Winter Solstice question

[RayProudfoot]

Hi Steve,

I'm running v3.8.8 build 865. Today is the winter solstice but Cumulus is reporting that tomorrow will be 5 seconds shorter than today.

It's not a huge problem but I'm just curious how this is calculated so any possible rounding or calculation issues are solved for C2.

[daj]

I found this on the web....

Not like a clock

But as the hours of daylight grow between now and June 21, 2010, don't look for the increase to grow by equal amounts of time on either side of noon each day. Sunrises will still come later each day into next month, even as sunsets continue to come later in the day, as they have since early this month.

The reason for the apparently out-of-sync changes? The shifting orientation of Earth's tilt with respect to the sun, along with the elliptical shape of Earths' orbit, combine to alter the times of sunrise and sunset unequally compared with what you might expect using a clock's 24-hour day.


PS - I am 7 seconds shorter tomorrow so you have 2 seconds on me!!!

[RayProudfoot]

David,

Thanks for the reply but I think you miss the point. You're quite correct when you say the sun continues to rise later each morning in late December. However the sun has already started to set later - 15:52 today compared to 15:50 a week ago at my latitude.

But the definition of the shortest day is that the next day cannot be shorter. It has to be longer. Only by a few seconds I grant you but nevertheless, longer. According to WeatherUnderground Dec 22 will be 2 seconds longer than today for my location. So i was curious about the formula that Steve uses as it appears to be slightly inaccurate.

[TNETWeather]

You can look up what the value should be using:

http://www.timeanddate.com/worldclock/a ... html?n=853

Says it should be +1 sec tomorrow (Cumulus 1.8.8 says -3sec).

The actual solstice takes place at: 2009-12-21 17:47 UTC

Link to london (same site):
http://www.timeanddate.com/worldclock/a ... html?n=136

Says length of day tomorrow is also +1sec

[RayProudfoot]

Thanks for the links Kevin. I independently found a site that confirmed the solstice occured at 17:47 today and that tomorrow would be a couple of seconds longer.

[TNETWeather]

It would seem to be an easy thing to calculate but I suspect that isn't.

I would imagine you need to have correct Lat and Long for the site along with the height of the location.

I've seen the formula before, but converting it into something that I can use is not something I'm good at.

[steve]

The routines I use are part of this library: http://tpsystools.sourceforge.net/

I'm pretty confident that I'm using it correctly, but I'm wondering if the times are a day out. Now it's the 22nd, mine is now showing tomorrow to be 2s longer than today. It's all much more complicated than you would believe, I have to convert from local to UTC and then to Julian days, and then I have to work out which day it's actually referring to when I convert it back.

As a kludge I could add a day to the days I'm comparing, but that's quite horrible without understanding why it's wrong.

If anyone wants to provide a better library, I'd be happy to use it.

[steve]

I experimented with changing the date back to around the summer solstice, and it wasn't until the 23rd June that it said the next day would be shorter, so it's not just a case of adding a day (the solstice was on the 21st).

So I'm afraid it's a case of live with it or provide me with an alternative implementation.

[steve]

I've tried the example application which comes with the library, and that also gives the 22nd as being shorter.

[Repairman77]

Think this confirms it as 21st December...

It's related to the height of the Sun (Astronomically speaking)...

http://en.wikipedia.org/wiki/Solstice

Mike.

[steve]

I think the problem may be that the library has a (perfectly adequate) accuracy of 1 minute, so the calculation to work out the difference in minutes and seconds (which I argued against adding, but caved in as usual) is meaningless.

[RayProudfoot]

The routines I use are part of this library: http://tpsystools.sourceforge.net/

I'm pretty confident that I'm using it correctly, but I'm wondering if the times are a day out. Now it's the 22nd, mine is now showing tomorrow to be 2s longer than today.

Hi Steve,

I think that's it. My station is reporting tomorrow (23rd) will be 1 second longer than today. So maybe you just need to subtract 1 day and it will be spot on.

[RayProudfoot]

I think the problem may be that the library has a (perfectly adequate) accuracy of 1 minute, so the calculation to work out the difference in minutes and seconds (which I argued against adding, but caved in as usual) is meaningless.

I posted my reply just as you added this. In that case, not to worry. It is only obvious on two days a year.