Monday, January 22, 2018

Climate Models Are Doing Great

After 2017's annual temperatures came in, Gavin Schmidt posted this on Twitter:

Here, historical forcings are used prior to 2000 -- the actual GHG concentrations, volcanic eruptions, etc. After 2000 the comparison uses the old IPCC Scenario A1B -- "...very rapid economic growth, global population that peaks in mid-century and declines thereafter, and the rapid introduction of new and more efficient technologies." (Truth is, there isn't much of any difference in these old AR4 scenarios by this time.)

Looks equally good with the CMIP5 models.

The model/observation difference varies depending on the particular year (or couple of years), but over the long-term it's looking pretty good.

Certainly good enough to see that we have a big AGW problem on our hands.

24 comments:

David in Cal said...

Speaking as a non-expert in climate models, it looks to me as if the actual temperature rise does not disprove the models, since it has been in the range predicted by the models. That's good. OTOH the actual temperature rise doesn't fully validate the models IMHO, because:

1. The models have large ranges. E.g., the current IPCC sensitivity range is 1.5 to 4.5 deg C. It's not hard to correctly predict something when the range is so wide.

2. The models' predictions are not a surprise. One could have done almost as well by simply following the trend of actual temperatures. Correctly predicting something surprising is a stronger validation for a model.

Cheers

climate data analysis said...

Well, David in Cal, if you look back in time a little further, the earliest models were published at a time when there had been a couple of decades with no obvious warming. In those cases, the predictions of global warming *did* represent a break with the past. You have already seen the work of Manabe and Wetherald (1967) which was stunningly accurate:

Evaluating the prediction of Manabe & Wetherald 1967

Another example is Wally Broecker's 1976 paper.

David Appell said...

David in Cal said...
"1. The models have large ranges. E.g., the current IPCC sensitivity range is 1.5 to 4.5 deg C. It's not hard to correctly predict something when the range is so wide."

Do you want scientists to accurately report uncertainties in both observations and models, or not?

Or would you prefer they lie about the inherent uncertainties in their science?

David Appell said...

David in Cal said...
"2. The models' predictions are not a surprise. One could have done almost as well by simply following the trend of actual temperatures. Correctly predicting something surprising is a stronger validation for a model."

a) Not a surprise?

Absent human influence, what natural factors account for modern warming?

b) Do you really not understanding why climate models can never "predict" future climate. I thought that anyone reading this blog understood this, but correct me if I'm wrong.

David in Cal said...

David --

a) As far as I know, science doesn't fully understand all the natural factors that affect climate patterns. But, so what? If I needed to predict the future of some variable, I would start by assuming that past patterns will continue into the future. That seems natural to me, since it was the basis of casualty actuarial work.

b) I confess I do not understand why climate models can never "predict" future climate. On the contrary, I thought that the reason we know that global warming is a problem is because our climate models predict a warmer climate in the future.

Cheers

David Appell said...

David: what natural factors does science not understand?

David Appell said...

David, climate models can never predict future climate because no one knows the future of our greenhouse gas in missions, aerosol emissions, and a land-use changes. All of these are put into models to calculate future climate. Scientist use assumptions, the so-called representative concentration pathways – Precisely because no one, you or I or anyone, can’t say how much the other two will be in mid it in the year 2026, or 2052, or 2087. Or La ninos or el Nino’s or the pacific decidable oscillations or changes in the suns intensity or major volcanic irruption’s, and such. All of these determine future climate change. Some average 20 over a long enough time, like El Nino’s and la Nina’s And the sun’s intensity and major volcanic irruption’s.

We know none of these things that light in the future. They all help determine climate. So we cannot project future climate.

climate data analysis said...

David, climate models can never predict future climate because no one knows the future of our greenhouse gas in missions, aerosol emissions, and a land-use changes

This is an argument for evaluating models based on their expected temperature as a function of CO2 concentration (or of net forcings) rather than of *time*.

We don't know exactly when CO2 will reach 500 ppmv (probably some time around 2050). But we can say with a fair degree of confidence what the global temperature will be when that happens ... probably approx. 0.8 degrees C warmer than now.

David in Cal said...

David -- here are some natural factors that affect climate patterns, which science not understand or does not fully understand

-- What causes ice ages to end
-- What factors cause warming after ice ages end
-- How much each of these factors contributes to this warming
-- How these factors interact
-- The magnitudes and interaction of things that affect CO2's impact on warming. I.e., the things that cause the uncertainty in climate sensitivity.
-- Feedback of global warming with moisture in the atmosphere
-- How warming of the upper atmosphere interacts with warming of the surface. (I believe scientists don't fully understand this, because, as I understand it, the actual relationship of surface warming and atmospheric warming does not match the models.)
-- Impact of global warming on other climate aspects, such as rainfall, windstorms, extreme events etc.
-- Regional impact of global warming. E.g., Why has the Arctic been warming faster than the rest of the planet? Was that pattern predictable in advance? Will that pattern continue? Will it reverse, and, if so, when will it reverse?

Cheers

Layzej said...

DiC,

I suppose it's possible to argue that we don't fully understand anything, but these things that you suggest we do not know are all things you can find on Google.

For example, you wonder whether polar amplification could have been predicted in advance. Well yes, it seems it was.

This paper by Wallace Broecker from 1975 was written at a time that had seen over 3 decades of cooling, yet predicted that exponential growth in CO2 would lead to a pronounced warming within a decade or so (nailed it!). It also predicted that by early in the next century CO2 rise will have driven the mean planetary temperature beyond the limits experienced during the last 1000 years (bingo!)


Well, this 1975 paper also references prior work by Manabe and Wetherald on polar amplification. So yeah. Could have been predicted. Was predicted.

This all ties back to your original comment: "If I needed to predict the future of some variable, I would start by assuming that past patterns will continue into the future. That seems natural to me, since it was the basis of casualty actuarial work."

Well yeah, but physics. After over 30 years of continued cooling, this Broecker fellow suggested that CO2 would lead to a pronounced warming. Now that pronounced warming is the new normal, you say that continued warming is not surprising. A steeper rise in the future won't impress because, after all, hasn't warming been accelerating since the 1900's?

David Appell said...

David in Cal, Re: 4:16 pm comment

Everyone knows all that you wrote.

So what's your point?

David Appell said...

David in Cal wrote:
"Feedback of global warming with moisture in the atmosphere"

This is the water vapor feedback.

It's definitely real, and a strongly positive feedback (water vapor is a strong GHG).

David Appell said...

David in Cal wrote:
"How warming of the upper atmosphere interacts with warming of the surface."

The stratosphere is cooling. This is one of the major predictions of greenhouse theory, and it's observed.

David Appell said...

David in Cal wrote:
"What causes ice ages to end"

Milankovitch factors -- especially warming at about 65 deg N -- accompanied by CO2 feedback.

Also very well known.

David Appell said...

David in Cal wrote:
"1. The models have large ranges. E.g., the current IPCC sensitivity range is 1.5 to 4.5 deg C."

Everyone knows this. What do you propose should be done about it, that scientists haven't currently done?

Uncertainties cannot be better than the uncertainties in the underlying data.

"2. The models' predictions are not a surprise. One could have done almost as well by simply following the trend of actual temperatures. Correctly predicting something surprising is a stronger validation for a model."

a) models can't predict. Ever.
b) why would you predict the temperature trend should continue, without a cause?

David Appell said...

Layzej: nice comment.

David Appell said...

David in Cal wrote:
"If I needed to predict the future of some variable, I would start by assuming that past patterns will continue into the future. That seems natural to me, since it was the basis of casualty actuarial work."

Every day from about 6 am to 3 pm, the outside temperature increases.

Would you therefore conclude that it will keep increasing throughout the late afternoon, evening and night?

climate data analysis said...

Layzej's comment has motivated me to post some graphs showing how accurate Wally Broecker's 1975 forecast was (I accidentally referred to this as Broecker 1976 up-thread).

Evaluating the prediction of Broecker (1975)

Check out Fig 3 and Fig 4. Despite writing at a time when there had been relatively little warming for several decades, Wally Broecker did an absolutely smashing job of forecasting rapid warming post-1975.

Broecker's climate sensitivity: 2.38. Actual per GISSTEMP: 2.41.

We have known for over 40 years what to expect, if CO2 emissions continued to increase. Uncertainty about the details doesn't negate the fact that the fundamentals have been understood by scientists for decades.

David in Cal said...

David -- You wrote, The stratosphere is cooling. This is one of the major predictions of greenhouse theory, and it's observed.

OTOH there's this comment, Overall, the study suggests that while tropospheric warming has not accelerated to the extent that models have predicted in recent years, there’s little evidence that it has slowed down. https://www.carbonbrief.org/study-why-troposphere-warming-differs-between-models-and-satellite-data

I do not know how to reconcile these two comments.

climate data analysis said...

Good grief.

Layzej said...

CDA: Nice article!

DiC: Stratosphere != troposphere?

SkS has a good write up: https://skepticalscience.com/Stratospheric_Cooling.html

climate data analysis said...

David in Cal wrote: "The models' predictions are not a surprise. One could have done almost as well by simply following the trend of actual temperatures. Correctly predicting something surprising is a stronger validation for a model."

OK, here's a comparison of David in Cal's prediction vs Wally Broecker's prediction:

https://imgur.com/rmmpv0s

The "David in Cal" line is an extrapolation of the 30-year timescale LOESS trend from 1975 onward. In other words, it is "simply following the trend of actual temperatures".

The Wally Broecker line is from his 1975 paper, informed by his understanding of the role of CO2 in setting the planetary temperature.

David Appell said...

data: nice graph!

David Appell said...

David in Cal wrote:
"I do not know how to reconcile these two comments."

I don't understand your point. The statements refer to different parts of the atmosphere.