Yesterday a co-worker and I were talking about how we weren’t able to enjoy the new snow because the weather had turned cold as soon as the snow stopped falling. Along the way, she mentioned that it seemed to her that the really cold winter weather was coming later and later each year. She mentioned years past when it was bitter cold by Halloween.
The first question to ask before trying to determine if there has been a change in the date of the first cold snap is what qualifies as “cold.” My officemate said that she and her friends had a contest to guess the first date when the temperature didn’t rise above -20°F. So I started there, looking for the month and day of the winter when the maximum daily temperature was below -20°F.
I’m using the GHCN-Daily dataset from NCDC, which includes daily minimum and maximum temperatures, along with other variables collected at each station in the database.
When I brought in the data for the Fairbanks Airport, which has data available from 1948 to the present, there was absolutely no relationship between the first -20°F or colder daily maximum and year.
However, when I changed the definition of “cold” to the first date when the daily minimum temperature is below -40, I got a weak (but not statistically significant) positive trend between date and year.
The SQL query looks like this:
SELECT year, water_year, water_doy, mmdd, temp FROM ( SELECT year, water_year, water_doy, mmdd, temp, row_number() OVER (PARTITION BY water_year ORDER BY water_doy) AS rank FROM ( SELECT extract(year from dte) AS year, extract(year from dte + interval '92 days') AS water_year, extract(doy from dte + interval '92 days') AS water_doy, to_char(dte, 'mm-dd') AS mmdd, sum(CASE WHEN variable = 'TMIN' THEN raw_value * raw_multiplier ELSE NULL END ) AS temp FROM ghcnd_obs INNER JOIN ghcnd_variables USING(variable) WHERE station_id = 'USW00026411' GROUP BY extract(year from dte), extract(year from dte + interval '92 days'), extract(doy from dte + interval '92 days'), to_char(dte, 'mm-dd') ORDER BY water_year, water_doy ) AS foo WHERE temp < -40 AND temp > -80 ) AS bar WHERE rank = 1 ORDER BY water_year;
I used “water year” instead of the actual year because the winter is split between two years. The water year starts on October 1st (we’re in the 2013 water year right now, for example), which converts a split winter (winter of 2012/2013) into a single year (2013, in this case). To get the water year, you add 92 days (the sum of the days in October, November and December) to the date and use that as the year.
Here’s what it looks like (click on the image to view a PDF version):
The dots are the observed date of first -40° daily minimum temperature for each water year, and the blue line shows a linear regression model fitted to the data (with 95% confidence intervals in grey). Despite the scatter, you can see a slightly positive slope, which would indicate that colder temperatures in Fairbanks are coming later now, than they were in the past.
As mentioned, however, our eyes often deceive us, so we need to look at the regression model to see if the visible relationship is significant. Here’s the R lm results:
Call: lm(formula = water_doy ~ water_year, data = first_cold) Residuals: Min 1Q Median 3Q Max -45.264 -15.147 -1.409 13.387 70.282 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) -365.3713 330.4598 -1.106 0.274 water_year 0.2270 0.1669 1.360 0.180 Residual standard error: 23.7 on 54 degrees of freedom Multiple R-squared: 0.0331, Adjusted R-squared: 0.01519 F-statistic: 1.848 on 1 and 54 DF, p-value: 0.1796
The first thing to check in the model summary is the p-value for the entire model on the last line of the results. It’s only 0.1796, which means that there’s an 18% chance of getting these results simply by chance. Typically, we’d like this to be below 5% before we’d consider the model to be valid.
You’ll also notice that the coefficient of the independent variable (water_year) is positive (0.2270), which means the model predicts that the earliest cold snap is 0.2 days later every year, but that this value is not significantly different from zero (a p-value of 0.180).
Still, this seems like a relationship worth watching and investigating further. It might be interesting to look at other definitions of “cold,” such as requiring three (or more) consecutive days of -40° temperatures before including that period as the earliest cold snap. I have a sense that this might reduce the year to year variation in the date seen with the definition used here.
We got another couple inches of snow last night, and while we don't yet have enough for me to ski to work, it certainly looks like winter. Winter in Fairbanks may be cold (OK, very cold), but it's also very sunny and because the angle of the sun is so low, the light is orange and gorgeous.
Yesterday I saw that CBS was going to show the Dolphins this morning. Ho hum, another almost unwatchable low-definition sporting event on our local CBS network. But it’s the Dolphins, so I watched anyway. When the picture started to flicker at 7-2, I switched to the other location for CBS, 13-1, and like magic, it was in High Definition! I don’t know if this is just for live sporting events (like the NBC affiliate in Fairbanks), but it sure looks great.
Even stranger, the Dolphins are currently beating the Giants 14–3. If they’re not careful, they might win a game this season!
The photo above was taken just before the intersection of Sheep Creek / Goldstream Roads and Murphy Dome Road, less than half a mile from where I work, and about four miles from our house as the crow flies. I’ve spent the last three hours backing up data, and getting our servers ready for possible evacuation. We’re both downwind (currently) and downhill from the fire, but it’s dangerously dry in Fairbanks right now, and if the wind shifts, we find ourselves evacuating our offices.
More as it happens…
The cold snap of 2008/2009 has finally broken. The plot below comes from my Ninety day weather summary plot for the Fairbanks Airport weather station (PAFA). That web page has a description of all the variables shown on the plot, but here I’m interested in the vertical red bars, which represent the high and low temperatures for the day. You’ll notice that from December 27th through January 11th, the temperature was far below normal (the 30-year normal high and low temperature are the solid blue, mostly horizontal lines. Record high and low temperatures are the green x marks. We didn’t come close to breaking any low temperature records.
So how does this cold snap compare to those in the past? Most people cite the cold snap of 1989 as the one they remember. Was the current period comparable?
My first attempt at answering this question was to generate a list of consecutive days where the low temperature for the day was below -40. It’s a fairly complex query because we need to compare a table against itself, and we need to report the results from both rows that match. Here’s the SQL:
SELECT a.dt, a.min_temp, a.temp, a.max_temp FROM isd_daily AS a INNER JOIN isd_daily AS b ON a.isd_id=b.isd_id AND a.dt=b.dt - INTERVAL ’1 DAY’ WHERE a.isd_id=’702610-26411’ AND a.min_temp<=-40 AND b.min_temp<=-40 UNION SELECT b.dt, b.min_temp, b.temp, b.max_temp FROM isd_daily AS b INNER JOIN isd_daily AS a ON a.isd_id=b.isd_id AND a.dt=b.dt - INTERVAL ’1 DAY’ WHERE b.isd_id=’702610-26411’ AND a.min_temp<=-40 AND b.min_temp<=-40 ORDER BY DT
The output is a list of consecutive dates and temperatures. From there, I wrote a Python script to read these periods and calculate the length, average temperature, average minimum, and the cumulative degree days below freezing. Using this metric, our recent cold snap was the coldest since the cold snap of 1989. Here are the two compared:
For period between 1989-01-19 and 1989-02-01 Days = 14 Average temp = -42.9 Average minimum = -48.0 Cumulative degrees below freezing = 1048.7 For period between 2009-01-02 and 2009-01-11 Days = 10 Average temp = -40.8 Average minimum = -44.8 Cumulative degrees below freezing = 728.0
You can see that this doesn’t include the first portion of the cold snap, because the low temperature on New Year’s Day was a balmy -38°F. So it doesn’t seem like it really captures the full length of the recent event.
The other approach I tried was to examine the entire historical record, calculating the average temperature for all periods of various lengths. It turned out that a window of 16 days resulted in the highest ranking. But even so, it was the 62nd coldest 16 day period between 1946 and the present. You can see the reason why the ranking is so low if you look at the results:
rank Date interval avg temp (°F) 1 1947-01-17 - 1947-02-01 -47.19 2 1947-01-18 - 1947-02-02 -47.06 3 1947-01-19 - 1947-02-03 -46.84 4 1947-01-16 - 1947-01-31 -46.82 29 1989-01-17 - 1989-02-01 -41.43 30 1989-01-18 - 1989-02-02 -41.27 32 1989-01-16 - 1989-01-31 -40.96 35 1989-01-19 - 1989-02-03 -40.58 47 1989-01-20 - 1989-02-04 -39.56 48 1989-01-15 - 1989-01-30 -39.53 62 2008-12-27 - 2009-01-11 -38.07
The top four coldest 16-day periods were all part of the same, longer cold snap, that lasted from January 16, 1947 through February 3rd of that year. Because the entire period was so cold, it yielded the top four coldest 16-day periods. I also included the intervening 1989 cold periods for reference. You can see that the 1989 event was also much longer than the 14-day interval identified by the consecutive days below -40 technique. The average temperatures over the 1947 events are quite impressive when compared to the recent average and the 1989 event. As bad as it seemed, we don’t really have much too complain about.