So… Rainbows. There’s not much to say: Everyone has seen them, and probably everyone has taken photos of them. And, in my experience, the photos almost never turn out as well as in real life. I’m including myself in the above; these following photos have been touched up more than usual. I played around with the brightness, contrast, and saturation to get the best view of the rainbow.
I keep trying to take a photo of a rainbow in different spectra. I’m usually hampered by the fact that unless I’m on vacation, I might not have my camera with me(I don’t want to get it wet if I can avoid it). Other times, I start taking photos, only to have the rainbow fade too quickly. Sometimes, though, everything lines up perfectly. I have my camera with me, all my filters are at hand, and the rainbow is just hanging there. In this specific case, these photos were taken beside Highway 16 in Jasper National Park, Alberta, just past the British Columbia border.
This rainbow is a Supernumerary Rainbow. If you look closely, the bands repeat themselves inside the arc of the rainbow. This repetition is related to the size of the droplets
Here is the rainbow in the visible spectrum:
This was taken with an unmodified camera, so there is no infrared leakage. Now, to look at the composition of the rainbow: When you break down the channels into Blue, Green and Red(in that order), you get: 
The repeating pattern seems to be stronger in the longer wavelengths- it is very strong in red, and almost invisible in blue. It’s hard to compare them, so I’m including a .GIF file below with these same photos to better compare the bands. These are all channels from the same photo, so they occur at exactly the same time.
Of course, as the sun changes position, the rainbow does as well( for these photos, the sun was descending, so the rainbow was rising). Also, as the atmospheric conditions change, the rainbow quality changes. This was minimal in the following photos, but remember that the rainbow won’t match up perfectly.
First, my short band pass filter, which allows light through in the violet and ultraviolet range, with a small amount of blue light. As mentioned before, the white balance settings cannot be set with this lens, so everything looks purple. This photo was taken about a minute after the visible photo.
There is no real sign of the repeating bows- if you look close on the left side, you can see one faint repeating bow. There are no true ultraviolet-only photos of this rainbow; I tried several times, but didn’t have a way to keep the camera still enough for a good photo(it might also be that the sky was too bright in UV for the rainbow to be visible, but I’m not sure). The exposure time for a UV photo was over 8 seconds.
So, instead, my UG11 filter, taken about 90 seconds after the visible photo. This filter cuts out visible light, allowing ultraviolet and Infrared light through.
The bands are much more visible now. I would assume that this is due to the longer wavelength infrared light and not the ultraviolet light. The trend seems to be that the shorter the wavelength, the less visible the repeating bands are.
If this hypothesis is correct, than the bands should be strong in the infrared range. Using my 680 nm infrared lens(with some red light passing through), three minutes after the first photo, the picture appears like this: 
The bands are much more visible at this wavelength. Four arcs are visible clearly, with hints of a fifth. Excuse the dust on the filter- I was worried removing the dust digitally would affect the rainbow visibility.
Finally, the full spectrum photo, without any filters. This was taken about 4 minutes after the first photo.
It looks similar to the visible spectrum photo, with the major difference being the colour of the trees(due to the infrared light). As for the rainbow itself, the red bands look much stronger- As with the trees this is likely due to the camera interpreting the infrared light as red, and giving the red channel a boost.
In all, looking at these photos as a sequence, I would assume that the supernumerary bands on a rainbow are strongest on the longer wavelength colours. This is just a hypothesis, however, based on one sample. I’m hoping to take more photos of supernumerary rainbows to compare the results.
To finish off, I’m including one more GIF file. This has all the photos in the post, for easier comparison. I reduced the colour in the short pass photo, so it looks more similar to the surrounding photos; otherwise, the pictures are the same as above.
Parallel Views 