Winter musings

Most people know that the winter solstice December 21 corresponds to the shortest day of the year in the northern hemisphere. But some go further and think that this date also corresponds to the time of latest sunrise and earliest sunset. That is not true. The earliest sunset occurs around December 8 and the latest sunrise around January 4.

The reason for this difference is that although sunset starts beginning to arrive later in the evening after December 8, sunrise is also getting later in the morning but at a faster rate, leading to decreasing overall day length. After the solstice, the rate at which sunset gets later outpaces the rate at which sunrise gets later, leading to increasing overall day length. After January 4, the days start getting longer at both ends, sunrise and sunset.

The coldest day of the year is on average around January 20, which is what I like to think of as mid-winter, rather than the solstice. So if we think of winter as the coldest quarter (91 days) of the year with the coldest day in the middle, then winter would be from December 6 to March 6. (See an earlier post for more details on temperature variation.)

This is why Groundhog Day on February 2 cracks me up. Supposedly, if Punxsutawney Phil does not see his shadow, spring will come early. But if he sees his shadow, then we will have to endure a long winter, in fact six more weeks of it. But six weeks from February 2 only takes us to March 16, hardly much longer than the regular end of winter on March 6.

But what I always look forward to is today, January 4. Knowing that the sun will now begin to rise earlier and earlier in the morning makes me feel that spring is around the corner. And after mid-winter on January 20, I feel that spring has pretty much arrived, whatever Phil might do two weeks later.


  1. slc1 says

    These figures may be true for Cleveland but they are not universal. For instance, in the Northern Virginia area, sunset starts to become later about December 15, and the statistically coldest day of the year is about Jan. 12.

  2. flex says


    I’ve always heard that if the groundhog sees his shadow there will only be six more weeks of winter.

    If he doesn’t see his shadow there will be a full month and a half of winter left.

  3. says

    I’m not sure you can say that the statistically coldest day in Northern Virginia is Jan 12. When I do a Fourier analysis on average high and low temperature (data from intellicast) for Leesburg, for the average high, the lowest day works out to Jan. 16. For the average low, it works out to Jan. 23. In reality, both get really, really flat near the peak, so only a Fourier analysis can tease out the difference.

    Here’s the Fourier reconstituted values for average high in that vicinity:

    Jan 11 41.90187
    Jan 12 41.85530
    Jan 13 41.81909
    Jan 14 41.79328
    Jan 15 41.77791
    Jan 16 41.77299
    Jan 17 41.77853
    Jan 18 41.79453
    Jan 19 41.82096
    Jan 20 41.85779
    Jan 21 41.90499
    Jan 22 41.96249
    Jan 23 42.03023
    Jan 24 42.10812

    And here it is for the lows:

    Jan 17 20.14611
    Jan 18 20.09159
    Jan 19 20.04603
    Jan 20 20.00959
    Jan 21 19.98242
    Jan 22 19.96468
    Jan 23 19.95649
    Jan 24 19.95795
    Jan 25 19.96916
    Jan 26 19.99021
    Jan 27 20.02114
    Jan 28 20.06200

  4. Brian E says

    All I know is it’s hot at the moment, and the sun isn’t setting to just before 9pm (daylight savings time). Sun’s up before 6am. Love this time of year. Except the 40+ days can be a grind.

  5. Mano Singham says

    I agree with ahcuah. The curve is really flat at this point. I did a crude estimate by taking the mid point of the days when the temperature did not change at its lowest point but the more accurate method is the Fourier analysis that he did. slc1’s date may correspond to when the time actually changed but since it is recorded only in increments of one minute, that would not be the actual turning point.

    I would expect that the shape of the temperature and time curves (and the locations of the maximum and minimum points) would not differ by location in the northern hemisphere but that only the flatness of the curves would vary.

  6. MNb says

    “this date also corresponds to the time of latest sunrise and earliest sunset.”
    You can notice that this is wrong when you are living in the Guyana’s, which are still on the northern hemisphere. For Moengo, where I live:

    Earliest sunrise: may/june.
    Latest sunrise: january/february
    Earliest sunset: october/november
    Latest sunset: july
    Shortest day:december 21
    Longest day: june 21.
    A difference of almost 40 minutes.

  7. Hamilton Jacobi says

    Is there a simple explanation for why the date of latest sunrise differs from the date of earliest sunset?

  8. slc1 says

    Actually, the figures I cited are for DC, which is some 30 miles SE of Leesburg, Va. and is at a somewhat lower altitude, at least for the Federal District along Constitution Ave. The ave high in early January in DC is 42 degrees and the average low is 28 degrees. It gets rather colder in the suburbs, especially at night.

  9. says

    OK, here are the corresponding numbers from DC (again, intellicast data). Average low bottoms out on January 20; average high bottoms out on January 15. Again, the curve is very flat. Here they are near the high:

    Jan 11 42.11003
    Jan 12 42.07335
    Jan 13 42.04786
    Jan 14 42.03349
    Jan 15 42.03013
    Jan 16 42.03767
    Jan 17 42.05600
    Jan 18 42.08496
    Jan 19 42.12442
    Jan 20 42.17422

    and here they are near the low:

    Jan 15 26.87801
    Jan 16 26.83229
    Jan 17 26.79664
    Jan 18 26.77107
    Jan 19 26.75560
    Jan 20 26.75024
    Jan 21 26.75495
    Jan 22 26.76970
    Jan 23 26.79442
    Jan 24 26.82904
    Jan 25 26.87345

    It’s really not that different than Leesburg (on timing, not magnitude, but that’s the whole point).

    The thing is, across most of the nation, using the 21st is a pretty good heuristic, which was my and Mano’s point.

Leave a Reply

Your email address will not be published. Required fields are marked *