Along those lines, it's really quite silly that we record temperatures to the nearest 0.1C because the accuracy is really about +/- 2C (19 times out of 20 anyway). Site location is huge. Rocks, buildings, pavement, trees, etc. really impact the temperature readings. Plus, with modern thermometers, they use semiconductors which are non-linear and always drifting. That makes calibration is pretty much impossible, especially during really hot days or really cold days. The old school alcohol and mercury thermometers were more accurate at the extremes, although they also have problems when it comes to evaporation over time to the top of the thermometer such that they start reading too low as time goes on if they're not replaced regularly.
Pretty much the only thermometer record I trust is the -63.0°C/81.4°F set in Yukon in 1947...
North America's cold extreme was not an isolated event in either time or space. Extreme cold conditions were the norm over Alaska and northwestern Canada for much of the Winter of 1946-47. A new low-temperature record for the Yukon had been set 13 December 1946 at Mayo, 260 km (163 miles) by air northeast of Snag, when a numbing minus 72 °F (minus 57.8 °C) was recorded. (In this essay, I am sticking to degrees Fahrenheit as the prime unit of temperature measure since that was the standard for all of North America at the time. The equivalent Celsius reading will be given in parentheses.)
Meteorologists attributed the extended cold weather to a strong zonal (westerly) circulation in the upper atmosphere across North America that trapped arctic air over Alaska and northwestern Canada for much of the winter. This built a cold dome of intensely frigid air over the Yukon. The air mass, which spilled into northern British Columbia, the Northwest Territories and eastern Alaska, reached its coldest during a period beginning about January 27 and lasting until 4 February. Many sites set record cold temperatures for January and February that still stand. For example, British Columbia's all-time officially recorded low temperature descended on Smith River on 31 January: minus 74 °F (minus 58.9 °C).
As the cold, heavy air mass lingered over the Yukon, its coldest air settled near the surface and drained into low-lying valleys. As February dawned, the weather observers at Snag reported clear skies and calm winds — ideal conditions for extremely cold temperature readings — with patches of ice fog. Temperatures during the night nearly slid off scale and during the day rose no higher than minus 50 °F (minus 45.6 °C).
Few dared venture forth on Groundhog Day (2 February) to look for shadows as the morning temperature dropped to the last mark on the standard alcohol minimum thermometer: minus 80 °F (minus 62.2 °C), a new record low for the continent. [In regions where extreme cold temperatures may occur, the thermometers used are alcohol-based rather than mercury-filled. Mercury freezes at about minus 39 °F/°C.] By 2 pm (YST), the temperature had risen to only minus 51 °F (minus 46.1 °C). Shortly after the day's temperature peak, the alcohol dropped rapidly over the succeeding five hours and then declined slowly through the night.
Weather officer-in-charge Gordon M. Toole kept vigil on the minimum thermometer that night. As he hurried from the log barracks to the instrument shelter some thirty metres away, Toole could feel the cold invade his parka. He recalled clearly hearing dogs barking in the distant village and a tinkling as his breath, frozen instantly in the cold, fell as a white powder to the ground below.
Shortly after 7 am on the morning of 3 February 1947, Toole cautiously opened the instrument shelter door, taking care not to breath on the instruments inside. Using his flashlight to illuminate the minimum thermometer, he saw that the sliding scale within the alcohol column (used to register the minimum temperature) was below the lowest scale marking: minus 80 °F ( minus 62.2 °C).
Toole quickly returned to the barracks and convinced a colleague to come out and witness the mark. Using a set of dividers to determine the slide's position below the minus 80 mark, Toole estimated the reading of about minus 83 °F ( minus 63.9 °C). The previous record low had lasted but one day!
Scientists at the head office of Canada's weather service had anticipated the possibility of a temperature reading under minus 80 °F. Therefore, they had recommended in such a situation that the weather observer mark the lowest level with a pen and then send the thermometer to Toronto for determination of the temperature. Toole found a pen would not work under such cold temperatures and used a small file to etch a mark in the glass.
Accounts of events recalled fifty years later differ as to the reaction of staff at the time of the event. Toole recalls no real interest, but according to weather observer Wilf Blezard (in an interview with reporter Greg Ralston of the Yukon News in 1997): "We had to put a lock on the door of the instrument screen because everyone was rushing out and looking at the thermometers...."
On the record-setting day, the morning observation reported a surface pressure of 1037 mb, calm winds and visibility of 32 km (20 miles). In some directions, visibility was reduced by patches of ice fog, most noticeably over the area where a dog-team was hitched. The snow on the ground measured 38 cm (15 inches) but, due to the intense dryness of the air, was decreasing at a rate of about 1.3 cm (a half an inch) per day. The record low was recorded at 720 am (YST), an hour and twenty-two minutes before sunrise. The high for the day would reach only minus 56 °F (minus 48.9 °C).
To confirm the record low temperature, the thermometer was removed and sent to the Canadian Weather Service headquarters in Toronto. Three months later after extensive testing, the temperature was officially certified at minus 81.4 °F (minus 63 °C), a new record for cold in North America, which still stands today.
The cold that day, however, was not solely confined to Snag. Fort Selkirk on the Yukon River, 180 km east-northeast of Snag, claimed a reading of minus 85 °F (minus 65 °C), but it could not be considered "official" because the thermometer hung on the outside wall of a building rather than being housed in a standard instrument shelter. Mayo apparently also fell to the minus 80 °F range on that historic morning, but the official temperature could not be confirmed as the weather station, its instruments and records were destroyed by fire around midnight 15 February. A surviving photograph of the instrument did show a reading around minus 80 °F, just higher than the Snag minimum. Other notable lows reported during the period are shown on the accompanying map and chart.
You do not have the required permissions to view the files attached to this post.