Agree to Disagree
Debate in a non-hostile environment
Alarmist thinking on the recent slowdown is one dimensional
There is a commonly held belief about Albert Einstein. That in response to an attack on his “theory of relativity” by 100 German scientists, Einstein is supposed to have said “Why 100? If I were wrong, one would have been enough.”
The situation with the recent slowdown is similar. If the recent slowdown really did NOT exist, then one scientific paper should be enough to prove that it doesn’t exist. Instead, we are being constantly bombarded with aggressive pseudo-scientific papers, which claim that anybody who believes in the recent slowdown is either a sociopathic criminal, or is getting money from Big Oil.
Just to make it clear, I am not a sociopathic criminal (at least, I don’t think that I am), and I have never received any money from Big Oil (but I am still waiting hopefully). I believe that there was some sort of a recent slowdown in the warming rate, which was strongest from about 2001 to about 2014 (depending on how you define it). I am NOT claiming that this recent slowdown disproves global warming. But I believe that this global warming “aberration” should be accepted by the scientific community, and investigated, rather than being “swept under the carpet”.
Slowdowns have more than 1 dimension. The main 2 dimensions are strength (how much the warming rate slowed by), and length (how long the slowdown lasted).
The strength can be measured by the change in the warming rate. For example, the average warming rate decreased from X degrees Celsius per century, to Y degrees Celsius per century. Or it can be measured as a percentage decrease from some average (or standard) warming rate. For example, the warming rate decreased by Z%, where Z% is a number like 25%, or 50%, or even 100% (a true pause).
The length can be measured using suitable units, like years. Or a start date and an end date can be used. But this raises the issue of how do you know when a slowdown started, or when a slowdown ended. Most slowdowns don’t have abrupt beginnings and endings. Slowdowns normally have a gradual beginning and end, and it can be very difficult to pinpoint exactly when they did or didn’t exist. This is one of the reasons why there is so much dispute about the recent slowdown. One man’s “slowdown”, is another man’s “normal variability”.
In this article, I will investigate the main 2 dimensions of slowdowns, strength and length. Many other studies have investigated both of these attributes. Will my analysis reveal anything new? One of the interesting things that I have discovered, is that the recent slowdown, is the second longest climate “event” since 1970 (when global warming became notable). A climate “event” is defined as a period of warming at a rate which is considerably above the long-term warming rate, or a period of warming at a rate which is considerably below the long-term warming rate.
The warming rate is normally very variable. It doesn’t stay near one value for long. But it did in the recent slowdown. This makes it like the Sherlock Holmes mystery, about “the dog who didn’t bark in the night” (in this case, “the slowdown that didn’t change warming rate rapidly”). The slowdown is a climate event of stability, in a world of highly variable warming rates. The following discussion will show this.
To investigate the recent slowdown, we need to use a temperature series. Everybody has their likes and dislikes when it comes to temperature series, and no matter which one I choose, somebody will attack my choice. For this analysis, I am going to use the GISTEMP Global Land and Ocean temperature index (LOTI). There is a good reason for this choice. If I can show that there is a recent slowdown in the GISTEMP data, then nobody will accuse NASA, Gavin Schmidt, or James Hansen, of adjusting the data to create a slowdown. People may believe that NASA, Gavin Schmidt, or James Hansen, might adjust the data to hide a slowdown, but they wouldn’t adjust the data to create one. Therefore, if I can show that there is a recent slowdown in the GISTEMP data, then you can be fairly certain that it is real.
Graph 1 shows the anomalies from the GISTEMP Global Land and Ocean Temperature Index (LOTI), plotted against Year. It also shows a LOESS curve calculated using a 10-year local regression. This LOESS curve provides a “smoothed” version of the temperature anomaly graph.
( don’t tell anybody, but look at how the LOESS curve is flat from about 2004 to 2011 )
Graph 2 shows the GISTEMP 10-year warming rates, calculated for each year. The 10-year warming rate is calculated for each year, by doing a linear regression for the period from 5 years before the reference year, to 5 years after the reference year. It also shows a LOESS curve calculated using a 60-year local regression, over the 10-year warming rate data. This LOESS curve provides a “smoothed” version of the warming rate graph, and shows what the long-term warming rate has been.
Now it is time for some mental gymnastics. Graph 3 shows the difference between the 10-year warming rate, and the long-term warming rate, for each year.
In simple terms, this graph shows whether we were warming faster, or slower, than the long-term warming rate, for each year. It was determined by calculating the difference between the blue line (the 10-year warming rate), and the orange line (the long-term warming rate), from Graph 2.
It should be immediately obvious that there was a recent slowdown, some time after the year 2001. Time to stop denying it, Alarmists!!! But was this slowdown “significant”. Please note, that I am NOT using the word “significant” in the “statistically significant” sense. I am not going to do any statistical tests on the slowdown, although I will calculate the standard deviation of the “difference” data, and plot the data with lines for 1, 2, and 3, standard deviations. I am interested in looking at the recent slowdown, and comparing it to other climate events, like speedups, other slowdowns, and periods of warming rate stability.
My interest is more in shapes and slopes, rather than statistics. Although I will use statistics when they are useful. There is a good reason why statistics are not very useful when studying the slowdown. Temperature data is very noisy. Looking for a 10 to 15 year slowdown in the temperature data, is like looking for a black cow, on a moonless night. There might be a black cow there, but it is very difficult to see. Alarmists claim, with absolute certainty, that there are no black cows in existence on a moonless night, because they can’t see one (or perhaps they don’t want to see one). And if you claim that a black cow does exist, then they will call you a “denier” (because you are denying the non-existence of a black cow).
Graph 4 shows the same data that was shown in Graph 3. The difference between the 10-year warming rate, and the long-term warming rate, for each year. But Graph 4 also includes lines to show the warming rates corresponding to 1, 2, and 3, standard deviations from the long-term warming rate.
It should be immediately obvious that the warming rate for the recent slowdown, was about 1 standard deviation below the long-term warming rate. This means that the warming rate for the recent slowdown, was NOT particularly statistically unusual. Have a look. There are a number of times, since 1970 (when global warming became notable), when the warming rate has been close to 1 standard deviation lower than the long-term warming rate.
You may be wondering whether I have suddenly become an Alarmist. I did, just for a second. But then I regained my common sense. Alarmists, stop thinking so one-dimensionally. There is more to a slowdown, than just the warming rate. Here is a clue. I talked earlier about slowdowns have more than one dimension. Something about strength, and …
… Length. One of the most obvious things about the graph of the 10-year warming rate, is that it is almost always changing rapidly, and making abrupt changes of direction. I think that it would be fair to say that it is almost always steep (going up or down), and the word “pointy”, could be used to describe the graph.
EXCEPT DURING THE RECENT SLOWDOWN. During the recent slowdown, the graph became NOT as steep (flatter?), and NOT as pointy (more rounded?).
Alarmists may wish to accuse me of trying to turn climate science into poetic imagery. Have I got any “hard” numbers to back up my fancy assertions? As a matter of fact, I do.
There is a lot of debate about just how low the warming rate must be, to qualify as a slowdown. Technically speaking, any amount of slowing is a slowdown. If you are driving your car at 100 km/h, and your speed changes to 99 km/k, then technically, you have slowed down. But in real terms, this is a trivial slowdown.
I am going to try and avoid the issue of specifying, in absolute terms, what criteria are used to define a slowdown. Instead, I will use 3 equally probable warming rate categories, to define what is a warming rate that is near the long-term warming rate, what is a warming rate that is considerably lower than the long-term warming rate, and what is a warming rate that is considerably higher than the long-term warming rate.
The data for the 10-year warming rate difference from the long-term warming rate, has an average very close to zero, and a standard deviation of about 1.53. If we assume that the data follows a normal distribution (it does approximately), then we would expect about 1/3 of the values to be below -0.66 (below the long-term warming rate by more than 0.66). We would also expect about 1/3 of the values to be above +0.66, and about 1/3 of the values to be between -0.66 and +0.66.
This allows us to categorise the data easily into 3 approximately equally probable categories. We can give each category a numerical label, and a description:
- -1 = a warming rate that is considerably lower than the long-term warming rate (below the long-term warming rate by more than 0.66 degrees Celsius per century)
- +0 = a warming rate that is near the long-term warming rate (within 0.66 degrees Celsius per century of the long-term warming rate)
- +1 = a warming rate that is considerably higher than the long-term warming rate (above the long-term warming rate by more than 0.66 degrees Celsius per century)
The criteria that I am using, may seem a bit theoretical. However, we can easily relate them to absolute values, and also to everyday situations. If you do a linear regression on the GISTEMP Global Land and Ocean Temperature Index from 1970 to 2018, then you will get an average warming rate of about +1.80 degrees Celsius per century.
So my criteria of “below the long-term warming rate by more than 0.66 degrees Celsius per century”, becomes (approximately) “below the warming rate of (+1.80 – 0.66) = +1.14 degrees Celsius per century (a decrease in the warming rate of about 37%).
My criteria of “above the long-term warming rate by more than 0.66 degrees Celsius per century”, becomes (approximately) “above the warming rate of (+1.80 + 0.66) = +2.46 degrees Celsius per century (an increase in the warming rate of about 37%).
And my criteria of ” within 0.66 degrees Celsius per century of the long-term warming rate”, becomes (approximately) “within the warming rate range from +1.14 to +2.46 degrees Celsius per century (a warming rate range which goes from about 37% below the long-term warming rate, to about 37% above the long-term warming rate).
These criteria may still seem too theoretical for some people. To make it more understandable, we can relate it to an everyday activity, like driving a car. Imagine that you are driving your car in the city, at an average speed of 50 km/h. Using my criteria, we would say that you are no longer traveling at the average speed, when your speed changes by more than about 18 km/h. You will be below the average speed, when your speed falls below about 32 km/h. And you will be above the average speed, when your speed exceeds about 68 km/h.
These criteria seem reasonable to me (for climate, if not for driving), and these criteria will be used in this article.
So, we will assign each month from 1970 to 2014, to one of the 3 warming rate categories:
- considerably below the long-term warming rate (-1)
- considerably above the long-term warming rate (+1)
- near the long-term warming rate (+0))
and then plot the resulting series on a bar chart. The result is interesting. See the following bar chart.
Look at the width of the bar for the recent slowdown. It is 61 months wide. Wider than any other blue bar on the chart, apart from the speedup bar that goes from about 1975 to 1981 (which is 69 months wide).
To help describe this bar chart, we will first define some terms:
•…A slowdown is defined as a number of consecutive months, where the warming rate was considerably below the long-term warming rate
•…A speedup is defined as a number of consecutive months, where the warming rate was considerably above the long-term warming rate
•…A climate event is defined as a slowdown or a speedup
•…The LENGTH of a slowdown (or speedup), is the number of consecutive months where the warming rate meets the criteria for being a slowdown (or speedup)
•…the STRENGTH of a slowdown or speedup, is defined as the average departure from the long-term warming rate
This means that the recent slowdown is:
•…the LONGEST SLOWDOWN since 1970
•…the SECOND LONGEST CLIMATE EVENT since 1970….(a climate event is a slowdown or a speedup)
•…the STRONGEST SLOWDOWN since 1970….(tied for first place with the slowdown that went from about mid-1990 to the start of 1993 (a width of 32 months))
The recent slowdown had an average strength of -1.4 degrees Celsius per century, and a length of 61 months.
The slowdown that went from about mid-1990 to the start if 1993, which I will call “the previous slowdown”, also had an average strength of -1.4 degrees Celsius per century, but it had a length of only 32 months.
<<<<<< — start of confession — >>>>>>
ok, I may have stretched the truth a little bit here. By about 0.001 degrees Celsius per century.
The strength of the recent slowdown, to 3 decimal places is -1.446
and the strength of the previous slowdown, to 3 decimal places is -1.447
So technically, the previous slowdown was stronger than the recent slowdown. But the previous slowdown only lasted for 32 months, whereas the recent slowdown lasted for 61 months (nearly twice as long).
So in terms of slowing the increasing temperature, the recent slowdown had a bigger effect, because it lasted longer.
<<<<<< — end of confession — >>>>>>
After looking at this bar chart, can you honestly say that there was no recent slowdown?
As a matter of interest, I tried to make the 3 categories equally probable, based on a normal distribution. On the bar chart there were 527 months in total.
- 145 months (28%) were in the “lower than the long-term warming rate” category
- 227 months (43%) were in the “near the long-term warming rate” category
- 155 months (29%) were in the “higher than the long-term warming rate” category
It appears that the distribution of warming rate differences (from the long-term warming rate), is more clustered towards the centre, than a normal distribution. But the distribution of warming rate differences does appear to be fairly symmetrical.
The results of this article, about there being a recent slowdown, are not affected by the distribution not being perfectly normal. If anything, the existence of the recent slowdown is strengthened, because there is a smaller probability of a month being classified as “considerably below the long-term warming rate”.
I can not end this article, without addressing a particularly incorrect idea about the recent slowdown, which is being spread by some Alarmists. This is the idea that the recent slowdown (which they call the “Pause”) only exists because of the 1998 super El Nino.
I won’t tell you who they are (but one of then goes by the name of “jgnfld”). He agreed with another Alarmist, who said:
“The “pause” was derived by starting from the top of the warmest el Niño of the 20th century, 1997-1998.”
This is absolute nonsense, and can easily be disproved. The strongest slowdown (the one with the lowest warming rate), went from 2002 to 2012, and had a warming rate of +0.14 degrees Celsius per century. Because it went from 2002 to 2012, it had nothing to do with the 1998 super El Nino.
The strongest slowdown which included the year 1998 (the one with the lowest warming rate), went from 1998 to 2013, and had a warming rate of +0.96 degrees Celsius per century).
So the false Alarmist slowdown (1998 to 2013), had a warming rate which was
(0.96 / 0.14) = 6.9 times greater than the warming rate of the real slowdown (2002 to 2012).
In summary, if the real slowdown (2002 to 2012) was a car that was traveling at 50 km/h, then the false Alarmist slowdown (1998 to 2013), would be a car that was traveling at 345 km/h. Perhaps this is one of the reasons why Alarmists don’t believe that there was a slowdown, they are not looking at the real slowdown.
Anybody who wants to read more about this issue, should read my article at:
In this article, I replace the high temperature anomaly for 1998, with an “average” temperature anomaly, and show that the slowdown still exists.