It seems like it’s been cold for almost the entire winter this year, with the exception of the few days last week when we got more than 16 inches of snow. Unfortunately, it’s been hard to enjoy it, with daily high temperatures typically well below -20°F.
Let’s see how this winter ranks among the early-season Fairbanks winters going back to 1904. To get an estimate of how cold the winter is, I’m adding together all the daily average temperatures (in degrees Celsius) for each day from October 1st through yesterday. Lower values for this sum indicate colder winters.
Here’s the SQL query. The CASE WHEN stuff is to include the recent data that isn’t in the database I was querying.
SELECT year, CASE WHEN year = 2012 THEN cum_deg - 112 ELSE cum_deg END AS cum_deg, rank() OVER ( ORDER BY CASE WHEN year = 2012 THEN cum_deg - 112 ELSE cum_deg END ) AS rank FROM ( SELECT year, round(sum(tavg_c) AS cum_deg FROM ( SELECT extract(year from dte) AS year, extract(doy from dte) AS doy, tavg_c FROM ghcnd_obs WHERE station_id = 'USW00026411' ORDER BY year, doy ) AS foo WHERE doy between extract(doy from '2011-10-01'::date) and extract(doy from '2012-12-19'::date) GROUP BY year ORDER BY year ) AS bar ORDER by rank;
And the results: this has been the fifth coldest early-season winter since 1904.
|O/N/D rank||year||O/N/D cumulative °C||N/D rank|
In addition to the ranks for October through today (O/N/D rank in the table), the last column (N/D rank) shows the same calculation without October temperatures. It’s always a good idea to examine how well a relationship holds up when the interval is manipulated in order to see if the results are an artifact of the choice of period. In this case, the rankings change, but not dramatically.
Tonight we may cross -50°F for the first time this year at our house, but even without the very cold temperatures predicted (but not record cold) through the weekend, the start of the 2012/2013 winter has been exceptionally chilly.