Paris Agreement aims to hold global warming well below 2 °C and pursue 1.5 °C. Given the clear evidence for accelerating climate impacts”

Clear evidence???

Reply to  Steve Smith
February 3, 2023 9:45 am

And not JUST clear (read: non-existent but you’ll believe it if we say so) “evidence” of supposed “climate impacts,” , but of ACCELERATING “climate impacts.”

Obviously an application of what Goebbels professed. The bigger the lies, the more likely they will be believed.

February 2, 2023 2:30 am

A clever way to deceive: it adds another “layer” between the modelers and reality. They can also train their AI to even more bias than the models themselves create. And finally, they can pretend it’s AI, so there’s no human bias, since a lot of people don’t know what is behind AI,
and think it’s some sort of “superior” intelligence.

February 2, 2023 7:16 am

I calculated the average ECS of the used CMIP6 models, see SI. This gives for the SSP1-2.6 scenario 3.9 and for the SSP2-4.5 it gives 4.2. In the latest approach for estimating ECS “from multiple lines of evidence” ( Lewis 2022) the best estimate is 2,2K/2*CO2. It makes no wonder that the “paper” ( better ferry tale) comes to a strange outcome.

Reply to  frankclimate
February 3, 2023 11:47 am

ECS is and always has been fiction, no matter the “calculated” value, because it is based on the foundational assumption “all other things held equal.”

Here in REALITY, this has never occurred, is not occurring, and will never occur. The feedbacks are negative and the ACTUAL, as opposed to hypothetical, “effect” of CO2 on temperature is ZERO according to observations.

Observations trump theory.

sciguy54
February 2, 2023 7:31 am

Projections based on current climate models predict that future trends will be similar to trends predicted by current climate models. This is surprising?

February 2, 2023 1:34 pm

The Deep State has convinced many that CO2 causes climate change. This allows them to claim that the theory has been verified by AI. It will be interesting to see how they spin it when the planet stops cooperating. Perhaps it has already. The temperature trend has been down since before the 2016.el Nino while CO2 continues its accelerating increase.

current & paleo T CO2.jpg