Guest essay by Sheldon Walker
For a long time, there has been strong disagreement over the idea of a recent slowdown in global warming. Many people believe that there never was a slowdown. While other people have argued about when the slowdown occurred, and why.
This article will use the GISTEMP global temperature series, to investigate whether the slowdown is real. I know that some people will throw up their hands in horror, at the thought of using the GISTEMP temperature series. I have deliberately used GISTEMP, because it has a reputation as a heavily adjusted, warmist friendly, temperature series. If I can show strong evidence for a recent slowdown in GISTEMP, then the recent slowdown can not be denied.
The first thing that we need to establish, is when the slowdown occurred. Some people insist that the slowdown started in 1998. 1998 was a very strong El Nino year, and temperatures climbed abnormally high because of this. Some people claim that the temperatures in the years that followed 1998, only appear to rise more slowly than normal, because of the abnormally high temperatures in 1998. This explanation is often used to push the idea that the slowdown was not a “real” slowdown. It was just caused by an abnormally warm 1998.
This is a wonderful story. Unfortunately, it is almost totally incorrect. The scientific way to find out when the slowdown occurred, is to look at the evidence in GISTEMP. We need to look at the warming rate for different date ranges, and try to identify a date range with a warming rate that is significantly lower than the normal warming rate. If we cannot find a date range with a warming rate that is significantly lower than the normal warming rate, then the slowdown can be regarded as fake news.
Following is an explanation of how date ranges are specified in this article.
Date ranges are specified using 2 years, e.g. 2000 to 2001.
A year with no month specified means January of that year.
So 2000 to 2001 means January 2000 to January 2001, which is a date range one year in length.
So 2002 to 2012 means from January 2002 to January 2012, which is a date range 10 years long.
This method makes the calculation of the length of a date range very easy, just subtract the first year from the second year, and the answer is the length in years.
But, the date range does NOT include the second year specified in the date range, except for January. E.g. 2000 to 2005 does NOT include the data for 2005, except for January 2005.
In order to find out when the slowdown was, I calculated the warming rate for every possible date range that started in or after 1990, that was 10 years or more in length. I ignored date ranges which were less than 10 years in length, because short date ranges have a more variable warming rate, and would not provide good evidence of a significant slowdown. This gave me a table of date ranges which I sorted by warming rate, from lowest to highest. I threw away all of the rows in this table, except for the first 10 rows. This left me with a table holding the 10 date ranges with the lowest warming rates. This table held the best possibilities for the slowdown.
The following table shows the 10 date ranges which had the lowest warming rates. Date ranges had to start in 1990 or later, and had to be of length 10 years or greater. The column headed “Warm Rate” is the warming rate in degrees Celsius per century.
To check whether the slowdown could have started in 1998, look at the following table. This table shows the warming rate for every possible date range of length 10 years or greater, that started in 1998.
The warming rates in Table 2 are considerably higher than the warming rates in Table 1. This means that the date ranges starting in 1998 are not as significant (in the slowdown sense), as the date ranges in Table 1. This shows that 1998 is NOT an important year for the slowdown.
Looking back at Table 1, we can see that the starting years are 2001, 2002, 2003, and 2004. All of these starting years are near each other. The ending years are 2012, 2013, 2014, and 2015. All of the ending years are also near each other. This is not unexpected. Most slowdowns have some “core” years when the slowdown is strongest. But the slowdown does not suddenly start full strength, or end suddenly from full strength. There will be a few years at the start and end when the slowdown is increasing in stength, or decreasing in strength. If you add these increasing/decreasing years to the core years, then you still get a slowdown, but one which has less strength than the core years.
Because it has the lowest warming rate of any date range (+0.09 degrees Celsius per century), I am going to use the date range from 2002 to 2012 as the “core” years of the recent slowdown. This makes it a 10 year slowdown.
To make this result easier to visualize, I have plotted some graphs which compare the slowdown decade to the 3 previous decades. Because all of the date ranges are 10 years long, it should be possible to compare “apples with apples”.
So I will be graphing the relative temperature anomalies and linear regression lines, for
- 1972 to 1982
- 1982 to 1992
- 1992 to 2002
- and 2002 to 2012 (the slowdown decade).
I have given each temperature series a common time axis, which runs from year 0 to year 10. I have shifted each temperature series to have the same starting value of zero. This makes comparison easier because all of the temperature series start at the same point.
4 temperature series and 4 linear regression lines on the same graph, is a little crowded. So I have created 2 versions of the graph. Both are based on exactly the same data, but highlight different aspects.
The first graph shows each temperature series (and the corresponding linear regression line) in a different color. The slowdown decade is shown in red.
The second graph is based on exactly the same data, and shows the linear regression lines for each temperature series. The slowdown decade is shown in red.
When looking at this graph, remember that the slope of the regression line is the warming rate for the temperature series.
If you look at Graph 2 and think, “that looks like 3 parallel sloping lines, and one flat line”, then you have made a very accurate observation.
Table 3 shows the warming rates for various date ranges.
You can see that the 3 earlier decades had warming rates of +3.07, +2.90, and +3.08 degrees Celsius per century. This compares to a warming rate of +0.09 for the slowdown. This means that the earlier decades each had a warming rate of between 33 and 35 times the warming rate of the slowdown.
While putting together Table 3, I found that the average warming rate from 1970 to 2017 was +1.78 degrees Celsius per century. This is considerably less than the warming rates for the 3 decades that I compared to the slowdown. The three 10 year decades that I compared to the slowdown were actually “speedups” (a greater warming rate than average), when compared to the average warming rate from 1970 to 2017.
Some people may consider it “unfair” to compare the slowdown to 3 speedups, when trying to determine if the slowdown is significant. I think that this is a valid viewpoint, and I will therefore repeat my analysis of the slowdown, but this time comparing it to the average warming rate from 1970 to 2017 (which was +1.78 degrees Celsius per century).
Graph 3 compares the warming rates (slope of the linear regression line) for 3 date ranges. I have made sure that this graph is drawn using exactly the same scale as Graph 2, so that there are no tricks to fool people into believing that there is a slowdown.
This graph shows the regression line for 1992 to 2002 (one of the speedup intervals), the regression line for 1970 to 2017 (the “average” warming rate for a long date range), and the regression line for 2002 to 2012 (the slowdown). I have given all of the regression lines a common origin, at (0.00, 0.00).
It can be seen that the speedup interval has the greatest warming rate, the long date range has the next greatest warming rate, and the slowdown has the lowest warming rate (quite close to zero). In my opinion, the graph makes it clear that the “average” warming rate is closer to the speedup warming rate, than it is to the slowdown warming rate. This means that the slowdown warming rate is quite different to the “average” warming rate.
As can be seen from Table 3, the average warming rate from 1970 to 2017 was 20 times the warming rate from 2002 to 2012 (the slowdown). I think that most people would consider this to be a fairly large difference. Imagine if your income was reduced to 1/20 of its current value, or if it was increased to 20 times its current value. Most people would find those changes significant.
To make the slowdown more understandable, consider the following analogy. Imagine that you are driving along on the motorway at 100 km/hour. Suddenly you encounter roadworks on the motorway. You are forced to reduce you speed to about 5 km/hour, and must keep to that speed for some time. That is the equivalent of the slowdown. But you are lucky, because you don’t have to stay at 5 km/hour for the next 10 years.
Is that enough to convince everybody that there was a recent significant slowdown? I strongly doubt it. Warmists will continue to deny the slowdown. The fact is, that this was a slowdown that lasted for 10 years, and that had a warming rate that was:
- not just 1/2 of the normal warming rate
- not just 1/4 of the normal warming rate
- not just 1/8 of the normal warming rate
- not just 1/16 of the normal warming rate
- but about 1/20 of the normal rate.
I am sure that this will be quickly ignored or forgotten. Look, there is a squirrel over there!
I would like to point out that this analysis of the slowdown is very easy to repeat. Anybody who doubts my results, can download the GISTEMP temperature series and calculate the linear regressions for the date ranges that I used. Anybody familiar with Excel (or a similar product) can easily repeat the calculations.