While reading over at TallBloke’s, there was this article:
That was largely a repost and link to this article:
That does a very nice job of showing why I found GISS product GIStemp to be “unfit for purpose” and not a valid temperature representation. It does algorithmic adjustments and extreme (IMHO broken) “homogenization” to a copy of GHCN monthly averages (and an average is no longer a temperature), with a snippet of Antarctic data added and a semi-blending of USHCN monthly averages. This after NCDC has done a boat load of “adjustments” and their own form of ‘homogenized pasteurized data food product’ homogenization on it. (This, often after the individual Met. Offices of various countries have already ‘fixed’ via adjustment and homogenization of the data they collected…)
The farther away from actual daily MIN / MAX readings, the further you get from temperatures. ONLY the actual readings of MIN / MAX and current temp are a temperature. ALL Averages are no longer temperatures, but are a statistic about the original temperatures and no longer represent temperature or heat content. So “don’t go there” is the best idea, but GIStemp goes there in spades and on steroids. This posting shows the result.
They go back to the Iceland Met Office and get the almost-original data only slightly adjusted and homogenized. (IMHO that is still an error, but a smaller one than doing it recursively a few times…) Then compare that to GIStemp “data food product”. The difference is dramatic. The reality on the ground in Iceland is not dramatic at all.
I don’t think I can improve on the article, nor do I think I need to quote the whole thing for effect or to preserve it, so just “hit the link” and read it. Also look at the nice and effective graphs.
Then realize that is what is being done globally in all of the three main suppliers of such “data food products”; NCDC, GISS, and Hadley. They all use very similar methods and start from the same basic input data, so are essentially the same product with only minor variations.
I have a load of articles looking at specific details of the GIStemp product and NCDC “issues”, and they can be found in the various ‘category’ listings at the sidebar of this blog. In particular the dT/dt alternative method, the GIStemp specific, and the NCDC specific. Also note that there is an ‘entry point’ at the top of this blog for GIStemp that has a map to the other articles.