Agree to Disagree
Debate in a non-hostile environment
Comparing temperature series
Most people have strong opinions about which temperature series are correct, and which are incorrect. Many skeptics claim the GISTEMP has been incorrectly adjusted, and shows too much warming. Many warmists believe that UAH is not calculated correctly, and shows too little warming.
Who is right? That is the question that this article will attempt to answer. There is only one thing that we can be certain of. No matter what this article finds, the outcome will be disputed by at least one group of people.
To make the comparison as valid as possible, this article will look at 4 “major” global temperature series. To make identification easy, each temperature series will be referred to by a 3 letter name. The 4 temperature series are: GIS (often called GISTEMP), HAD (often called HadCRUT4), UAH, and RSS.
GIS and HAD are both surface temperature series, which use temperatures from the Land and the Ocean. UAH and RSS are both satellite temperature series, based on the same satellite data, but calculated in different ways. As well as comparing the individual temperature series, this comparison will also provide a comparison between the surface and satellite temperature series.
Because the satellite temperature series began just before 1980, all temperature series will be compared beginning in January 1980, and going up to September 2018. September 2018 is the most recent month with data for all 4 temperature series, when this article was written.
To provide a detailed comparison, this article will do more than just compare the Global temperature series. Because the Northern and Southern Hemispheres can be different from each other, the comparison will include comparing the Northern Hemispheres for each temperature series, and comparing the Southern Hemispheres for each temperature series.
Because we are dealing with many different temperature series, we need a convention which will make it clear which temperature series we are talking about.
- There are 4 major temperature series (GIS, HAD, UAH, and RSS).
- Each major temperature series has 3 temperature series
(GL – Global, NH – Northern Hemisphere, and SH – Southern Hemisphere).
- Each temperature series has a number of formats
(RA – Raw Anomalies, MA – Modified Anomalies, RL – Raw LOESS, ML – Modified LOESS).
- The term “Modified” refers to values which have been adjusted so that the January 1980 LOESS curve value is equal to zero. This gives all temperature series a common starting point.
Each temperature series will be given a name, of the form “XXX-YY-ZZ”
- where XXX is the major temperature series (GIS, HAD, UAH, and RSS).
- and YY is the region of the earth (GL – Global, NH – Northern Hemisphere, and SH – Southern Hemisphere.
- and ZZ is the format of the data (RA – Raw Anomalies, MA – Modified Anomalies, RL – Raw LOESS, ML – Modified LOESS)
So the UAH Southern Hemisphere modified anomalies, will be called UAH-SH-MA. Where possible, each major temperature will be plotted using a constant colour. The colour scheme is:
- GIS = blue
- HAD = green
- UAH = red
- RSS = orange
How will the comparison be done? All 4 temperature series have temperature anomalies which jump around a lot, from month to month. This makes comparison difficult. This article will “tame” all of the temperature series, by calculating a LOESS curve from the monthly data. The LOESS curve calculation will use a local regression size of 10 years (120 months). A local regression size of 10 years was chosen because this makes the LOESS curve responsive to temperature changes, but not too responsive. This will give a “smooth” LOESS temperature curve, which doesn’t jump around too much, but still shows what the temperature has done. Then all of the LOESS curves will be adjusted to have a temperature anomaly of zero at January 1980. This starts all of the temperature series at a common point.
To show how the temperature series were processed, we will use GIS as an example. There is little point in showing this for all of the temperature series, because they were all processed in exactly the same way, and showing them all would fill the article with unnecessary graphs.
We start with the raw GIS global temperature anomalies. A LOESS curve is then calculated from this raw data. The GIS RAW data and the GIS RAW LOESS curve are shown on the following graph.
Next, the Y-axis value of the raw LOESS curve is determined, for January 1980. This value is subtracted from all of the raw LOESS temperature anomalies, to give the modified LOESS temperature anomalies. This causes the LOESS curve to shift vertically on the graph, so that the Y-axis value for January 1980 is zero.
The same value is subtracted from all of the raw temperature anomalies, to give the modified temperature anomalies. This causes the plotted raw temperature anomalies to shift vertically on the graph, to align with the modified LOESS curve.
The following graph shows the GIS MODIFIED temperature data, and the MODIFIED LOESS curve. If you compare this graph to the previous one, then you should see that they are exactly the same, except for the position of the data on the Y-axis. It is important to realise that shifting a temperature series vertically, to give it a common LOESS starting point, does not change the temperature series in any important way. It is similar to adjusting the base period for a temperature series.
The following graph shows the first comparison, the 4 Global temperature series. The immediate impression is that 3 of the major Global temperature series agree quite well (GIS, HAD, and RSS). The one that is most different from the average, is UAH. It is interesting that UAH agreed with the others until about the year 2000. Then UAH went sideways, while the others kept increasing. Even while it is different, UAH runs parallel to the others (has a similar warming rate), from about 2006 to 2018.
We will next look at a similar comparison for the Northern and Southern hemispheres.
The following graph shows the second comparison, the 4 Northern Hemisphere temperature series. The immediate impression is that this is similar to the Global comparison. 3 of the major Northern Hemisphere temperature series agree quite well (GIS, HAD, and RSS). The one that is most different from the average, is UAH. It is interesting that UAH agreed with the others (fairly well) until about the year 1997 (in the Global comparison it agreed until about 2000). Then UAH went sideways, while the others kept increasing. Even while it is different, UAH runs parallel to the others (has a similar warming rate), from about 2006 to 2018.
We will next look at a similar comparison for the Southern hemispheres.
The following graph shows the third comparison, the 4 Southern Hemisphere temperature series. This comparison is different to the last two. Instead of having 3 temperature series which agree quite well, and one which is different, we now have 4 temperature series that agree quite well. The 3 temperature series which agreed in the Global and Northern Hemisphere comparisons, agreed very closely. The 4 temperature series which agree in the Southern Hemisphere comparison, don’t agree as closely as the other comparisons, but there is general agreement. We no longer have one temperature series which is clearly different from the others.
This is a strange situation. The temperature series UAH, agrees with the other temperature series for the Southern Hemisphere, but disagrees with the other temperature series for the Northern Hemisphere and Global.
Since the Global temperature series is the average of the Northern and Southern Hemispheres, we would expect the difference between UAH and the others to be twice as big in the Northern Hemisphere series, as in the Global series. If you check the graphs, then you will see that this is approximately right, where UAH disagrees with the others.
At this point, I think that it is worth comparing all of the Northern and Southern Hemisphere temperature series on one graph. See the following graph. The colours for each major temperature series are the same, but the Northern Hemisphere series are solid lines, and the Southern Hemisphere series are dashed lines.
I am going to point something out, which will probably upset some people. It is meant to be a simple observation, and should not be taken as a criticism of anything or anybody. The UAH Northern Hemisphere temperature series looks more like a Southern Hemisphere temperature series. I don’t know why it is like that.
While I am upsetting people, I will make another observation. Since about 2006, the UAH Southern Hemisphere temperature series has been almost identical to the GIS Southern Hemisphere temperature series. One temperature series is preferred by skeptics, the other by warmists, but they are almost identical.
That is all that I am going to say for now. I don’t have the answer to the question about which temperature series are correct. I have my suspicions, but no hard evidence. I have written this article, so that you can see the evidence, and draw your own conclusions. I would like to hear from anybody who can make sense of the evidence.