Watts Up With That?
April 24, 2010

UPDATE: I’ve been in contact with the Eureka Weather Station manager, and despite the incredible nature of the temperature jumps that are outside of normal experiences, such as July 14 2009,* they do appear to be real. The METAR coding errors are another issue. I have many pictures now, plus a plan map, and a better meteorological insight than what can be gained from the meager online metadata. I’ll have an updated post later tonight or tomorrow – Anthony

Ecotretas find more cases of missing “M”s and mangled data in METAR reports making it into climatic temperature data.* I’ve reposted his findings below and added some of my own notes [in brackets] to strengthen his findings. This time it’s Eureka, Nunavut, Canada. From “Up Here” magazine:
The first High Arctic Weather Station in history, Eureka was established in April 1947 at 80-degrees north latitude in the vicinity of two rivers, which provided fresh water to the six-man United States Army Air Force team that parachuted in. They erected Jamesway huts to shelter themselves and their equipment until August, when an icebreaker reached Eureka – as it has every year since – and brought permanent buildings and supplies. For decades after that, small, all-male crews would hunker down for entire winters, going a little stir-crazy from the isolation.
Photo by Eric Charlton from Flicker used under creative commons license*

As I pointed out last Saturday there’s a common reporting flaw in world meteorological stations that use the METAR weather data format to report their hourly temperatures. Just one bad report in a cold location in the Arctic or Antarctic is enough to throw off the whole month’s worth of data when averaged. And it is monthly data that is used for climate. The all time high temperature error identified below has found it’s way into Wikipedia as “factual” when it is clearly wrong.

Some METAR and other kinds of unidentified errors also seem to be finding their way into official Environment Canada data. I’ve made screencaps. Given the importance of this weather station as the last surviving* GHCN station that far north, also used in GISS, you’d think better quality control would be done, particularly when EC has a statement about data quality on each data page. What has been found seems to point to a corrupted dataset there.

- Anthony


Eureka by Ecotretas

[Images above from GISS - click to enlarge]

Eureka, in Nunavut Canada, is a very special meteorological station. As can be seen in the first image above, it is responsible for the very big stripe on the very top of Canada. As can be seen on the second image, the data from only one station is responsible for a very big percentage of global temperatures… Both graphs can be obtained here.

So, one would imagine that data for this station is quality proof. We found two sources for temperature data: at Weather Underground and at the National Climate Data and Information Archive.

The first interesting data about this station is its record high temperature, which according to Wikipedia was reached on July 14, 2009, with 20ºC. [image below added by Anthony]

screencap from Wikipedia’s climate section on Eureka station -click to enlarge*

[UPDATE: Wikipedia editor removed the sentence above within 8 hours of this posting. See discussion here]

On Weather Underground, the monthly page for that day does say that 20ºC was the maximum temperature. But when you check the METAR data, the maximum temperature was 14ºC.

[Here is where it really gets strange, I've added two screencaps from Environment Canada, on for the monthly data, another for the daily data from July 14th, 2009. They don't match! The 20.9C value never appears in the July 14th hourly data. Click images below to enlarge, EC's July 2009 Monthly report on the left, July 14th, 2009 daily/hourly data on the right. Perhaps EC corrected the error in the daily/hourly data, but missed the monthly?* - Anthony]

[click images above to enlarge]

[It also shows up as 20.9C on the EC yearly report for 2009, shown below]

Checking the Environment Canada page, the maximum for the date was 14.4ºC.

Things were different on the day before, July 13th. Maximum temperature for Weather Underground was also 20ºC, while at Environment Canada was 19.6ºC. But if you check the graphs below, some special heat occurred at 10PM, when temperatures soared some 15ºC!

(click images to enlarge)

[Image below added by Anthony. Here is the METAR report for Eureka, via Weather Underground, the error is highlighted in yellow, note the jump in temperature followed by a fall the next hour]

Eureka METAR coded and decoded 07-13-09 click to enlarge*

[In the image below added by Anthony, note the implausible temperature jump in the Environment Canada data for July 13th, 2009 at 22:00 where it was 5.9C at 21:00, jumping to 19.6C at 22:00, then back down to 4.1C at 23:00. Looking at the weather conditions of clear sky and moderate sustained winds from the North ~ 24 km/hr, there does not appear to be a meteorological explanation. Looking at the METAR data above, it appears the temperature was rounded up by the observer to 20C from 19.6. This makes me wonder if the event could be caused by something like a plane, truck, or snow-cat parked briefly near the sensor. In the Arctic, vehicles are left to idle, as turning them off allows them to freeze up, sometimes never to be restarted. ]

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.

The fact that the obvious error on July 14th is now cited in Wikipedia as an “all time record high”, when it doesn’t appear in the EC hourly data, is troubling. – Anthony