Beam me to Planet Gore !
This affects all data from the Artic.
As Anthony Watts pointed out at Watts Up With That, the Eureka station registered the biggest rise in temperature probably seen on the Earth’s surface: 86ºC in one hour, on March 3, 2007!
Now this data is available on Weather Underground, but seems not to exist in Environment Canada. The graph differences are clear below:
But that seems not to be the case in other examples. Take January 1st, 2007, for instance. Both Weather Underground and Environment Canada agree: there was a mighty spike at noon. Seems like the “M” problem affects both:
[Here's the METAR data with the missing "M", note at 11AM the M reappears]
There are times where differences are not so big, but the “M” problem is still there. Check the images from Weather Underground and Environment Canada for September 26, 2006:
Other times, changes are so significant, that something must be wrong. Check out the temperature rise on June 20, 2005. On the left, the weekly graph from Weather Underground shows a great surge in temperatures, confirmed by the Environment Canada graph for the day.
=====================================
Thanks to Ecotretas for his sleuthing, I’m sure there are many more like this yet to be discovered. It seems with Eureka, more is going on than METAR errors. The temperature rises reported seem impossible given the meteorological conditions, and because they seem to be automated, suggest sensor error or perhaps sensor environment contamination (like a vehicle or other heat source). If you look at this 1997 image from Wikipedia (and click it to get the super hi-res version and pan around) you’ll see a number of vehicles near buildings. Where is the temperature sensor? – I don’t know, but if someone can find out it might shed some light on this mystery.
Well, well——IF YOU REPEAT A LIE OFTEN ENOUGH
Yes, it does! Not only do temperature spikes (bad data) affect averages, when the number of stations recording data is small, the effect is amplified to give higher readings on average all around.
If stations are closed, IIRC, the practice was to average data from adjacent stations to fill in the dataset. With averages similarly corrupted by questionable data, the overall picture would reflect a trend which did not, in fact, exist.
Thanks for posting this!