The most known explanation for the Shadow Bands, which is visible right before and after a total solar eclipse, is that some air particles refract the light of the thin sun-crescent, preventing them from reaching the ground. And it's for the same reason which cause stars to twinkle.

And if the following:

  1. Good seeing is unfavourable, since no turbulence cells develop. ("bad seeing is good for shadow bands"). Therefore observation places at sea level are more favourable than higher locations.
  2. Moderate wind velocities in middle atmospheric heights let the shadow bands become well visible. Very strong wind results in fast movements so that the eye cannot follow any longer. With zero wind speed the shadows are nearly without motion and therefore hardly remarkable.

...(from https://www.strickling.net/shadowbands.htm) is proven, it most probably means that this explanation is the right one.

For the star-twinkling part, I understand it perfectly, because there's just one point of light out there which can easily be obscured. But for a crescent, I don't understand it. Because no matter how thin the crescent is in the East-West direction, it is still composed of many points of light, which extends in the North-South direction, which shouldn't be so easily obscured.

Especially I don't understand why they should be bands/lines/snakes, I know that the shape of the sun at that moment is similar to a line, but the shape of the obscuration shouldn't be related to that, but to the orientation of the air particles which causes it. And it's hard to believe that they are oriented in bands (but if they are, it may solve the first part of my question too).

Edit: I perfectly understand the shadow bands visible on the ground of a water body (like a swimming pool): The water waves are big enough that the entire sun is relatively small enough to produce those bands. But if the sun would be like an oval where the long side is ten times bigger than the sun is in reality (5 degrees), and the longer dimension would be perpendicular to the orientation of the waves, it would probably not produce shadow bands, no matter how small it is in the other dimension.

And if the water wouldn't have band-like waves, but zero-dimensional bumps, it wouldn't produce shadow bands at all, just shadow spots.

So does the atmospheric turbulence have to match these criteria in order to produce shadow bands? If not, how do they still work despite that?

  • $\begingroup$ "some air particles refract the light of the thin sun-crescent, preventing them from reaching the ground." Do you have a reference for that claim? Why would refraction prevent light rays from reaching the ground? Refraction deflects light, it doesn't block it. $\endgroup$
    – PM 2Ring
    Apr 13 at 0:34
  • $\begingroup$ By the time of writing I wasn't sure where the refracted light goes. But in any case it's true that it prevents light from reaching the ground at some locations (the shadow bands). $\endgroup$
    – George Lee
    Apr 14 at 14:46
  • $\begingroup$ Oh, I forgot to write @PM2Ring $\endgroup$
    – George Lee
    Apr 14 at 20:12
  • $\begingroup$ @George All the light reaches the ground, but the distribution of the light is uneven because of the refraction. Say we have two adjacent patches of ground, A and B. Without refraction, the light intensity at A equals the intensity at B. With refraction, some of the rays that would've hit A instead hit B, so the intensity at A is reduced and the intensity at B is increased. $\endgroup$
    – PM 2Ring
    Apr 15 at 3:37

2 Answers 2


Think about the patterns that you have seen on the bottom of a swimming pool when the sun shines through ripples on the surface. These are a related phenomenon. When the sun shines through a medium that is "uneven" it is refracted and a pattern of brighter and darker shapes appears. The particular appearance of particularly bright and dark lines is due to the formation of "caustics". Now the pattern that is created by ripples is much stronger as the refraction at the water's surface is very strong. But the same occurs when sunlight shines through the atmosphere. You won't notice it when the sun in un-obscured, since the size of the sun does have a "smoothing" effect. But it does become visible when the sun is reduced to a sliver that is a fraction of an arcminute thick.

The orientation in bands is due to the fact that the sliver of light is a band, extremely narrow, but quite long (relative to it's width) All shadows are affected by this change.

  • 1
    $\begingroup$ "typo" is a kind word... $\endgroup$
    – James K
    Apr 8 at 4:57
  • $\begingroup$ Hi, I'm the OP (having problems with my 'New Editor' account). Thanks for explaining the shadow bands phenomenon, but I'm still unsure about the answer on my specific question. I guess that the atmosphere also have a wavy surface extending mainly in one dimension like the surface of water (not zero dimensional bumps like I thought before). And I guess that the shadow bands are only happening when the orientation of the sliver of the sun aligns with the orientation of the atmospheric surface waves, and not when they're perpendicular to each other. Are these statements correct? $\endgroup$
    – George Lee
    Apr 10 at 20:14

How can atmospheric turbulence obscure the sun-crescent immediately before a total solar eclipse? and why are they 'bands'?

tl;dr: Refraction by atmospheric turbulence redirects the almost-parallel light just before eclipse making it more concentrated in some areas and less in others rather than by "obscuring" it. Stars twinkle for the same reason. If a twinkling star was really bright, it would create similarly visible patterns.

JamesK's answer is correct but lacks supporting sources.

We cause these wavy alternating lighter and darker patterns that aren't really band-like caustics. Random refraction from inhomogeneities in the Earth's atmosphere happens all the time, but only when the light becomes collimated (almost parallel) can we see it. Random deflection of light causes more light in some places and less light in others. There's not really any loss of light or blockage or obscuring, it's just uneven distribution.

The image below shows a plate of normal, transparent glass with a "wiggly" uneven surface generate a portrait of Alan Turing using caustics. For every dark area, there's a nearby light area where the light has been re-directed, not blocked.

From Wikipedia's Shadow bands

Shadow bands are thin, wavy lines of alternating light and dark that can be seen moving and undulating in parallel on plain-coloured surfaces immediately before and after a total solar eclipse. They are caused by the refraction by Earth's atmospheric turbulence of the solar crescent as it thins to a narrow slit, which increasingly collimates the light reaching Earth in the minute just before and after totality.

The shadows' detailed structure is due to random patterns of fine air turbulence that refract the collimated sunlight arriving from the narrow eclipse crescent.

The bands' rapid sliding motion is due to shifting air currents combined with the angular motion of the Sun projecting through higher altitudes. The degree of collimation in the light gradually increases as the crescent thins, until the solar disk is completely covered and the eclipse is total.

Stars twinkle for the same reason. They are so far from Earth that they appear as point sources of light easily disturbed by Earth's atmospheric turbulence which acts like lenses and prisms diverting the light's path. Viewed toward the collimated light of a star, the shadows bands from atmospheric refraction pass over the eye.

Also, from Physics SE's What is this wavy light coming through my blinds?

enter image description here

and from my answer therein

The video Taming light reflection to create images discusses engineered caustics (refractive rather than reflective in this case). See also this post from the EPFL Geometric Computing Laboratory:

from the EPFL Geometric Computing Laboratory


From James Gurney's blogpost Caustic Reflections

From James Gurney's blogpost Caustic Reflections http://gurneyjourney.blogspot.com/2010/07/caustic-reflections.html From James Gurney's blogpost Caustic Reflections http://gurneyjourney.blogspot.com/2010/07/caustic-reflections.html

From forums.sketchup.com's Water Reflection on Surfaces:

from forums.sketchup.com's Water Reflection on Surfaces https://forums.sketchup.com/t/water-reflection-on-surfaces/79291

  • 1
    $\begingroup$ @GeorgeLee it's more that the density of the air fluctuates, rather than the "surface*. This causes the refraction direction of different parts of the air to differ. There's no alignment required, just a nearly parallel (collimated) beam of light. We get that from stars (point sources) and the slivers and dots of light around the edge of the moon moments before totality. $\endgroup$
    – uhoh
    Apr 10 at 20:38
  • 2
    $\begingroup$ @GeorgeLee No it does not mean that. The inhomogeneities will not have directionality. Visibility comes from the collimation of the light, and that happens preferentially in one direction. Just like the GIF in my answer - the shades are slits in one direction and long in the other, so the isotropic puddle waves, viewed through the narrow shade slits, show up as "band-like" patterns. Better start by showing what the eclipse "bands" actually look like, they may not be as distinctly "band-like" as you imagine. $\endgroup$
    – uhoh
    Apr 11 at 23:33
  • 1
    $\begingroup$ Here's the classic shadow bands lithograph by Müller from the 1870 eclipse strickling.net/shadowbands1870large.png from Dr Wolfgang Strickling's page strickling.net/shadowbands.htm That site has shadow band photos & videos, but they're all very vague. $\endgroup$
    – PM 2Ring
    Apr 15 at 3:52
  • 1
    $\begingroup$ @PM2Ring The drawing from 1870 is very exaggerated, and we can't really learn from it. Anyway, my question was not if the band are parallel, but rather why. $\endgroup$
    – George Lee
    Apr 15 at 23:25
  • 1
    $\begingroup$ @GeorgeLee OK great! I was about to add a bounty to your question (I love adding bounties!) but I see that you've done it already. If you still don't get a satisfactory answer, I'll add another one once yours expires. In the mean time, why don't you edit your question a bit to emphasize your why parallel? aspect. Your question's title certainly doesn't make that very clear at the moment. You had a 2nd chance to emphasize that in the bounty message but you left that blank as well. I have a hunch the bands will be parallel to the slit-like crescent but it's just a guess, not an answer. $\endgroup$
    – uhoh
    Apr 15 at 23:32

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .