Recently I've noticed a peculiar effect while taking twilight flat-fields. I should explain that, using the Lowell 1.1-m telescope, I usually take batches of evening twilight flats with the telescope aimed at the neutral-gradient 'Chromey spot'. I typically need flats in seven filters, so they have to be done in a specific order and taken pretty much without screw-ups --- if you start like a minute late, or there's some goof in the process, you won't get through them all. The order is: CN (3880A), Sloan u, Sloan z, B, V, VR (a wide filter very similar to Sloan r), and Ic. The CN filter is a narrow one used for comet comae/jets, and is taken close to sunset (we sometimes also use other narrow comet-emission filters farther toward the violet that are done even 45 minutes before sunset!). Following CN, one has to work steadily to get 15 images (usually) in each filter. Luckily the camera shutter is good, so I can start flats with exposures of 0.5 seconds. For exposures less than ~0.8 seconds, I let the sky drift across the frame by leaving the telescope tracking off. For longer exposures, I use tracking but dither the telescope by an arcminute or so while the images are reading out.
OK, so I've done this on hundreds of nights, so I kinda have the pattern down, including the estimated exposures progressing along the filter sequence. If things are completely clear to the horizon, there is a bit of a break after finishing the u images before starting the z filter exposures. A couple of times recently I have noticed, however, that if there are obscuring clouds low in the west, blocking the twilight to some extent, the z filter exposures are roughly triple what's expected, and must be started right away following the u images. This caused me to panic initially, because it seemed successive filters would have much longer exposures, too, meaning the final ones would be unacceptably long. But on going from z to B, the exposures are indeed as short as expected. Finally getting to the Ic frames, the exposures were again longer than expected. hmmph...
So if there are clouds in the west, the twilight itself ought to be somewhat dimmer simply from the shadowing overhead caused by the clouds, essentially a sky-wide crepuscular ray. OK, so one can start the flats a bit sooner than otherwise, not a big deal. But why the apparent trend with color? I would expect that it would be the _bluer_ wavelengths that get dimmed more than the long wavelengths just from the Rayleigh scattering (it's why the sky is blue to begin with). So the puzzle is the stronger dimming in the far-red Ic and z (roughly 8900A) filters. There must be some atmospheric-physics thing I'm missing here. Can anyone explain this better?
\Brian
Brian,
I haven't taken or used twilight flats in several decades but scattering can also be Mie and probably other types.too. Perhaps there was some variation in or change in scattering at the wavelengths you noticed. Don't have any ideas on what physics might be involved to cause what you observed but might require some chemistry changes somewhere along the optical path.
Attached is a link to a Penn State Meteo 300 level class notes that might be a start, other pages in the class notes might be applicable too.
https://www.e-education.psu.edu/meteo300/node/785
Best regards,
Jim DeYoung (DEY)
Thanks to Jim DeYoung for his suggestions, which I did ponder. However, there might be a simpler, straightforward answer to this puzzle. I described the mystery to one of our postdocs, who knows optics stuff down at the microscopic level as regards scattering off asteroid surfaces, but who is also a sky-watcher. His suggestion was that there was no scattering or other complicated aerosol optics involved, but just the shadowing of the twilight. Briefly, during ordinary cloud-free twilight, the lower western horizon has the more-or-less reddish sunset colors that one expects. But in the case where there are foreground clouds down low, that red light is blocked from coming across the sky to where the telescope is pointed. So, sure enough, the sky is darker toward the red end of the spectrum compared to the no-clouds case. To first order, this sounds right. QED.
\Brian
Brian,
Well, you did have the cloud shadow described in the original post and I read right over it thinking you had already excluded that as the cause or the shadowing wasn't along the line-of-sight. Now since the Sun wasn't illuminating that part of the sky due to cloud shadows there still may be chemistry changes along the light path. I'm no expert on lower-atmosphere aerosols and state changes from being illuminated by Sunlight or not but perhaps we are both correct with my suggestion being maybe only at the 1% level or less!
What you describe is one of the reasons I don't use sky flats anymore. Since I use small portable telescopes I use what is equivalent to a dome-card flat system. I have no dome so I set things up in the garage or the great room and so can control illumination levels and somewhat the color of illumination that the flat target card is illuminated by. The modern LED lights come in with different color temperatures and I usually use a mix of 5000K and 3400K lamps to illuminate the card. My filters set is made up of Clear, B, V, Rc, Ic Astrodons. Seems to work.
DEY
We already know from tests on our 1.1-m telescope that dome flats using a screen illuminated by lamps are no good compared to twilight flats. But there seems to be some residual in the twilight flat-fielding, evidently from some difference between how the twilight sky looks to the telescope versus the night sky (differing scattered/reflected light issues in the telescope). But that's another subject, and it's own little research project!
\Brian