A question that popped into my head: if I see a picture of the sun close to the horizon, in an unknown place, can I know if it was taken at sunset or sunrise?
Do sunrises and sunsets look the same in a still image? Can one tell them apart?
Asked
Active
Viewed 8,540 times
65
-
3Previously: https://photo.stackexchange.com/questions/26776/can-one-tell-just-by-looking-at-a-picture-if-its-a-sunset-or-a-sunrise – user3067860 Nov 09 '21 at 21:39
3 Answers
40
If you have a sufficiently advanced camera, then you can distinguish a sunrise from a sunset from a still frame. I will assume that the Sun and the horizon are visible.
The Sun is rotating at roughly 2 km/s at the equator.
This rotation imparts Doppler shifts in the light from the Sun, even in a still frame. So with a sufficiently advanced camera which can detect those Doppler shifts (for example, an IFU).
Then you can measure the rotation axis of the Sun. Now, depending on where on Earth you are (latitude and whether this is sunrise or sunset) the rotation axis of the Sun will appear at a different angle relative to the horizon.
So, not only could you theoretically you figure out whether it is sunrise or sunset, but you could also measure the rotation speed of the Sun and your present latitude.
The Sun and the Earth rotate in the same direction (counterclockwise when viewed from above the North Pole). Thus, the top of the Sun relative to the horizon will rotate in the same direction as the observer.
I will use this fact to describe what you would see due to the Sun's rotation:
During sunset, you must be on the side of the Earth moving away from the Sun. From this angle, the top of the Sun (relative to the horizon, not necessarily North!) will appear blueshifted relative to the bottom of the Sun, since the apparent top of the Sun has a velocity towards Earth (same direction as your side of the Earth).
During sunrise, you must be on the side of the Earth moving towards the Sun. From this angle, the top of the Sun will appear redshifted instead.
Side note: I think in a statistical sense you could distinguish between sunrises and sunsets based on effects mentioned in other answers/comments: temperature of the air, stillness of the air, presence of particulates. Statistical meaning you could theoretically do better than a 50/50 guess with a single image, and better if you allow multiple images taken at the same place or with other variables controlled for. So although those answers/comments do not provide a sure way to distinguish sunrise from sunset, I think they suffice to show that the two phenomena are different on some level.
Alwin
- 4,955
-
1It's a good answer (+1), but maybe should have stopped after the 5th paragraph, the word 'latitude'. After that it seems flawed as the top and bottom of the sun would both appear to be travelling the same relative to the earth. – John Hunter Nov 08 '21 at 08:45
-
1Are you sure about this, the rotation of the earth would cause equal doppler shifts to all parts of the Sun, true the Sun's rotation matters, but that's a version of the first part of the answer... – John Hunter Nov 08 '21 at 09:00
-
2Rotation of the Sun, not the Earth. The 2nd part of the answer is just to spell out exactly what the effect of the rotation of the Sun would have. When I refer to "top" i don't refer to North, but "top" as viewed relative to the horizon. – Alwin Nov 08 '21 at 09:08
-
"the rotation of the earth would cause equal doppler shifts to all parts of the Sun" I agree with this. My answer might be mis-communicating something. – Alwin Nov 08 '21 at 09:20
-
7Also: don't try this at home, at least not with a normal camera. You'd need the correct filters, e.g. an H-alpha. But it's true, with an H-alpha telescope, you do notice the Doppler effect, and it's not possible to "focus" (by selecting the required wavelength) on prominences on both sides of the sun at the same time. – Eric Duminil Nov 08 '21 at 11:51
-
2So it's a frequency shift of 2/300,000 ~ 0.66/100,000 that can be measured? Awesome :-). – Peter - Reinstate Monica Nov 09 '21 at 14:40
-
Note that the Sun-Earth distance varies over the year, and this variation can defeat this technique. – Edgar Bonet Nov 09 '21 at 14:50
-
1Edgar: no, that variation does not change the gradient of Doppler shifts across the surface of the Sun. You can detect the Sun's spin rotation no matter what Earth's velocity is. – Alwin Nov 09 '21 at 19:21
-
1Fascinating, thanks! I was hoping for something like this. It a bit of a shame it requires special equipment but I enjoyed reading nonetheless. – AlonMln Nov 11 '21 at 17:38
-
1
26
Yes, the temperature of the air that the sunlight goes through to reach our eyes would be different.
At sunset the air would be warm, at sunrise it's colder and that causes the light to refract differently.
This website shows more about it
John Hunter
- 13,700
-
37
-
1@ ProfRob The diagram shows that the sun would be in a different position, it could probably not be seen by eye, but in principle there would be a difference. Also there is more dust and other particles in the air at the end of the day. That would make the sun look redder at sunset than at sunrise. – John Hunter Nov 07 '21 at 14:14
-
3Cool, I didn't think of that, though it does assume that I know where the sun is supposed to be, no? – AlonMln Nov 07 '21 at 15:04
-
@ AlonMln yes, it's an answer about 'in principle', if enough data were available. 'In practice' it would be difficult just from a photo for example, unless we used other clues from the photo, the amount of dew on grass etc... – John Hunter Nov 07 '21 at 15:07
-
2For pictures taken at different times of the year or at different locations, the temperature for the sunrise picture could have been higher than that for the sunset picture, – Thomas Nov 07 '21 at 17:05
-
Yes, you would need data about the average air temperature at sunrise and sunset for the time of year, and the usual apparent position of the Sun – John Hunter Nov 07 '21 at 17:10
-
7@JohnHunter, Re, "more...particles," The average redness might help you to distinguish a large collection of sunrise photos from a large collection of sunset photos; but the amount of dust and whatever in the air depends on more than just the time of day. Compare photos of sunrises taken in the year after a major volcanic eruption to photos of sunsets taken the year before the eruption, and the sunrises probably will be more red. – Solomon Slow Nov 07 '21 at 17:11
-
16
-
12@JohnHunter: "if enough data were available": this data must obvously include the time of day, otherwise you wouldn't know where the sun is supposed to be. But then if you know the time of day... – TonyK Nov 07 '21 at 22:53
-
2Re the air being colder at sunrise, that's not necessarily so. Suppose you have a warm front move into the area during the night? Which does happen: hereabouts it's not surprising to find that it's warmed up overnight, so what you thought was going to be a nice snowfall when you went to bed turns out to be rain in the morning. – jamesqf Nov 08 '21 at 02:46
-
3It's interesting, but I don't think it's relevant. It only shifts the exact time at which sunrise and sunset happen. Without a clock and without knowing where the sun would be without refraction, it tells you nothing. Sunset and sunrise still happen when the apparent solar disk begins to touch the horizon. This is true with or without refraction, with cold or hot air. – Eric Duminil Nov 08 '21 at 11:41
-
2From my personal experience, sunrises pretty much universally look "straightforward", pretty much regular sphere of the Sun rising above the horizon. Sunsets, with warmed air distort the sphere a lot, resulting in shapes that often hardly resemble the Sun, sometimes splayed like a broken egg, sometimes "trimmed" a bit over the horizon, sometimes even cut into multiple horizontal stripes. Though normal "sphere vanishing behind the horizon" happens too - although it tends to tint the world look more orange than sunrise. – SF. Nov 08 '21 at 13:16
-
1You're focussing on the look/location of the Sun, but the rest of the sky, landscape, and more generally weather may contain clues about the temperature, too. This is a good start of an answer, but would be more useful if expanded to all the visible things that may look differently. – gerrit Nov 08 '21 at 13:39
-
3@jamesqf Air is not always colder at sunrise, but it usually is. Given 1000 photos of sunrise and sunset, it's probably possible to classify them with more than 50% accuracy, but 100% will not be feasible. – gerrit Nov 08 '21 at 13:42
-
1@gerrit: Sure, there are factors that will let you do better than random guessing, but I think the OP wants near-certainty. I don't think that's possible, in general. – jamesqf Nov 08 '21 at 16:08
-
1To which question are you answering "yes" ? It's a problem because the question is repeated 4 times, twice in one way, twice in the opposite – Caius Jard Nov 08 '21 at 16:31
-
3
11
I did not perform any calculations, but if the exposure is long enough, one can see a tiny fragment of sun trajectory. In the northern hemisphere the sun moves from left to right, so the fragments of trajectory will look different at sunrise (left bottom to right top) and sunset (left top to right bottom).
EDIT (11/7/2021): Let me add some calculations. The angular size of the Sun is 0.5 degrees. The visible angular velocity of the Sun is 360 degrees per 24 hours, so the Sun moves by 1/10 of its angular size in 12 seconds. So the direction of the Sun's trajectory can be visible in photos at comparable exposure.
akhmeteli
- 26,888
- 2
- 27
- 65
-
7While this is technically true it does feel a bit like cheating. The question asked for a still image and you are argueing that a real life photograph is not actually a still life but rather an exposition of a small time segment. – quarague Nov 08 '21 at 07:27
-
19
-
9
-
3
-
1The question specifies "in an unknown place", so you can't assume the hemisphere is known. – TonyK Nov 08 '21 at 17:56
-
1@TonyK : this is a valid objection, but I would say it is rather technical as 90% of the population of the Earth live in the northern hemisphere – akhmeteli Nov 08 '21 at 18:13
-
1That's a ricidulous objection in itself! But the relevant statistic is how many people live north of the Tropic of Cancer, isn't it? That is much less than 90%. – TonyK Nov 08 '21 at 18:17
-
@TonyK : no, the number of people living north to the equator is about 90% – akhmeteli Nov 08 '21 at 19:12
-
I assume you are replying to my deleted comment that claimed a figure of 87% (for the equator, not the Tropic of Cancer). This comes from Wikipedia, which takes it from the World Population Yearbook for 2019. Perhaps your source is this factoid, which gets it directly from this BigThink article; the BigThink article, however, says "almost 90%". Which could mean anything. – TonyK Nov 08 '21 at 19:37
-
6But really, the very idea that 10% of the world's population simply don't count is, frankly, offensive. – TonyK Nov 08 '21 at 19:38
-
3Assume that it's equally probable that any human could have taken the photo, from this it follows that it's probable that it was taken in the northern hemisphere. Factor that in to the answer with bounds on the certainty. Next, try to be less easily offended . – Lamar Latrell Nov 08 '21 at 20:08
-
1
-
4@akhmeteli please ensure your time machine complies with ISO-8601 then :p – Lodinn Nov 09 '21 at 16:44
-
@Lodinn : You see, this site provides dates to me in the US format (these are probably local settings), so until they change that, the time machine won't change the format either:-) I believe one can tell the US format from the European one as the former uses slashes. – akhmeteli Nov 09 '21 at 17:45
-
5@akhmeteli it's not the European format, it's the basically-everywhere-except-the-US format (if you count YMD and DMY as the same). And it's not distinguished by slashes, I live in Canada where both are unfortunately used and have to disambiguate by using the 3-letter month. You are of course perfectly entitled to use that format as an American resident, but it is ambiguous to an international audience – llama Nov 09 '21 at 19:12
-
5@LamarLatrell If you simply want odds: It's sunset. Because a lot more people are awake and out-of-doors taking photos around sunset than are around sunrise. – user3067860 Nov 09 '21 at 21:38
-
Please try to take a sunrise or sunset picture, at any aperture, with an exposure time of 12seconds, and please try to guess anything about this pure white image. – Eric Duminil Nov 12 '21 at 10:06
-
@EricDuminil : I don't know much about photography, but if everything else fails, why not use a filter? https://digital-photography-school.com/how-to-use-a-10-stop-nd-filter-to-take-long-exposure-sunset-images/ – akhmeteli Nov 12 '21 at 13:13
-
@akhmeteli: It's a step in the right direction, but this will fail too. Note that in the linked pictures, you don't see the sun as a sharp disk, but a blob of pure white, with undefined shape, which is much larger than the solar disk. You cannot extract any information from the motion blur of this blob, because the motion blur isn't visible. It's all just a bunch of 0xFFFFFF pixels. – Eric Duminil Nov 12 '21 at 16:52
-
-
@akhmeteli Because a filter doesn't change the fact that the sun will be many orders of magnitude brighter than everything else in the picture. If you want to show the landscape, the sun and its surrounding will be pure white. If you want to expose for the sun, you'll need many many filters, and everything else will be pure black. It won't be much of a sunrise or sunset picture. The only filter which could help, would be one which is cut specially for the sun and has its exact shape and size. I've never heard of such a filter. Some filters have a gradient, and filters only the upper part. – Eric Duminil Nov 12 '21 at 17:13
-
@EricDuminil : This does not sound very convincing. We don't need a detailed image of the sun, we just need to be able to find its approximate contour (deviation from the round shape). – akhmeteli Nov 12 '21 at 17:29
-
"This does not sound very convincing." I LOLed. The burden of proof lies on your side. You have apparently not much clue about photography, and simply handwave my arguments. Please try what you proposed if you're so sure it works, and show us an example picture! Everywhere in the Tropics, there's at least one day during which the sun rises from bottom right to top left. Look at https://www.datagraver.com/case/world-population-distribution-by-latitude-and-longitude-2020 to see for how many billion people your method would fail. It's not a problem though, since it fails everywhere. – Eric Duminil Nov 12 '21 at 18:03
-
@EricDuminil : I admitted that I don't know much about photography, but I know enough about physics, so when you tell me that no filter can resolve this issue, I just don't believe you, sorry. – akhmeteli Nov 12 '21 at 20:42
-
@EricDuminil : "Because a filter doesn't change the fact that the sun will be many orders of magnitude brighter than everything else in the picture." So what? A longer exposure makes brighter both the image of the sun and the images of other (stationary) objects. Let us consider some run-of-the mill sunset photo made with a certain exposure. If we use instead a 100 times longer exposure AND a filter transmitting only 1/100th of the light, the contrast between the images of the sun and the other objects should remain pretty much the same. Or do the laws of photography trump the laws of physics? – akhmeteli Nov 13 '21 at 07:29
-
@akhmeteli the contrast indeed stays the same, and that's the problem. The contrast is too high for the dynamic range of the camera, and you cannot get detail in both the landscape and the solar disk and corona at the same time. And I don't think you'd detect a small amount of motion blur in this white blob. Atmospheric refraction and lens distortion will also change the shape of his blob. Stop arguing and start shooting. I'd be happy to be proven wrong. – Eric Duminil Nov 13 '21 at 09:44
-
@EricDuminil : Thank you for agreeing that the contrast stays the same with longer exposure and lesser filter transmission. But then it turns out that your argument is too strong: if the dynamic range is not sufficient for long exposure and dark filter, that means the dynamic range is not sufficient for the short exposure and no filter, either. So one cannot properly photograph sunset in principle, according to you. As for refraction and lens distortion, they probably would not hide the asymmetry between the left and right sides of the sun image containing the sunrise vs. sunset information. – akhmeteli Nov 13 '21 at 16:03
-
@EricDuminil : "Stop arguing and start shooting." Look, you raised an interesting and reasonable (on the face of it) objection to the approach of my answer. I carefully considered your objection and explained why it does not kill the approach. I make maybe five (lousy) photographs per year with my phone camera, so, with all due respect, I am not going to buy a tripod, a cable release, a (graduated neutral density?) filter, and what not to convince you. – akhmeteli Nov 13 '21 at 16:12
-
1/2. "that means the dynamic range is not sufficient for the short exposure and no filter". Indeed. Please look at sunset and sunrise pictures. Can you see details in both landscape and sun+halo? You typically can't, and the sun is just full of blownout highlights. It looks realistic, because our eyes don't see any detail in this region anyway, and a pure white blob is what we expect the sun to look like. It erases any potential motion blur, though. If the sun is not completely white on the picture, it might be due to having enough clouds in front of it. – Eric Duminil Nov 13 '21 at 16:20
-
2/2 But then, atmospheric distortion would be so strong that you couldn't trust the shape anyway. (e.g. https://en.wikipedia.org/wiki/Mirage_of_astronomical_objects#/media/File:Mock_mirage_of_the_setting_sun.jpg). Lens distortion can also be pretty strong. (e.g. https://www.flickr.com/photos/kds8404/31888894577/) Please zoom in on the sun region. Do you see any detail there? It doesn't even look like a disk, and the pure white area is larger than 4 or 5 solar disks. – Eric Duminil Nov 13 '21 at 16:24
-
@EricDuminil : I don't buy it. https://snapshot.canon-asia.com/article/en/sunrise-sunset-achieving-dramatic-contrasts-in-street-photography (Scene 2 - sun and bridge) seems fine to me. Good cameras have decent dynamic range. – akhmeteli Nov 13 '21 at 17:43
-
"I don't buy it". Yeah, don't let facts contradict your opinion. I listed 4 reasons why your idea won't work. I also mentioned the possiblity of having clouds in front of the sun in order to dim its apparent brightness. That's how Galileo observed the sun and discovered sunspots. But the atmospheric distortion can be so strong in this case that you simply cannot trust the shape of the sun disk. You don't want to try the experiment? Fine. You could try to find a sunset or sunrise picture with motion blur, then. I've found many pictures where it wouldn't work, none where it could. Goodbye! – Eric Duminil Nov 13 '21 at 17:55