From Dr. Roy Spencer’s Global Warming Blog
Roy W. Spencer, Ph. D.
Since the blogosphere continues to amplify Gavin Schmidt’s claim that the way John Christy and I plot temperature time series data is some form of “trickery”, I have come up with a way to demonstrate its superiority. Following a suggestion by Heritage Foundation chief statistician Kevin Dayaratna, I will do this using only climate model data, and not comparing the models to observations. That way, no one can claim I am displaying the data in such a way to make the models “look bad”.
The goal here is to plot multiple temperature time series on a single graph in such a way the their different rates of long-term warming (usually measured by linear warming trends) are best reflected by their placement on the graph, without hiding those differences.
A. Raw Temperatures
Let’s start with 32 CMIP6 climate model projections of global annual average surface air temperature for the period 1979 through 2100 (Plot A) and for which we have equilibrium climate sensitivity (ECS) estimates (I’ve omitted 2 of the 3 Canadian model simulations, which produce the most warming and are virtually the same).
Here, I am using the raw temperatures out of the models (not anomalies). As can be seen in Plot A, there are rather large biases between models which tend to obscure which models warm the most and which warm the least.
B. Temperature Anomalies Relative to the Full Period (1979-2100)
Next, if we plot the departures of each model’s temperature from the full-period (1979-2100) average, we see in Plot B that the discrepancies between models warming rates are divided between the first and second half of the record, with the warmest models by 2100 having the coolest temperature anomalies in 1979, and the coolest models in 2100 having the warmest temperatures in 1979. Clearly, this isn’t much of an improvement, especially if one wants to compare the models early in the record… right?
C. Temperature Anomalies Relative to the First 30 Years
The first level of real improvement we get is by plotting the temperatures relative to the average of the first part of the record, in this case I will use 1979-2008 (Plot C). This appears to be the method favored by Gavin Schmidt, and just looking at the graph might lead one to believe this is sufficient. (As we shall see, though, there is a way to quantify how well these plots convey information about the various models’ rates of warming.)
D. Temperature Departures from 1979
For purposes of demonstration (and since someone will ask anyway), let’s look at the graph when the model data are plotted as departures from the 1st year, 1979 (Plot D). This also looks pretty good, but if you think about it the trouble one could run into is that in one model there might be a warm El Nino going on in 1979, while in another model a cool La Nina might be occurring. Using just the first year (1979) as a “baseline” will then produce small model-dependent biases in all post-1979 years seen in Plot D. Nevertheless, Plots C and D “look” pretty good, right? Well, as I will soon show, there is a way to “score” them.
E. Temperature Departures from Linear Trends (relative to the trend Y-intercepts in 1979)
Finally, I show the method John Christy and I have been using for quite a few years now, which is to align the time series such that their linear trends all intersect in the first year, here 1979 (Plot E). I’ve previously discussed why this ‘seems’ the most logical method, but clearly not everyone is convinced.
Admittedly, Plots C, D, and E all look quite similar… so how to know which (if any) is best?
How the Models’ Temperature Metrics Compare to their Equilibrium Climate Sensitivities
What we want is a method of graphing where the model differences in long-term warming rates show up as early as possible in the record. For example, imagine you are looking at a specific year, say 1990… we want a way to display the model temperature differences in that year that have some relationship to the models’ long-term rates of warming.
Of course, each model already has a metric of how much warming it produces, through their diagnosed equilibrium (or effective) climate sensitivities, ECS. So, all we have to do is, in each separate year, correlate the model temperature metrics in Plots A, B, C, D, and E with the models’ ECS values (see plot, below).
When we do this ‘scoring’ we find that our method of plotting the data clearly has the highest correlations between temperature and ECS early in the record.
I hope this is sufficient evidence of the superiority of our way of plotting different time series when the intent is to reveal differences in long-term trends, rather than hide those differences.
Well done. Your previous discussion was convincing. This is compelling.
I’m sorry but this is unbelievable.
You/we have been debating the ‘best way to baseline a time series’ for 2 weeks now when this has 0 to do with the point of the statement related to the graph that began this crap show.
I’m sorry Dr.Spencer, you seem like a genuinely honest man who has been under serious attack by your (less than honest) peers and I understand your desire to ‘set the record straight’ so to speak. (even though I disagree with you entirely on CO2 having a MATERIAL impact on the climate)
But you have been captured, institutionalized really, in to playing their game. It’s total & utter misdirection.
The point you should be pushing, the point that matters is that the models show a 43% (if my memory of your article is correct) greater warming vs the data over the compared time period and are thus simply not fit for purpose. That has NOTHING to do with how their y-intercept is adjusted. The slope of a line doesn’t change based on where you start it, it is an independent variable that stands on it’s own.
Instead of posting reams of virtual ink like this I would strongly advise you stick to the real point & call out their misdirection. Here’s a statement that you can adjust as you see fit for anyone trying to claim your plotting of the data is wrong…
“The y-intercept of the graph has nothing to do with its slope. They are independent variables. Your continued attempts at this misdirection belies just how desperate you are, and exactly who is trying to fool society. The climate models show a 43% greater warming (based on their SLOPE) over the time period in question vs the actual, real and measured data. They are not fit for purpose and need to be discarded entirely, at least for the purposes of setting any kind of policy, energy, climate or otherwise.”
Again adjust as you see fit. If it’s directed to Gavin I might throw in a “A scientist of your eminence undoubtedly knows this.”
Do NOT back down, do NOT give in. Do NOT play their game. Keep your eye on the ball. But DO call out their bullshit & misdirection, show the world who really are the liars, cheats and scoundrels.
I understand your point, and thank you for expressing it. But to someone who doesn’t understand the nuances of the arguments, they might think, “Spencer made a good point on that graph, but Schmidt made a good point on the other graph.” I’m not willing to concede defeat on either. But you are correct, it’s the trends that matter, and this issue to some extent distracts from that point.
Your post was specific to Gavin’s criticism of your graphing method. It wasn’t about the model reality divergence that UAH has shown many times. Carry on.
My apologies if I’m sounding mean but I think that when someone is institutionalized it takes a ‘slap in the face’ (virtual or otherwise) to wake them up (I happen to be dealing with a type of ‘institutionalization’ of my own so I have some experience here).
But what ‘nuance’ is there in the statement that ‘the models show a 43% greater warming then reality and are not fit for policy making’? That has 0 nuance. Comparing the slopes isn’t even something you can graph, its a single number. Showing this ‘divergence’ visually has no value.
Furthermore anyone who doesn’t understand the difference between the slope & y-intercept is either NEVER going to ‘get it’ (unfortunately they simply don’t have the mental ability) or are being DELIBERATELY obtuse (Schmidt, AlanJ below, Nick Stokes who somehow thinks the labeling on your graph matters and probably anyone who has bought in to the ‘extreme climate’ propaganda). They will NEVER concede your point on this no matter how well you make it. I’m sorry to tell you but you’ve ‘lost’ this debate because there is NO debate being had. There’s you making good points and the detractors simply putting their fingers in their ears shouting “nyah, nyah I can’t hear you!!!”.
And I’m sorry to be the one to tell you but anyone who MIGHT have been on the fence & thinking that ‘Schmidt made a good point about Spencer misrepresenting the data’ is now very likely to be convinced that’s true because you’ve given up the playing field by playing along & making this about the y-intercept instead of going straight for Schmidt’s throat and pointing out that the ‘centering’ (y-intercept) has 0 to do with the slope (how fast the planet may be warming) and that Schmidt knows this and is being deliberately obtuse in order to fear monger.
In your blog posts or papers you might publish, graph this how you choose and Schmidt can graph his how he chooses, but provided you both properly compare the slopes then there is no debate or ‘nuance’ to the argument that the models run hot & are not fit for purpose.
As much as I’m not entirely on board with some of what you believe in respect to CO2 and the climate you have a FREAKIN’ PHD! You’re time is FAR better spent & far more valuable analyzing the main points not ‘y-intercept nuances’. That’s first year University or high-school stuff.
Way back in high school (early ’60s) normalizing data sets was taught in both science and math classes as the best way to compare different data. It was also pointed out at various places in books and articles on my main interest of the time, cosmology.
The models are paid-for, subjective fabrications meant for scare mongering by allied entities of the IPCC
We are scientists, not alchemists
We must ignore those models
We must use empirical data, correctly graph it, present it as objective reality
I agree with you regarding models being misleading, but I think you have missed the point, which was simply to show that the Spencer/Christie technique is better at representing data. To this end, they used the models, since the data is just what it is, and there can be no dispute regarding measurement.
Reducing climate change to a single meaningless number is playing a silly game.
Climate is complex. Measuring surface or satellite temperature is complex and error prone. Placing faith in a single anomalous number relative to some average temperature value is as unscientific as the belief that a small change in the amount of a trace gas in Earth’s atmosphere can directly alter the radiative energy balance.
“Placing faith… is as unscientific as… belief…”
A thought experiment :
If the ‘consensus’ belief was that the climate is currently cooling rather than warming, could CO2 equally be blamed for affecting the radiative balance negatively by means of stratosphereic cooling, requiring us to decarbonize to save the planet from an ice age?
I have only looked closely at the GISS E2.1 model. It has glaciation covered with the breakdown of the Gulf Stream and the Kuroshio Current by 2100. This results in most of Europe and North-eastern Russia cooling. Not so much in Canada.
So no matter the cause of glaciation, the bases are covered as far as CO2 is concerned.
Glaciation is an inevitable consequence of the precession cycle. As the oceans in the NH warm, there is a lot more moisture in the atmosphere in late September that ends up over land in October and November, depositing as snow on any land cooler than 0C. There is a strong upward trend in early season snowfall and a less pronounced upward trend in snowfall maximum extent. in the NH.
Rickwill, my question is, could a scientifically illiterate population, with faith in the “experts”, believe that a cooling planet is caused by CO2, just as they now believe in a warming planet.
People who ‘defer’ their opinions to ‘experts’ would happily believe that a magic pixie-gas could cause devastating cooling if the experts decreed so. To the extent they care to understand global warming, they think in terms of simple analogies: CO2 is like a blanket, so more blankets must necessarily make the globe hotter, right? But CO2 is also comparable in its physical properties to a conductor of heat such as the metal used in cold iron railings, so instead of a ‘greenhouse gas,’ they could be panicked over a ‘gatepost gas’ sucking all the precious heat away and radiating it from the top of the atmosphere!
If “consensus science” ruled and any contrary ideas were dismissed out of hand, where would we be now?
Any comments from Dr. Schmidt?
I doubt Gavin would weigh in. He has fled from several previous invites to public debate.
He just takes pot shots from the safety of his RealClimate blog, where he can just ban anyone trying to debate him.
Good for Dr. Spencer to go after him anyway. Gavin can run, but not hide.
Roy, the amount of energy defending this single point is excessive. No one is arguing that the method produces an incorrect result, just that the effect is to visually maximize how spread out the individual lines are from each other. This doesn’t matter in the slightest if you plot the observations alongside the full ensemble, the issue is when you plot only the multi-model mean alongside the ensemble, then that spread, physically quite meaningless, seems to imply something that isn’t true (i.e. the models are diverging wildly from observations).
As I said in the previous thread you could make the exact same arguments you’re making here but in reverse – argue in favor of placing the x-intercept of all the trends at the most recent year. It would give you the same visual effect you’re describing, so why not do that?
Because it would NOT give the same visual effect!
The hottest running models would give the coolest temps at the start and show a much smaller divergence to the cooler running models at some future date, e.g. 2080.
Your method cuts the visual error in half. (This would also be true if the models ran cold!)
It is totally idiotic, anti-scientific, and anti-mathematical to think the trends FINISH at the same point.
Your mathematical nil-logic is shining through.
If you want to compare trends, you work from a common starting point.
Dear bnice, please graph a function of the form f(x)=mx+b and point out to us the start and the finish. Please also remember that the edge of your paper is not the domain, which is the set of all real numbers.
YAWN, you really are just being STUPID..
Climate model trends start when they were made.
To compare to UAH, you start at the beginning of UAh with them all the same value..
What you are saying is that the climate clowns deliberately started their models way below actual temperature in 1979.
Even they are not that incompetent.
But you most definitely are.
I agree he’s wasting far too much time debating you knuckle draggers about how to select the y-intercept when it’s the slope that matters but hey it’s his time.
In the meantime perhaps you can enlighten those of us not in the climate cabal how comparing a group of model outputs from different models against reality and because at least 1 of them by chance may match reality you can claim ‘see the models are correct’ (as a group). That’s not science that’s astrology.
Using the model mean would also be incorrect if it was used in an attempt to claim ‘see the models are correct’. Dr. Spencer is using it to show just how badly as a group they do not match reality. Seriously the model mean is running 43% hot. That means 50% of them are running even HOTTER. They are crap from the get go.
And it is NOT the spread on a graph that shows the models are ‘wildly diverging’ its the comparison of the slopes that demonstrates 50% of the models are ‘wildly diverging’ from reality (the rest aren’t much better..”hey I’m within 43% of reality whoopeee!!!”).
That Dr. Spencer chooses a graphical display that unequivocally demonstrates that visually is just icing on the cake.
If you want to compare graphs for predictive value, you give them all a common starting point and observe how they diverge from each other and reality. If, however, you are looking for how accurate the graphs’ starting points were, you give them a common end point and observe how far off they are from the true starting point. Spencer is trying to demonstrate the former, so he is correct to give them all a common starting point. To do otherwise demonstrates statistical torture and mathematic incompetence.
And yes, it’s the slopes of the graphs that matter. But to visually demonstrate the predictive accuracy of the model graphs, you need to have a common starting point, not end point.
Except these are graphs of anomalies, so the differences in values aren’t indicators of predictive skill. As noted, you are making precisely the error everyone else makes looking at Roy’s graph. You think “the dark blue line is much warmer than the light blue line.” But this is not information the graph contains. It’s just constructed in a way that makes you assume that. So it probably isn’t a smart data visualization technique.
The fact that you can all instantly see the concern with plotting the graph the other way round, but think it’s appropriate to plot it the way Roy does, just drives the point home even further. You’re deluding yourselves because you want to incorrectly interpret the data. For you all and for Roy, the misdirection is a feature, not a bug.
And if you cross them in the middle,..
…all it shows is just how pathetic and incompetent your hindcasting to known data was. !
The models are known to be wildly inaccurate at hindcasting. Somehow, this does not discredit them for forecasting. I suppose hope springs eternal, and all that.
“seems to imply something that isn’t true (i.e. the models are diverging wildly from observations).”
That’s not what Roy’s figures imply. Roy’s figures imply that the theory deployed in the models is poorly constrained, i.e., the physics is deficient. A deficient theory produces a large uncertainty in the modeled extrapolations of future air temperature, evidenced by the diverging projection spread.
AlanJ “seems to imply something that isn’t true (i.e. the models are diverging wildly from observations).”
So you’re saying the observations and models match?
That should be obvious when trended together IF true.
Gavin tried to disguise the fact that the models don’t match observations. “Wildly”.
Here you go:
All plotted together, not just the model mean, as Roy does. The observations are well within the model ensemble envelope.
Again, I don’t think it really matters if you use Roy’s weird baseline convention or not – it seems to be very confusing to all of you, but it isn’t confusing to me. So I recommend that he uses a convention that confuses people less, but it doesn’t really matter. The big issue is that Roy doesn’t plot the model spread, he plots only the multi-model mean, so there is no indication of the structural uncertainty in the models.
The Earth is currently in a 2.56 million-year big ice age named the Quaternary Glaciation with over 20 percent of the land surface frozen. This current ice age won’t end until all of the natural ice melts. https://en.wikipedia.org/wiki/Quaternary_glaciation
On some reasonable scale, say -40 to +50 like most outdoor thermometers, the average world temp has been rock steady within a couple of degrees for 10,000 years….and a degree warmer wouldn’t put it outside its “no problem, actually improved for both plants and animals” excursion range…
Hi Dr. Spencer,
This question is a little off topic, on the UAH dataset the file is https://www.nsstc.uah.edu/data/msu/v6.0/tlt/tltpenamg.2023_6.0 the latest daily values available for that calendar year. We pulled all the daily slices from 1978 – 2023 but do not see a file for 2024. Is there an incremental file for current year dailies?
Working in a team building a climate model and we would like to get the daily values as soon as they are available.
Leron Wells
Propagation… eqn. 1 emulations of the CMIP6 projections with SSP370 forcings.
The Figure below is Roy’s Figure D.
The thick red line is a direct projection using the 0.42 coefficient derived from Manabe and Wetherald (1967). The thick blue line is an emulation of the CMIP6 multimodel SSP370 mean. Both were normalized to zero at 1979.