Well, sometimes you get a surprise.
We saw that the Southern Hemisphere was showing no net warming, but the Northern Hemisphere data had most of the ‘warming signal’ and that it had been enhanced in v3 compared with v1. I expected to find mostly flat changes in various S.H. continents / Region Codes.
What I found is in some ways more interesting. Not only is Africa not in a warming trend in v3, but the trend is actually LESS than it was in the v1 data. Typically when something is being “screwed up” the choice is between “malice and stupidity”. Forensics looks for evidence that “something happened”, but then you hand it over to management to decide if that “something” was a case of malice (or sometimes criminal intent) or was it just stupidity (accidents happen… but stupid helps raise the rate.).
Now a very clever crook will try to arrange what looks like “stupid” to be involved, so that if someone sees suspicious things, they get involved in ‘fixing the stupid’ and don’t notice the nefarious. But that is extraordinarily rare. Most of the time “Stupid just is”. Per Hanlon’s Razor the presumption has to be that Africa got flattened because the folks “finding warming” in the data are just doing crap science and not deliberately putting it in, though that possibility still exists.
It is always possible that in some backwards way when Finagling the Data they didn’t notice the continent scale impacts and were looking at grid/box impacts and this was just an accidental artifact of Finagling; so it doesn’t prove stupidity instead of malice, it just leans that way.
Most of the change between the data sets happens in the distant past. Temperatures within about 1/3 C of zero back to 1900 with about a 1/2 C drop in the cold ’60s-’70s. Guess they don’t have a lot of major international airports with acres of tarmac and thousands of cars and jets flying in / out on 2 minute intervals in most of Africa ;-)
Yes, I know, most of the country is close to the equator and things just don’t change much there… but that’s sort of the whole point, isn’t it? That “Global Warming” isn’t global if it doesn’t happen at the equator and doesn’t happen in the Southern Ocean and only really shows up in Canada and Siberia where there “are issues” with the data record. We’ll see who warms their data the most as we visit other continents. 2 down and 5 to go.
But add Africa (all of it) to the place that gets a “Climate Guilt Indulgence”. As the Southern Hemisphere collectively gets an indulgence, that just leaves N. America, Europe, and Asia as potentially “causal” and deserving of Penance Taxes. Personally, I’m hoping Asia shows up “warming” as then we can repudiate our $Trillion of debt to China as a “Climate Debt”… but that would be politically incorrect, so I don’t expect to see that in the data. Still, “expecting at the data” doesn’t do much good. So we’ll see.
The other “big deal” in this comparison is just that from the 1880s-1908 era to 1990 the v1 data showed nearly 1 C of warming in Africa. The v3 data shows dead flat. So which is it? Was v1 lying or is v3 buggered? Or is it just that we can get 1 C of “wandering” in the results based on what thermometers are put in the data series?
IMHO the data we have does not support making any assertion about past temperatures with a precision of more than 1 or 2 C and that is largely due to the very few thermometers that existed even until recently. This dP/dt code only uses anomalies of a single thermometer record to itself, and it starts with the most recent and best readings to make that anomaly basis. It does not remove “splice artifacts” that comes from joining a segment of trend in one thermometer with that from another thermometer.
End effects like starting measurements in a cold year and ending them in a hot year can put in an artificial bias to a series. Average several of those together and you will get a ‘warming trend’ based on nothing but splicing those end effect influenced segments. I deliberately don’t do any splice artifact removal processing attempts simply to keep the process clear and clean and secondarily because I want to know if there are splice artifact ‘features’ in the data. IMHO, codes like GIStemp find “global warming” simply due to just such splice artifacts, but they are hidden in the “homogenizing” code where folks don’t see them. (The question of ‘malice or stupidity’ on that behaviour of the codes will be left for others to decide).
To the extent that these trends in the data reflect such splice artifact prone behaviour from the data, they also reflect that same risk in the present climate codes. Goose, meet gander… So I do think it is very valid to look at these graphs and simply say: “The trend varies by 1 C from release to release for an entire Continent. The data are crap and we can’t say anything about fractional degrees C of warming.”