APOD: A Triple Sunrise Over Gdansk Bay (2009 Aug 04)

[green1]

In the next image, I illustrate the the parabola that would be required to bring any reflected images to a single image. (This is assuming that the balcony railing is perpendicular to the window, which I think is reasonable.) The sun would have to move to the point on the right of the photo to be at a right angle to the window. (It would also have to stay at the horizon.)
While looking at this I caught what several posters have mentioned. At the angle of attack to the window, any reflected images would be to the right of the original, not to the left. Add to this the focal length of the camera lens would not even focus on any image reflected on the window. As several proponents of the reflected window image have called attention to blurry reflections that are probably drapery on the inside of the room; these reflections are not sharp because they are too close to the camera.

angle1.jpg (84.47 KiB) Viewed 2941 times

[HellCat]

As an APoD devotee, astronomy buff, and closet scientist, I admit to being out of my league discussing the optical components of these pictures.

But as a pilot, and respectful watcher of the weather, we may all agree that there are still meteorological phenomena out there that are not completely understood or appreciated. A friend in Colorado had his neighborhood hit by what sounds like a horizontal tornado a few weeks ago.

So, assuming the panes of glass were not warped, the camera normal, photographs untouched, and other observers relatively sane, I'd like to cast a vote for atmospheric disturbance.

Lenticular cloud: http://antwrp.gsfc.nasa.gov/apod/ap090203.html
"new" cloud: http://news.nationalgeographic.com/news ... -pictures/

[jude5001]

note how each image of the sun is higher than the image to the right of it.
note the clouds, there appears to be two distinct layers. my guess is atmospheric lensing similar to a rainbow (ie. double rainbow).
interesting pic

[Chris Peterson]

But as a pilot, and respectful watcher of the weather, we may all agree that there are still meteorological phenomena out there that are not completely understood or appreciated.

Certainly. But as a meteorological phenomenon, this would be nearly unbelievable. Certainly, nobody has remotely suggested anything even close to plausible; there is simply nothing that could even theoretically cause this effect. And there are clear indications that this isn't atmospheric, such as the reflection of the horizon mirage from the primary disc in the second, and the matching cloud patterns between all three discs. And, of course, there is the posting of very similar images that are known to have been caused by internal reflections in glass windows.

[Elias Chasiotis]

From the glossary of the book "Color and light in nature" by D. Lynch and W. Livingston (Cambridge University Press, second edition, 2001), p.266:
"multiple suns (moons) unexplained occurrences of multiple images of the sun".

I would like to know what was in this book. Are they talking about multiple images above, below, left or right?

i own the book, and chasiotis has quoted the entire glossary entry in this book for the phrase "multiple suns/moons". there is only one entry for "multiple suns/moons" in the index, and it is to page 266, in other words, the glossary entry.

there is no photograph or treatment of "multiple suns/moons" anywhere in the book. there is an extensive treatment of mirages, all of them vertical (not horizontal) image displacements.

the cite is a red herring.

On page 3 of the discussion i have a reference on the book "Light and color in the outdoors" by Marcel Minnaert, who describes an observed and photographed case of multiple suns beside and above the sun. Also cases of observation of multiple crescents of the moon are reported there, including two sketches.

[grump]

A few people have mentioned that the multiple images cannot be caused by reflections in a window because they are on the wrong side. They assume that the window is roughly parallel to the buildings seen mid foreground in the wide angle photo, and do not consider the possibility that the window might be roughly perpendicular to the aforementioned buildings, as might be found in a corner or bay window. This might explain the way the drapes or whatever appear reflected in the glass.

It is clear to me that the photo was taken behind a glass window, so its just a matter of working out how the reflections were formed.

[almightyon]

[...]
I again point out that the cloud shadow in front of the reflections is different than that of the primary. In fact if a curved line is drawn at the top of the shadows (think "sine wave") it shows a continuous contour consistent with a single bank of clouds, showing that the reflected light is behind the clouds rather than in the windowpane.

triplesunrise_diacon lines.jpg

The clouds are no problem for the reflections theory. I've analized the color of the 1st sun, and the very most of it has passed the camera's threshold for brightness: that's why you can't see in the 1st sun the cloud that appears in the second. And you can't see clouds bleeding with luminosity out of the 2nd sun because it's not bright enough. The suns AND clouds should blend on the reflections, and the clouds of the 1st one should be cloned - but the brightness difference is so big that it just disappears. So far, the glass on the balcony theory wins easily - it was shown that there seems to be in fact a glass, it matches the angles, etc.

And for who says they're on the wrong side - they're not. The window should be oriented approaching from the right to the photographer to show the reflections that are shown - and it seems right that the window is oriented this was looking at the balcony pictures.

[Bill Melater]

How about that other thing in the sky that's just about the same size as the Sun?

As we saw from the photos of the eclipse 12 days later, the apparent diameter of the Moon is a bit smaller than that of the Sun this time of year.

Checking the July data for Gdansk, the Moon was three days past full (94% illuminated) and three hours short of setting at the time our puzzling photographs were taken.

Could we be seeing an atmospheric reflection of the Moon rather than a refraction of the Sun? Unlikely (I really don't believe it), but then what we are dealing here does seem to fit the category of "unlikely."

As for Mr. backstagediver's calculations of distance, he is neglecting the inverse-square relationship of apparent size to distance, plus the fact that the light passes between the panes at an angle. Using my own measurements of relative sizes, the window panes would need be only 955 thousand miles apart, a far more realistic distance.

[Ageingted]

Multiple images can occur if a camera has a series of a lens and filters attached to it. The normal lens being used, a polarising filter and possibly a protective clear glass lens. Couple this with the possibility of taking the photograph through a window and you have multiple reflections. I have done this several times before and it took me quite a while to work out how it happened.

[Alnilam]

To drollere :

Why guess about the dynamic of the camera and the reflection ratio of a window ? Just mesure it ! With PhotoShop, I mesured the brightness of several points in the picture : three for each sun image as illustrated on the picture above. Black point in the limbe, blue point in the top cloud, green point in the left cloud, the cross to center the pattern on each image. There are my results :

1black : x=1064, y=741, R=255, G=255, B=255
1blue : x=1064, y=725, R=255, G=255, B=180
1green : x=1024, y=761, R=255, G=255, B=146
2black : x=964, y=735, 252, 255, 158
2blue : same relative position as 1blue, 255, 126, 64
2green : same relative position as 1green, 242, 79, 46
3black: x=897, y=727, 255, 107, 49
3blue : same relative position as 1blue, 248, 98, 39
3green : same relative position as 1green, 209,72,40

And my conclusions :
- the depth of this picture is obviously 24 bits (1 byte for each color, 0-255)
- multiple parts of the picture are over-exposed (255) and only non-over-exposed parts could be usefull.
- based only on blue color of blue&green points, the rate between image2 and 1 is about 1/3 (36% and 32%)

So this measurments are consistant with the theory of a window reflections, all optical conditions being equal between the firts and the second (transmission and reflection index). Until then my opinion was not this theory but multiple clues converge to it.

[mfatihokhider]

Hi !
Due to the inherent " slow" light velocity towards our eyes, which will takes light 492 seconds, for us to detect it, the sun image will certainly be distorted by variety of causes. These might encludes anything that might affects the gravity waves between us and the sun: during this breif light trajectory.

[Alnilam]

To luchio :

My opinion is that your picture (I gained the same) demonstrate that there is obviously reflections in the picture and it's a great probe for the window theory. But I think that the "gohst image" is not a reflection of the balcony because the distance between them isn't consistant with the distances between multiple sun images. I think that the "gohst image" is a reflection of things placed into the room, brightly lightened by the rising sun and reflecting in a dark part of the window (because of the wall behind it). I can see something looking like the base of a dark tripod.

To Green1 :

We can make no hypothesis of the angle of the window. Why should it be parallel to the balcony ? It could be if it's a sliding model but if not ? With hinges, the position of the window could be almost every angle !

[Alnilam]

To owlice :

I confirm that there is realy a reflection of the "second sun" on the sea surface. Look at the above picture (I ajusted the contrast +60%).

[Tod]

I am reminded of what happens to images made of specific colors when I look away from straight ahead through my rather thick (in the order of 13 diopters) glasses. If a round object were composed of many red, green, and blue LEDs, for example, it would look white to me as I looked at it straight on. As I turn my head and look through my thick glasses which form an ever more steep prism shape the red, green, and blue disks would separate as each of them is being thrown at a slightly different angle since the refractive transition is at an angle.

Perhaps there is something in the atmosphere particularly conductive at the colors represented by the two suns - one fairly yellow and the other deep orange. If there were a sharp front between some say very dense cold air and very dry warm air whose line(s) are at an angle to the viewer perhaps they would separate those colors as my glasses do.

I suppose they would mix back together as it the sun rose above the strange surface activity.

Just a thought,

Tod

[Alnilam]

In French we says "un dessin vaut mieux qu'un long discours" (about : "a simple drawing is better than long tellings"), so I tried to verify if the theory of multiple same images of the sun could be right, on the picture itself. I take a copy of the "second sun" (because not over-exposed neither too faint) and put it respectively over the "first sun" and "third sun", to see if the clouds pattern could match. I did that with PhotoShop. I take an elliptical part of the picture because I used the "hourglass" pattern (or "Etruscan vase" as someone writes) to have a better fit with overexposed "sun 1". I reduced light of the copy to match the fainter "sun3".

I think that the result is clear : clouds pattern match for each case (2/1 and 2/3) as we can see on the left edge of each sun image and right edge of "sun 1". Until then, my opinion was not the window theory but I have to change because of multiple clues for it (faint rate between reflections as suggested by drollere, ghost image as demonstrated by luchio, and finaly pefect clouds pattern matching). A little explanation of my wrong theory (atmospherical unusual effect) : by a special chance, the clouds pattern is very compatible between each reflection and the background landscape so it produces the illusion of continuous layers of clouds.

A last question : Are Robert and Jerry not supposed to have some data about the conditions of a picture (presence of a window by example ) before publishing it in the very famous APOD ? (I send them few pictures with no result... next time, I will try another trick of of the light )

[blastoff]

to GREEN. if the other suns ARE "reflections" bouncing out of the glass, they would indeed still be in focus according the camera, because they are just the same image as the actual sun image, plus an extra few millimeters gained within the window. so thats 93, 000, 000 miles and a little extra. it would still be in focus.
THAT SAID, im still on the all natural bandwagon. it just seems odd to imagine that window coming out next to the balcony -perpendicular to it. and if it was, it seems all objects on the right of the picture would appear blurrier and even warbly, due to the acute angle of this wacky window through which they are seen. even odder than a few sheer "panes" of different atmosphere reflecting the sunlight back and fourth between them on its way to the camera? ...

[walfy]

I just did a very quick crude experiment with my Nikon D700, 70-200 VR lens, using a flashlight as the sun, and taking two pictures of it from outside my house. I have single-paned windows, side by side. One slides open, which doubles up the two panes. So I turned off all the lights, set the flashlight atop the fridge, stepped outside and took the first shot through one pane of glass at about a 45 degree angle. Then the same shot through two panes of glass, same angle. Word of warning: I did this all hand-held with lens vibration reduction turned on, set the D700 at very high ISO (1600), shutter speed 1/30th of a second. So the shots are not as sharp as they'd be if I had used a tripod. Perhaps I shall try again tomorrow night with a tripod. The angle of the windowpanes was roughly 45 degrees from the plane of the camera sensor. This was a very crude, quick experiment.

Here is a QT movie of the photos: http://www.freywine.com/Crude3SunExperiment.mov

The 5 photos, in order of appearance, consist of:

1. Flashlight photographed through single pane of glass, showing dimly the horizontal top of the fridge (our "horizon" in these shots).
2. Flashlight through 2 panes of glass, includes horizontal top of the fridge.
3. Flashlight through single pane of glass, imaged at 100% (original resolution of camera).
4. Flashlight through 2 panes of glass, imaged at 100% (original resolution of camera).
5. In this final image you can see the window through which I photographed the flashlight, which is still turned on and resting atop the pile of nuts on the fridge.

I assumed the angle of my window was similar to the APOD image (perhaps a wrong assumption). The two reflections of the flashlight are to the right of the original. But on the APOD triple sunset photo, the two sun reflections are to the left! It would be helpful if the photographer of the "triple sunrise" submitted more photos of the apartment where the plane of window could be discerned in relation to the sunrise. Better yet, take another sunset photo, and try to duplicate the results!

A few quick observations:

In my image, all three "suns" (the original and both reflections) seem equidistant, unlike the APOD sun image.

But very similar to the APOD image, the luminosity of the two reflections diminishes at a very similar rate.

How likely is it that a large-scale atmospheric event over a body of water would tweak the rays of the sun in a way very similar to what these double-paned windows can do?

[geckzilla]

To the people who think the angle of the window matters, you are missing the point and quite wrong. It's not the relationship of the window itself to the viewer but the two panes' relationship to one another that really matters. I mean, the viewer to window does matter, but not in the way you think it does.

The angles in my picture are definitely not calculated to be an exact illustration. Actually, I'm sure this illustration is wrong in a number of ways, but to be honest I don't even know exactly what I was looking at when I was doing this. I can't tell which pane of glass the reflection is on and I'm admittedly feeling pretty dumb that I can't figure it out. If anyone has an illustration which explains it more accurately, I'd really like to see it.

[green1]

Let’s illustrate various window configurations that might result in ghost images due to refraction.
With the window perpendicular to the lens, there would be images on both sides.

RAYS1.jpg (34.14 KiB) Viewed 2473 times

With the window at the same angle as the balcony rail (most likely) any images would be to the right.

RAYS2.jpg (32.31 KiB) Viewed 2473 times

The window would have to be at right angles to the balcony railing for the images to be to the left of the primary. If this was the case, the reflected curtains would have to be in the camera’s field of view, which they obviously are not.

RAYS3.jpg (31.36 KiB) Viewed 2473 times

In addition to this, we are forgetting that for multiple ghost images you need an aperture which in this case would be at the horizon. I will admit that the curvature of the atmosphere acts as an aperture of sorts. Unfortunately, for the simple lensatic effect the ‘window reflection’ proponents are proposing, the aperture has to be oriented vertically to get horizontal images. The curve of the atmosphere is horizontal, which would result in vertical ghost images.
Now to the point; the images shown are lensatic images from refracted light through varied density air found at this latitude in early summer. You have a cold mass to the north (the arctic sea) which in fair weather cools the air in a stable mass during the night. This air is dense and has a distinct boundary which acts as a lens separate from the primary lens of the earth’s atmosphere.

[geckzilla]

Hi green1, you still missed a vital part of the circumstances, and that's to have two panes of glass which are NOT perfectly parallel to one another. Please see my post above.

[green1]

In the example of reflected light we see the same scenario. For the images to show up on the left of the primary the window must be at right angles to the (apparent) face of the building.

REFL1.jpg (30.82 KiB) Viewed 2463 times

[Hi green1, you still missed a vital part of the circumstances, and that's to have two panes of glass which are NOT perfectly parallel to one another. Please see my post above.] Gekzilla, even with panes that are not perfectly aligned the reflection (or refraction) is going to be the same, with only minute variance. The reflection is always 'downstream'. Try throwing a ball against a wall to illustrate the point. Without a spin, the ball will bounce of the wall at an angle equal to the angle of attack. Light does the same thing. even refracted light acts similarly, the angle of change is equal to the average of the wavelength. As the wavelength changes so does the degree of refraction. (This is how a prism separates the colors in sunlight. Longer wavelengths bend at a sharper angle than the shorter wavelengths)

[geckzilla]

That's not what happens though. Why don't you try it yourself? It's not that hard. You can even use some pretty small pieces of glass. I've got a couple of 3.5"x4.5" picture frame glass panes that I use so it's easy to handle them.

[Naos]

Hello,

@ green1 : why do you use a point source at finite distance to represent the sun in your last drawings ? it seems quite strange to me.
I would have used a set of collimated beams coming from a direction +/- 0.25 degree (= 0.5 degree, apparent angular size of the sun)


@ geckzilla
I have tested with Zemax optical calculation software your idea of "NOT perfectly parallel panes of glass". This kind of configuration CAN produce ghosts on the Left OR on the right depending on the sign of the tilt angle between the two panes of glass. (Note: if using the other axis of rotation, we could deviate the ghosts up and down too !)
Moreover : it does not depend on the global inclination of the window (Plus 45 deg, or Minus 45 deg). I calculated that the tilt angle must be equal to half the apparent angular distance between successive ghosts.

Just a word about non equal angular distance between suns : glass used for windows are not optically perfect, the thickness is not everywhere uniform, thus creating some local slopes on the surface. This can have two effects :
- first one is to blur the image because of converging or diverging shape, so the image of the sun is not well focused anymore.
- second one is to introduce a small additionnal deviation to the direction of the beam, so the exact uniformly spaced sequence of ghosts is broken.
I calculated that a small bump of 0.2mm thick over a surface of 10cm wide can deviate the image of about 0.25 degree
So, the successive positions of the ghosts and their quality of focus can slightly change depending on the local properties of the glass thickness crossed by the beam before entering the camera.

PS : I only wanted to show that a window CAN produce the effect, not that this the ONLY explanation.

[geckzilla]

Naos, can you take a screenshot of the simulation? I'm very curious about this software now. Something I don't understand is why the reflections go towards the narrowing part of the panes rather than the widening side no matter what. I watched it do it but I can't understand it.

[TCS.net]

As photographer I have often experienced reflection of this sort within my lens. As much as I doubt this is an in lens reflection(due to the same reflection at different focal lengths) I know that a window can have much the same effect. Is any more information such as a window or the specific camera body and lens/es available?