Some are trying to defend Judith Curry’s claim that early 20th century global warming (i.e. between 1900 and 1950) was “almost as large as the warming since 1950.” An ardent attempt from Jaime Jessup reveals a mind hard at work, trying to make the evidence fit belief.
Let me explain.
When trying to measure climate change (whether natural or man-made), we have to separate the signal from the noise, that up-and-down jiggling and wiggling that never ceases but never really gets anywhere. The signal we’ll call climate, when we add the noise to it we get the weather.
I looked into the issue by using a smoothing method (modified lowess smooth) to estimate what the signal was, which I’ll show here as a solid black line for NASA data (first on my list, I did all five major global data sets):
I don’t think the early 20th century warming qualifies as “almost as large” when the warming since 1950 is 2.8 times as big. Do you?
I didn’t just use NASA data. I used NOAA, and HadCRU, and Cowtan & Way , and Berkeley Earth, and they all gave the same answer. “Modern warming” was bigger than “early warming” by a factor ranging from a low of 1.8 times the size (for the HadCRUT4 data), to a high of 2.8 times the size (NASA data). Any data set you choose, including HadCRUT4, the early warming is not “almost as large” as the warming since 1950. So, I stated outright that Judith Curry’s claim is bullshit.
Jaime Jessup takes exception with my analysis. He shares a graph with us:
Then he makes sure we know it’s a real graph:
This is not a fake graph. It is actual data via Tim Osborn of the UEA. The blue line is a decadally smoothed average. What it shows is that between 1910 and about 1945, global temperature increased by just over 0.5C. Between 1950 and now, global temperature has increased by approximately 0.75C, looking at the decadal average. 0.75 is 1.5 times 0.5C, …
He seems to think I’m showing fake graphs. But the graphs I showed are real too, featuring real data from NASA and NOAA and Cowtan & Way and Berkeley Earth and yes, from UEA too (that’s the HadCRUT4 data, the same data he’s using).
He used a 10-year smoothing filter (which? one wonders). But we’ve studied global temperature data enough to know that a 10-year filter is too short, it’ll show too much fluctuation that’s really due to noise, not signal. This is method #1: help noise obscure the signal.
No matter. I’ll do my own 10-year filter, a Gaussian filter, and mine sure looks a lot like his:
He estimates the early warming was “just over 0.5C.” My calculation puts it at 0.48. He gets a ratio, late to early warming, of ~1.5. I get 1.54.
And that’s giving the early warming every chance, using a smooth with too much noise and allowing only the data set with the lowest ratio of the five available. The late warming is still 54% bigger, early warming just is not “almost as large.”
Now the most fascinating part. The factor of 1.5 (the absolute lowest he could get it) wasn’t good enough to satisfy Jaime Jessup, so he changes it yet again, but in bizarre fashion.
If we look at the anomalies only, notice that 1910 was a little less than -0.5C, 1945 was about +0.1C, 1950 was about -0.05C and 2018 – the latest year in the series – is +0.6C (January to November). So, very similar warming 1910 to 1945 and from 1950 to 2018. Whichever way you look at it, the warming from 1950 (or 1979 if you prefer, which I do, because it was uninterrupted rapid warming, just like the E20C warming) to present is of similar magnitude and rate to the early 20th century warming. They are not vastly dissimilar as claimed by Tamino.
Jessup resorts to basing his estimates by “eyeballing” values, based on individual points — which of course maximizes the impact of noise. This is method #2: Cherry-pick times and values to enhance your case. It enables him to claim that the range covered by global warming in the early 20th century is given by the dashed blue lines in this graph:
Look at the red line, showing Jessup’s already-too-wiggly smooth. Look at the range it covers prior to 1950. Look at the range Jessup uses to estimate early warming. Do you think he’s being honest? I don’t.
As for the later warming, Jessup is using these values (again, the dashed blue lines):
Look at the red line. Look at the range it covers after 1950. Look at the range Jessup uses to estimate late warming. Do you think he’s being honest? I don’t.
If you want to know whether or not I’d call the early and late warming “vastly dissimilar,” you’ll have to define precisely what you mean by “vastly dissimilar.”
But I can tell you this: the idea that early 20th century warming was “almost as large” as recent warming, in all its forms and with all its dressings, is not just bullshit, it’s dim-witted bullshit!
This blog is made possible by readers like you; join others by donating at My Wee Dragon.
Open Mind 
the denialati work in the realm of obscure language in this instance. To say almost as large is wonderfully vague since it provides no frame of reference. Per Wittgenstein, as I understand his ideas: that which can be stated, should be stated clearly: so, the correct presentation of this data might be: the early period of warming produced 0.31C degree of warming and the late/current period has produced 0.86C degree of warming to date.
These numbers speak for themselves. Rhetorical representation of these numbers is the kind of thing about which we should remain silent per W. I don’t think JC got the Wittgenstein memo.
Let’s see…
If I cherrypick a start from a local min and measure to a local max 35 years later in one case (i.e., 1910-1945) and start from near a local max and measure to another local max in another case (i.e., 1950-2018) I get apples-to-apples comparisons, right?
Who does he think he is fooling? Does anyone really think he is actually fooling himself?
FWIW, your analysis of the relative magnitude of the changes between the two periods (ca. 1910-1945 and ca. 1970-current) seems correct to me. However, I wonder if there is another way of looking at it? Consider the piecewise linear analysis method that you have presented previously. This provides slopes of the linear ranges, and presumably (?) confidence intervals for these slopes. So how do the relative slopes of these two periods compare, and are they significantly different? Visually, from your earlier analyses, they look different, but how different are they? And would this be a meaningful way of looking at it?
Also, I wonder about the relative accuracy of the data over these various periods, and to what extent this might influence what we can conclude about relative rates of change.
The “earlier analysis” of BEST data have trends 1910-40 = 0.14ºC/decade and 1980-to-date= 0.19ºC/decade. Concerning accuracy of the temperature data and implications for these trend numbers, there would be reason to doubt the accuracy of the earlier data and the size/speed of the rise except much of the warming is found in the high Northern latitudes and this would account for the speedy SLR which is also on record for this time.
Of course, the denialist folk trying to subvert the 1910-40 global temperature increase to their own cause don’t try very hard to nail down the attribution of that event. Note Blog-Mom Judy is prompted to embark on this recent round of hand-wringing over early 20th century warming by belatedly spotting Hegerl et al (2017) ‘The early 20th century warming: Anomalies, causes, and consequences.’. And when the finding of that paper are discussed (“Attribution studies estimate that about a half (40–54%; p> .8) of the global warming from 1901 to 1950 was forced by a combination of increasing greenhouse gases and natural forcing, offset to some extent by aerosols.”) she goes into denialist mode – “not THAT much!! … My own back-of-envelope calculations suggest a smaller number.” (Mind, to be correct, the paper does not appear to explain the method of reaching that finding).
And well might Judy disagree! If half the early 20th century warming is attributed to forcing, that doesn’t leave enough for natural variation which can then be boldly subtracted from the post 1950 forced warming by dint of her wonderous big natural wobble.
Reblogged this on jpratt27 and commented:
Yep it’s bullshit alright
What is the best argument for why the post WW2 cooling was so anemic: 1945 to ~1965? If NV caused the GMST to rise from 1910 to 1940, a perfect half for a 60-year cycle, what explains NV’s failure to send it right back to 1910-level anomalies?
JCH,
The failure of internal variation to wobble the climate into a greatly cooler state 1945-65 pre-supposes that the warming 1910-40 was mainly due to the same internal variation wobbling it up. It would be interesting to see the grand workings of Blog-Mom Judy on this matter. Yet, despite this being allegedly an exceedingly important but unresolved matter for (Judy’s) climate science and thus for the governance of the whole wide world, I don’t think it has ever been written up and published, even in blog form. (If I’m wrong on this, it would be good to have the reference/link.)
The view I take of it (which to me makes a nonsense of all Judy says on this matter) runs something like this:-
Consider that post-1975 warming is running at +0.18ºC/decade and the warming 1910-40 was 75% of that rate +0.135ºC.
(a) The man-made climate forcing was increasing at 25% of post-1975-rates suggesting that a third of the 1910-40 forcing can be attributed to AGW.
(b) Prior to the 3 decades 1910-40, there was 6 decades averaging half that rate of man-made forcing but which was masked by volcanic forcing. When the volcanic forcing fades, and say this is 1910-40, this would double the effective forcing 1910-40. We now have 50% of today’s warming rate when we measured 75%. And that is pretty-much what you find in the IPCC AR5 attribution graphs (eg Fig 10-1a).
(c) We are thus left with a rate of warming of +0.035ºC/decade which I’m sure could do with some explanation above-&-beyond invoking measurement-uncertainty.
(d) If we consider this +0.035ºC/decade was part of a wobble and turned into a negative effect over the next three decades 1940-70 (à la Judy), and noting that the rate of man-made forcing was lower through the first 2 decades of that period and perhaps extend it to the end of that period because of negative volcanic forcing (Mount Agung), perhaps we should expect a cooling through this period of (+0.025-0.035=) -0.01ºC/decade. Yet such a small speculative wobble is simply not grand enough for contrarian denialists to trumpet in the manner to which they have become accustomed. It simply won’t do!!
Of course what truly “won’t do!!” is the eye-poppingly unscientific antics of contrarian denialists on this matter. But hey, what do I know?
It has been discussed at Climate Etc. many times. This one example.
JCH,
“Discussed?” Indeed!! There is discussion there, many many words spattered across the interweb!!!
The example you link to is so typical of all we get from blog-mom Judy. This exemplar post (from 2011) addresses Smith et al (2011) ‘Anthropogenic sulfur dioxide emissions: 1850–2005’. So what do we get from darling Judy? A whole lot of graph-eyeballing followed by an unsubstantiated conclusion “(T)he large cooling from 1940-1950 cannot be explained by aerosol forcing.”
This is not the “writing up” this subject deserves but a lot of hand-waving followed by a large dollop of hot air. Yet, as I set out above, Judy sees this as a subject of great importance, not just for climatology but for the whole wide world!!! That she does nothing but handwave on a subject she argues is so important says it all. She is but a denialist.
Dr Kevin Cowtan of York uni has an online temperature trend tool.
The trend from 1910 to 1950 is 0.125c/decade and the trend from 1950 to 2018 is also 0.125c/decade.
This is using IPCC’s preferred HAD Crut 4 Krig global.
http://www.ysbl.york.ac.uk/~cowtan/applets/trend/trend.html
It uses the methods section of Foster and Rahmstorf.
Yes, little if any warming from 1950 to 1974.
Yes, always the same cherrypicking trick.
entire graph (from 1850) is easy to read:
NO warming (actually a little cooling) from available data up to approx. 1910, similar to the pause (a real pause, not the 2000 and over fake one) during 1950-1980;
2 very strong (much more than the peak warming during deglaciation times superficial warming) TS increasing between approx. WWI and WWII and after ’70 ending to nowadays.
Of course if like clever Jeff we compare a linear trend between different intervals, 4 decades vs approx 6 decades, with the last ones starting with 2 FLAT decades, we find similar warming rate.
And what about if we compares apples with apples?
1882-1950 vs 1950-2018 results are not nice for cherrypickers (0.068 C/decade vs 0.125 C/decade).
Avoiding cherrypicking: warming speed for the two periods are roughly similar
with the recent one, starting from approx. 1978 faster in comparison to 1910-1940 so called “early warming”, being approx 0.19 C/decade vs 0.14 c/decade (in any case more than 30% higher). Cumulative superficial warming are not “almost as large” being approx. 0.75 C vs 0.42 C (80% higher). due to the fact that the first warming time was shorter and slower that actual one.
Well the early warming period is a lot for a negligible rise in CO2 from 300 to 311ppm.
On the graph both warming periods (early and current) look significant and comparable to me.
And how do you know that a rise in CO2 from 300 to 311ppm is negligible?
Answer: you don’t.
“look comparable to me”
said the blind man.
Jeff,
The CO2 effect – by itself – is logarithmic. The same absolute change at lower concentrations has more effect than at higher concentrations. Curry mentions this, but there are also feedbacks to factor.
CO2 isn’t the only player. To the best of our knowledge other factors conspired to enhance the early warming (solar/volcanic aerosols), while the same factors contributed little or negatively to the latter period. Curry mentions this, too, but downplays it.
Curry, Tamino and Jaime are comparing total change for the periods, while you are talking about the rate of change. Why shift the goalposts?
@ 0.125 C/decade
1910-1950: 0.50C total
1950-2018: 0.85C total
70% more warming in the latter period.
(setting aside uncertainties)
The main point i.m.h.o. is, that any such ratio is utterly meaningless, because it has no predictive value whatsoever – and it’s the prediction, that counts. They do not, as some seem to insinuate, come from extrapolation of any ratios of warming in the past, but from physics, i.e. from tested models.
So in this case you have shown typical argumentative failures again, but the denialist argument should be rejected right away.