3 ms·

> We can completely ignore the fact that over the years the average location of all the contributing stations has moved slightly northward to colder territory[.

> We can completely ignore the fact that over the years the average location of all the contributing stations has moved slightly northward to colder territory[.] Woah, that's not something I'd considered. Are weather stations that contribute to NOAA operated by the volunteer public, kind of like Weather Underground? Either way, that's pretty interesting that there's a distinct northward migration in at least one "demographic."

gok 8y ago
That graph is...questionable. The entire Y axis only covers 55 kilometers. I’m sure the NOAA has more principled ways of accounting for station movement.
- jtbayly 8y ago
  So the answer appear to be that, yes, he can sucker an awful lot of people.
- martinpw 8y ago
  55km is not insignificant though. Some very rough estimates - 55km is 0.5% of the distance from the equator to the pole. If the temperature range between those two extremes is 50C then the shift corresponds to a net cooling of 0.25C, which seems like it would have a noticable impact on those graphs. Obviously there is more to it than that, but I don't think it makes it a non-issue.
prolikewh0a 8y ago
>Are weather stations that contribute to NOAA operated by the volunteer public, kind of like Weather Underground? This would provide extremely poor data. A lot of people on Wunderground are running ~$100 Chinese weather stations that don't have proper radiation shielding, dubious barometers, and poor anemometers. While they're perfectly fine for hobbyists and accurate to some degree, a volunteer network for climate data would give non-scientific results.
avip 8y ago
Tbh that location graph is meaningless.
trynumber9 8y ago
The population has been migrating south. Why are the stations migrating north? https://en.wikipedia.org/wiki/Mean_center_of_the_United_States_population https://en.wikipedia.org/wiki/Mean_center_of_the_United_Stat...
- puetzk 8y ago
  Note: this is only my guess, but it seems quite plausible that inhabited areas have been well-sampled for a long time, but some combination of 1) automated stations and satellite communication making it much easier/cheaper to sample remote locations 2) use of whole-globe numerical weather models which require complete initial conditions, even for regions where nobody much cares about the output 3) increased interest in climate research would lead to a greatly increased number of Alaskan stations operated by NOAA (and of course corresponding moves by other national agencies)
roter 8y ago
But everything is area-weighted, so if we move a southern station northward, that means the southern stations previous neighbours will need to carry more weight. The station moved northward will steal some of the weight from its new northerly neighbours. So, it should only impact the accuracy of the results, not necessarily shift things colder.
- martinpw 8y ago
  You are correct, but the point of the article is that the people manipulating the data deliberately avoid these types of corrections. They see a cooling effect in the raw data and say the scientists who apply corrections are the ones who are actually fudging the data.

gok 8y ago

That graph is...questionable. The entire Y axis only covers 55 kilometers. I’m sure the NOAA has more principled ways of accounting for station movement.

jtbayly 8y ago
So the answer appear to be that, yes, he can sucker an awful lot of people.
martinpw 8y ago
55km is not insignificant though. Some very rough estimates - 55km is 0.5% of the distance from the equator to the pole. If the temperature range between those two extremes is 50C then the shift corresponds to a net cooling of 0.25C, which seems like it would have a noticable impact on those graphs. Obviously there is more to it than that, but I don't think it makes it a non-issue.

prolikewh0a 8y ago

>Are weather stations that contribute to NOAA operated by the volunteer public, kind of like Weather Underground? This would provide extremely poor data. A lot of people on Wunderground are running ~$100 Chinese weather stations that don't have proper radiation shielding, dubious barometers, and poor anemometers. While they're perfectly fine for hobbyists and accurate to some degree, a volunteer network for climate data would give non-scientific results.

avip 8y ago

Tbh that location graph is meaningless.

trynumber9 8y ago

The population has been migrating south. Why are the stations migrating north? https://en.wikipedia.org/wiki/Mean_center_of_the_United_States_population https://en.wikipedia.org/wiki/Mean_center_of_the_United_Stat...

puetzk 8y ago
Note: this is only my guess, but it seems quite plausible that inhabited areas have been well-sampled for a long time, but some combination of 1) automated stations and satellite communication making it much easier/cheaper to sample remote locations 2) use of whole-globe numerical weather models which require complete initial conditions, even for regions where nobody much cares about the output 3) increased interest in climate research would lead to a greatly increased number of Alaskan stations operated by NOAA (and of course corresponding moves by other national agencies)

roter 8y ago

But everything is area-weighted, so if we move a southern station northward, that means the southern stations previous neighbours will need to carry more weight. The station moved northward will steal some of the weight from its new northerly neighbours. So, it should only impact the accuracy of the results, not necessarily shift things colder.

martinpw 8y ago
You are correct, but the point of the article is that the people manipulating the data deliberately avoid these types of corrections. They see a cooling effect in the raw data and say the scientists who apply corrections are the ones who are actually fudging the data.