By now I have pictures of 4 separate locations, and still do not have a complete record. I do not even now how to adjust for what is known. It’s a safe bet that the roof of a downtown office building was warmer than the entrance gate to the small airport. But how much warmer?

The existence of century-old weather data was at first beguiling. Just adjust for knowable biases, and use statistical treatment over a number of stations to tease a signal out of the background noise. But after looking at a good number of these stations, my personal conclusion is that it’s an impossible project. They have almost all been moved multiple times and it’s impossible to reconstruct previous environments.

Of course, there are other reasons to document these stations. The COOP system is a remarkable success story of volunteers who work in the public interest. Many of their personal histories are fascinating, particularly those of the 19th century. Somebody should write a book. : > )

That’s what people really need to correct the temperature not what GHCN and the all the rest do which is lazy maths.

Like Max and Min temperatures tells the whole story- there are temperature inversions.

“GSN: GHCND Version”

as everything after that heading stays bolded.

@xyzlatin:

Not all data sets are created equal. The GHCN keeps the short records. That is then fed into GIStemp, that tosses out anything sorter than 20 years (in the STEP2 processing just prior to the PApars.f program). So depending on where you stick your ladle in the river you get different results. IMHO the more “upriver” from the effluent producers the better

@Allan Williams: Even the “GHCN UN-Adjusted” data are rather adjusted. I’ve yet to find actual “raw” sources. The “QA process” can fabricate data based on averages of nearby stations and can change the values of items.

http://chiefio.wordpress.com/2010/04/11/qa-or-tossing-data-you-decide/

Don’t know the size of the impact, though.

@gallopingcamel: Yes, the GHCN set has dropped from over 6000 at peak activity to about 1200 now. Though in the last couple of years they have started adding stations back in (not clear if they are adding warming stations… but it looks like it.)

@ThosTHos: how to pick warming vs cooling sites? Easy. Very easy. First off you could just pick places with planned economic growth. Like, oh, Airports… I’m sure it’s just an accident that 92% of active thermometers are now at airports in France (and a similar number in the USA) for the GHCN data set… Furthermore, as the GHCN is a “historic creation” (as the apologists for the station dropping like to point out – dodging the point that it’s still a CHOICE what goes in…) it’s easy to run a code like “first differences” on the station data and find what’s warming and what’s not. I’ve just done this for GHCN for the entire world. Took me about a month, but I was only working on it part time:

http://chiefio.wordpress.com/2010/04/11/the-world-in-dtdt-graphs-of-temperature-anomalies/

The code I wrote for this is called dT/dt and it’s maybe 2 dozen lines long of active code. Oddly, I found many cooling countries. More oddly, most of them have their data end prior to today… In Southern Africa we have a very unphysical warming in one country that persists to today, but it is surrounded by countries where the falling or flat series end. They, of course, will be “filled in” from the one that was rising in codes like GIStemp.

I’m sure it’s just an oversight…

For now I am continuing my analysis without this information. It would have been so nice to run a full comparison, but that can be added in when I’m done with the rest.