This week in #CoolClimateData we’re continuing our dive into the large collection of up-to-date climate data hosted by Professor Ole Humlum at Climate4you.com. Today it’s the compilation of surface air temperature data sets, which Professor Humlum divides into three quality classes. In the first class are the satellite data sets we discussed last week, where Humlum has long pointed to the UAH data set as the most reliable, a judgment recently confirmed by NOAA scientists. The second and third quality classes are temperature datasets based on thermometer measurements on land and sea. He judges the best of these to be that produced by the Hadley Centre in the UK, which is called the HadCRUT record. (This frankly ugly half-acronym is for the Hadley Centre/ University of East Anglia Climate Research Unit Temperature measure.) Humlum favours their ocean record because it does a better job of matching the Argo float record, another cool data set we’ll discuss later. Mind you, like all the thermometer-based data sets HadCRUT shows the imprint of much monkeying around and “adjustments”. But they aren’t as bad as the kludges applied to the other data sets, which in modern climate science is actually praise.
The HadCRUT4 record since 1979 looks like this:
As you see, there was a warming trend through the 1980s and 1990s, then the famous, if contentious, “hiatus,” a long pause until 2016 when an El Niño caused a big warming spurt, after which the record has been gradually settling down ever since.
Humlum produces what he calls “maturity” diagrams, in which he compares the current version of the data to the version the same team published in 2008. If the data are being tweaked to show a greater warming trend, it would show up as cooling adjustments in the current version for earlier years and warming adjustments for later ones.
In the case of the HadCRUT record, the adjustments do indeed add a bit of a trend, as illustrated by this chart which shows not temperatures or temperature changes, but adjustments to the temperature by year. It was produced by taking the 1850-2008 HadCRUT record as it is currently published and subtracting the same HadCRUT record as published in 2008. So if a given year is red it means the current HadCRUT figure for that year is higher than their 2008 figure and if blue that it’s cooler:
Suspiciously, there’s a lot of blue before 1980, especially after 1895, indicating cooling adjustments, though also a lot of red. But from 1980 onward they are all warming adjustments. And it’s noteworthy that the period from 1940 to 1975, when falling temperatures triggered a cooling scare despite rising CO2, seems to have been, well, adjusted upward. But at least there are warming adjustments in the 1800s as well.
Now here’s the same chart but for the NASA global temperature record rather than the HadCRUT. And it indicates a much bigger hot red thumb on the scale:
The IPCC and climate science crowd see nothing wrong with all the adjustments and manipulations of the temperature record, even when the apparent warming trend turns out to be mainly a result of fudging the numbers. Which is not cool, especially since if you asked a group of competent statisticians to create the adjustment chart you’d expect if someone were manufacturing a warming trend you’d pretty much get the one immediately above.
So we prefer to use the satellite record whenever possible, in line with Professor Humlum’s recommendation.
Following Tony Heller's lead, you should mark on these graphs periods when the glaciers were retreating due to a warming trend, and periods where they were advancing due to a cooling trend. You will see that the advances and retreats do not match the reconstructed thermometer record very well. This method empirically refutes the adjustment methods employed by the alarmists.
If adjustments were wholly legitimate, they would publish them, and the algorithyms used.
"Adjustments" are just another word for tampering.
A number of issues raised here need clarification.
1. The article says "The HadCRUT4 record since 1979 looks like this:" but then shows the a plot of the HadCRUT5 data.
2. A sweeping generalization (or implied generalization) about adjustments is likely to be misinterpreted. What's needed is an explanation of why the data is adjusted - the output is temperature estimates, not adjusted data - and the potential errors in the adjustment method (e.g. false assumptions).
3. The meaning of "HadCRUT" is only partially correct. The "Had" part is for the UK Met Office Hadley Centre, which deals with sea surface temperatures, and the CRU for the Climatic (not Climate) Research Unit at the university of East Anglia, which processes data from weather stations. They each develop their own datasets - HadSST4 and CRUTEM5 - and merge them to create HadCRUT5.
4. The change in HadCRUT data, specifically the global average temperature anomaly, surprises me only to the amount of blue in the plot. Temperature adjustments/ estimates are typically biased downwards due to the inability of the methods to deal with gradually increasing non-meteorological distortions but being much better at dealing with the sudden changes that "correct" those situations (e.g. a station relocation or vegetation cleared). A weather station move to a cooler location will mean that all past data gets adjusted downwards, those becoming the estimated temperatures that would have been measured with the current instruments in the current location. Temperature anomalies are calculated relative to 1961-1990 averages, so if the data in that period is "adjusted" then all of the anomalies will change. (P.S. Not that I said "all past data gets adjusted downwards". This means that the distorted data is adjusted by the same amount as the very early undistorted data, which means the false trend caused by non-meteorological influences is still present.)
5. I believe that HadSST4, the sea surface temperature component of HadCRUT5, also had some new tweaks to bring them to what they would be at the (ostensibly) same depth. HadSST3 had 100 alternatives but HadSST5 has 200 alternatives, all based on guesses about what method was used to measure the temperature in the instances that no methodology was noted. Each guess then means a different adjustment. The commonly used version, which I think means the version used for HadCRUT5, isn't the mean value of the 200 guesses but the median value (although it makes about 0.02 C difference in most years). This type of work might also account for some of the difference between HadCRUT versions.
Your "Cumulative graph of HadCrut monthly average global average surface temperature changes" is EXTREMELY interesting.
https:notrickszone.com/wp-content/uploads/2021/1/Arch_2.png
How was it generated? It looks like a smoothing of very noisy data