Reposted from the NoTricksZone
By P Gosselin on 19. February 2022
By Frank Bosse, via Die kalte Sonne
It’s known that GISS does not process the temperature data as they are determined but rather “corrects” them. Such corrections may make sense, but to what extent they influence warming trends remains an open question. Gavin Schmidt, the director and responsible person for the measurement series, recently tweeted some peculiar things:
Image: Twitter screenshot
He notes that the overall trend decreases with the corrections (green compared to the raw data (black dashed). Now, however, it is the case that the anthropogenic influence on the warming only becomes noticeable after 1950.
If we now digitize this part of the overall series, a completely different picture emerges:
Although the trend decreases from 1880 to 1950, this is relatively insignificant for calculations of climate sensitivity, for example, because very little anthropogenic forcing occurred during this time.
After 1950, however, we see a 13% trend shift due to the “corrections”. This was also quickly communicated on Twitter and one has to wonder why Gavin Schmidt mentioned this irrelevant reduction in early trends so prominently. Smokescreen?
What the NASA GISS graph says to me is that NASA would have us believe that some time around 1945 meteorologists around the world suddenly lost the ability to accurately read a thermometer but fortunately have gradually recovered that ability in the 50 years since.
Or maybe NASA GISS just decided to cleave the temperature record at that point and tilt the second half so it more or less joined up at the current date.
And if anybody is interested I have several bridges to sell. All in prime locations around the world.
It has nothing to do with reading a thermometer. The post WWII difference is primarily the result of a transition from bucket observations to engine and buoy observations with the former/later having a low/high bias relative to the later/former. A secondary contributing factor is the change in time-of-observation and sheltering/instrument changes. The net effect of all factors is a significant high bias in the warming trend over the entire period of record and slight low bias in the warming trend in the later part of the period of record. It has nothing to do with observers ability to record temperature observations.
Tell how these “biases” were determined. Are they claimed to be compared against calibrated thermometers and recommended methods?
What are the uncertainties associated with each type, especially buckets and engine readings? Does “adjusting” for so-called bias eliminate the uncertainties?
What happens to the uncertainty in a GAT temperature when extremely uncertain SST’s are averaged in? The ocean cover about 2/3rds of the earth. How does high uncertainty temp measurements for this amount of the globe affect the uncertainty of a GAT that uses them?
Haung et al. 2015
Haung et al. 2020
Your first link doesn’t show what you claim it does.
“Finally, estimates of uncertainty of its SST reconstruction (so-called parametric uncertainty) were not provided in ERSST.v3b, and therefore parametric uncertainty was not included in the total uncertainty of SSTs in ERSST.v3b. Studies have shown that the parametric uncertainty is an important component of the total uncertainty as demonstrated in the latest Hadley Centre dataset, HadSST3 (Kennedy et al. 2011). These have been estimated in this new ERSST.v4 in the accompanying Part II paper (Liu et al. 2015, hereafter Part II).” (bolding mine, tpg)
parametric uncertainty: “Parameter uncertainty includes measurement errors, sampling errors, variability, and use of surrogate data.”
In other words the uncertainties are GUESSES! Guesses introduce BIAS.
I haven’t had a chance to look at the second link. My guess is that it won’t be any different.
Let me make sure I have your argument correct. Are you saying your primary criticism with Haung et al. 2015 is their choice of words?
Nope. It’s the fact that the corrections are *estimates*. Estimates are guesses, they are *NOT* measurements. Guesses introduce bias.
Both your references use models as observational data. The 2015 one uses Model GFDL CM2.1 and the 2020 one uses LSATs from “coupled model simulations”.
No uncertainties from these models were discussed whatsoever.
The 2015 had the following:
– global averaged SST slightly overestimated
– may come at a cost of accurately portraying biases
Do these statements lead one to be confident in the overall conclusions?
I’ll sya it again, the overall reason for this is to obtain long records that don’t have overweighted short term records. This is not a valid scientific reason to create new information that replaces measured and recorded data. If bucket data is considered unreliable, then throw it away. If ship intake temps are unreliable, then throw them away. Trying to use a model to determine what these temps SHOULD HAVE BEEN is a joke. They are NOT creating new data, they are manufacturing information to replace measurements with.
Couple this with homogenization, i.e. guessing at what a value *should” be, where the uncertainties of the stations being used to do the infilling and you get data that simply cannot be believed.
Peculiar how Watts has not done any analysis of the statistical significance of the difference in trends that he is pointing out.
Peculiar how Watts didn’t look at the trend starting in the mid 70’s, when anthropogenic warming unequivocally emerged from the background.
Peculiar how he uses the word ‘peculiar’.
Peculiar how the article is by Pierre Gosselin 🙂