27

I was looking at the sunrise and sunset times where I live (Aberdeen, Scotland) and I noticed something odd: the time between sunrise and sunset in winter is longer than the time between sunset and sunrise in summer. I chose the solstice in both cases and noticed that the summer solstice night is 6 hours and 4 minutes whereas the winter solstice day is 6 hours and 41 minutes. The discrepancy is 37 minutes and looking for a similar ask on google I found nothing and here I could only find this question that asked why later sunsets occur than on the 21st of June and someone pointed out that the aphelion of the Earth's elliptical orbit occurs near the summer solstice, but if this was the reason for the difference, then surely it's be incredibly noticeable, and not just because we'd be missing almost an hour (and perhaps more since this barely accounts for 1/4th of the day so since I'm guessing it'd be proportional we'd have almost 3 hours of difference in day length between the two and clocks would need to be adjusted).

Thank you for the answers! (also I've added screenshots of my app showing the times because idk)

Summer solstice sunrise and sunset times Winter solstice sunrise and sunset times

uhoh
  • 31,151
  • 9
  • 89
  • 293
Nierninwa
  • 373
  • 3
  • 5
  • 1
    What's the name of that app? Different apps can use different calculations, and may not agree on sunrise / sunset times. – PM 2Ring Jun 12 '22 at 06:12
  • The answers on the linked question mention that the day length isn't only affected by the eccentricity of Earth's elliptical orbit. It's also affected by the obliquity: the tilt of the orbit. Also, sunrise / sunset times are affected by your latitude. – PM 2Ring Jun 12 '22 at 06:27
  • 3
    I would suppose that atmospheric refraction plays a role here. When we see the Sun rise, it’s geometrically lower than our horizon; atmospheric refraction brings it up a little. The opposite is true at setting: When we see the Sun set, it’s already set geometrically. This would add a few minutes to the duration of daytime. – Pierre Paquette Jun 12 '22 at 06:44
  • 1
    @PierrePaquette Atmospheric refraction plays a huge role here. Any website that calculates sunrise / sunset times for years in the future / in the past, and for any location on the surface of the Earth will use a canned value for the location of the Sun below the horizon to determine when sunrise / sunset occurs, typically 50 arc minutes below the horizon. – David Hammen Jun 12 '22 at 12:58
  • @PM2Ring I know that my answer to the linked is correct. It's an effect I have studied. The effect is even greater around the December solstice as that is close to when the derivative of the equation of time is close to its most extreme value. That question however has little in common with this question. It's definitely not a duplicate. – David Hammen Jun 12 '22 at 13:23
  • @PM2Ring You helped a lot with you comments on my answer to the linked question with regard to improving the quality of that answer. I learned a long time ago the mantra to "check your ego in at the door" regarding constructive criticism. A young coworker recently found what he thought was a bug in my flight code. He was very deferential, maybe overly so. He mentioned it in scrum. I replied "Yep. That's a bug. Thanks for finding it!" Those who don't follow the mantra are those who are most likely to create nasty bugs. My ego is "checked in at the door" regarding your comments to my answer. – David Hammen Jun 12 '22 at 13:38
  • @David Oh, I agree it's not a dupe. I just wanted to mention that declination & latitude are important. But I should've mentioned refraction & horizon dip due to altitude, too. As I'm sure you know, sometimes refraction near the horizon can produce spooky effects: https://en.wikipedia.org/wiki/Fata_Morgana_(mirage) – PM 2Ring Jun 12 '22 at 21:32
  • @David I also appreciate it when people point out errors in my answers, especially in my code. It's just so easy to make sign errors, or to multiply instead of divide by a factor near 1, even when you've been coding for several decades. – PM 2Ring Jun 12 '22 at 21:36
  • Funny thing, I've always wondered this and indeed I lived by Aberdeen specifically for some years. And indeed I've already asked this question on an SE site years ago (perhaps physics?) and never got a good answer. I can report that if you live near Aberdeen, in fact, the effect is extremely obvious. It's clear as day (joke :) ) that the shortest daylight is not nearly as short as the shortest nightdark. – Fattie Jun 15 '22 at 12:59

1 Answers1

47

Why is the summer solstice night shorter than the winter solstice day?

Aside: If you use your sunrise / sunset calculator you will see that day and night are not equal at the equinoxes. The reason for this disparity is the same as the reason for disparity between the length of shortest summer night versus the length of the shortest winter day.

Explanation

The disparity is primarily a consequence of how sunrise and sunset times are defined and calculated. The calculated sunrise time is defined as the moment when the top of the Sun is expected to first appear above the flat horizon. Similarly, the calculated sunset time is defined as the moment when the top of the Sun is expected to finally sink below the flat horizon. The two key factors that lead to the noted disparity are that

  • The Sun is not a point source. It appears to be a sphere with an apparent diameter of about 32 arc minutes. So even on an airless Earth, sunrise and sunset would occur when the center of the Sun is about 16 arc minutes below the horizon.
  • The Earth's atmosphere typically (but not always) refracts light from celestial objects near the horizon. The troposphere usually has temperatures that decrease with altitude. This temperature gradient refracts light (and also sound; this is why you can sometimes hear far-away trains, but other times you can't). Most sites that calculate sunrise and sunset times use a fixed value of 34 arc minutes for refraction.

Adding the sun's apparent radius and typical refraction angle results in the center of the Sun being 50 arc minutes (0.833 degrees) below the horizon at the calculated times of sunrise and sunset. These two factors account for almost all of the observed disparity between the shortest summer night and the shortest winter day. Most sites do not go beyond the standard 50 arc minutes below the horizon calculation, so on those sites (which is the vast majority of sites that calculate sunrise / sunset times), those two factors alone account for all of the observed disparity.

User d_e made a great comment, which is that the effect depends strongly on latitude. At the equator the path of the center of the Sun at sunrise and sunset is always perpendicular to the horizon (ignoring atmospheric refraction). Even after accounting for the non-zero diameter of the Sun and atmospheric refraction, the path of the Sun is very close to vertical at sunrise and sunset. The 100 arc minute offset (2*50 arc minutes) amounts to a small time difference. Aside: The longest days at the equator occurs on the equinoxes rather than the solstices, which is when the days are shortest.

Looking at locations outside of the tropics and the polar regions, the longest and shortest days are always on the June and December solstices in the northern hemisphere, and vice versa in the southern hemisphere. Latitude has a significant effect as sunrise and sunset occur at increasing angles with respect to the vertical as absolute latitude increases.

As an extreme, consider Rovaniemi, Finland. This town is located just south of the polar circle. The most recent sunrise was at 1:31 AM on 6 June 2022. The next sunset won't be until 1:04 AM on 7 July 2022. That's nearly a 31 day long "day". A portion of the Sun remains just above the horizon even at local solar midnight during that 31 day long interval. On the other hand, a portion of the Sun will appear above the horizon on the December solstice for over 2 hours and 14 minutes.

Issues

There are multiple issues with this standard 50 arc minute offset. A lesser issue is that the Sun does not always appear to be 32 arc minutes across. The eccentricity of the Earth's orbit makes this apparent size vary.

A much bigger issue is the 34 arc minutes used for calculating atmospheric refraction. The 34 arc minutes used for atmospheric refraction is based on the average environmental lapse rate (how much temperature drops, on average, with increasing altitude) of 6.5 kelvins/kilometer. On a very nice day the lapse rate can be much closer to the dry adiabatic lapse rate of 9.8 kelvins/kilometer as opposed to average environment lapse rate. On a rather nasty day with a temperature inversion (lots of smog, fog, low clouds, etc.), the lapse rate can be close to zero, or even negative. The variation in lapse rate can make calculated sunrise and sunset times off by five minutes.

That said, there is no way to calculate sunrise/sunset times for anywhere on the Earth and for years into the past / into the future without using a canned value for atmospheric refraction (and for the Sun's apparent radius). So that's what those websites do.

Finally, if there's a mountain or tall building off to your east (or west), the sunrise and sunset time calculations do not account for that mountain or tall building. The websites that calculate sunrise / sunset times assume the Earth is a triaxial ellipsoid with no obstructions.

Glorfindel
  • 4,790
  • 3
  • 25
  • 40
David Hammen
  • 33,900
  • 3
  • 74
  • 125
  • 3
    To add a point to this great answer: it might be explained how the (34m + 16m)*2 of arc (=1.66 degrees) , account for such big a discrepancy of 34 minutes of time. For naively one can assume 15deg equates to 1 hour, ergo 1.66 deg only to several minutes. But because when we are not in the equator and especially far as 56 deg lat; the declination circles are not perpendicular to the horizon, hence this leap of 50m (each side) of altitude, equated to new declination and hence which might be more dramatic. In the equator this 50m effect should be only few minutes. – d_e Jun 12 '22 at 18:50
  • 2
    I just checked the OP's sunrise / noon / sunset times with Horizons. It looks like the app is using an elevation of zero. But Aberdeen is hilly & an elevation of 100 m subtracts ~3 minutes from the sunrise time (and adds it to the sunset time). – PM 2Ring Jun 12 '22 at 22:24
  • 2
    @PM2Ring As I mentioned in my answer, most apps that calculate sunrise / sunset times assume a spherical Earth with no terrain variations. – David Hammen Jun 12 '22 at 22:42
  • 1
    @d_e I added info about how latitude affects the disparity. I intentionally picked a town just south of the Arctic Circle to demonstrate the sensitivity to latitude. Thanks for the suggestion. – David Hammen Jun 12 '22 at 23:50
  • "there is no way to calculate sunrise/sunset times for anywhere on the Earth and for years into the past / into the future without using a canned value for atmospheric refraction (and for the Sun's apparent radius)" -- agreed on the atmospheric effect, but it is wrong to say that the Sun's apparent radius cannot be computed specifically for each date. Most calculators do not do so, but that doesn't mean there is "no way". The radius part should have been left out of that sentence. – nanoman Jun 13 '22 at 04:17
  • @nanoman The Sun's angular radius is the lesser of the two key factors. Assuming a constant value for the Sun's angular radius induces at most a 0.3 arc minute error. Assuming a constant value for atmospheric refraction induces a much greater error. – David Hammen Jun 13 '22 at 06:29
  • @DavidHammen Agreed -- I'm just saying the sentence wrongly implies that it's impossible to do any better than a "canned value" for the radius. I would split it into two points: impossible for refraction, and possible but not worth doing for radius. – nanoman Jun 13 '22 at 06:32
  • Thanks a lot for this! It's much clearer now! @ d_e that's really cool! I was about to do the calculation before reading your addition to this post so thanks for pointing it out! So does that mean that mean the declination circle around 57°N is closer to 80°? @ PM2Ring yes, although to be fair, it's also by the sea so there is sea level elevation, too. Also thanks for this link, I've been looking for an interactive height map for ages! – Nierninwa Jun 13 '22 at 17:12
  • @Nierninwa I only discovered it myself yesterday. Good elevation data is annoyingly hard to find, although there are a few apps floating around. Re: declination circles, they're centred on the celestial pole, which has an altitude equal to the observer's latitude (with a slight movement of a few arcsecs due to nutation). Star trails show the declination circles. – PM 2Ring Jun 14 '22 at 00:26
  • The Earth is also closer to the Sun during northern hemisphere winter, making the Sun look larger and thus the angular diameter between centre and top of the disk larger. However, I haven't done the maths how many seconds (or maybe minutes) of extra difference this would add. – gerrit Jun 15 '22 at 07:54
  • @gerrit See the discussion with nanoman from a couple of days ago. – David Hammen Jun 15 '22 at 08:38
  • The section about refraction gives the impression that refraction is mainly caused by temperature gradients. Surely temperature gradients affect the magnitude of the refraction, but you would still have refraction without temperature gradients due to the pressure gradient. – jkej Jun 15 '22 at 12:16
  • 1
    @DavidHammen one meta point. Your sentence "The disparity is primarily a consequence of how sunrise and sunset times are defined and calculated." may to some readers give the sense: "it's not a 'real effect', it's just a psychological trick, and the disparity seen in the apps is due to a calculation consequence". However (as you know) that is NOT the case. The effect is real, blatant and substantial if you live in Aberdeen. Shortest daylight (as any human would think of it colloquially in daily life) is much longer than shortest darkness. – Fattie Jun 15 '22 at 13:06