Taking Flat Fields


When taking CCD images, and particular, when trying to use those images for scientific purposes, it is important to reduce the amount of unwanted signal and unwanted noise from each image. Optical path “noise” (some of which is actually signal), is such a problem that many astronomers really never come to grips with it. Their data suffer, and the end result is poorer science. This treatise will spell out the simplicity of taking good flat fields to reduce optical path noise and CCD sensitivity issues and will also walk you through a couple of methods to get flats done.

Optical Path Noise:

Telescopes, CCD chips and filters all block light as well as transmit light. They also harbor dust, finger prints, and other unwanted shadow producing things in the light path. The result of such optical path obscuration is an unevenly illuminated CCD chip. This is a real nightmare for anyone doing photometry, in which a standard star of known brightness might measure a bit faint one night because it was being imaged on top of a dust speck on the filter glass! Optical path vignetting and other physical path obstructions will also cast large, non-discernable shadows onto your CCD, causing poor even illumination.

CCD Sensitivity:

In the spatial realm both on the multi-pixel and single-pixel level, a CCD chip will display uneven sensitivity to incoming light. This can depend on the thickness of the substrate and uneven cooling among many other issues. This creates issues very much like those mentioned already in the optical path noise section above.

The Solution:

Take flat field images and divide them out of your images. A flat field is an image taken of an evenly illuminated object like the dusk sky, or a special illuminated white card hanging on the wall of the observatory. These images are taken through the telescope:

· at the same temperature as your nightly work,

· through the same filter/s as your nightly work,

· at the same focal point and at the same rotational angle being used all night,

· and with integration times to allow the flat to reach an average of between 20 to 50% full well capacity of your CCD chip. Flat images should never bloom, but should also not be less than a second in integration time.

For precision work, 20 to 30 or more flats through each filter should be taken each night you are collecting science data. Each flat of a given filter should then be averaged together to create a master flat which is then divided out of your light frame on a pixel by pixel basis. These details are usually all handled automatically by your software. I will assume you are using MaxIm DL software revision 5+ for the following examples.

In Practice – Taking Sky Flats:

Taking flats is easy. Here is a step-by-step method to take sky flats which has worked well for me for years. You need no special equipment other than that you already own to take CCD images.

1. Wait until the sun is setting, but still just above the western horizon.

2. Turn on your observatory: EVERYTHING. The mount, the fans, the CCD, the PC, lights normally on, etc.

3. Cool down your CCD to the night time working temperature. Wait 10 minutes for it to settle to the working temperature.

4. If you are using filters, you should take flats in order of densest filter to most transmissive. I work in the order of Ha, B, V, R, then lastly I. Set your filter wheel to the first filter.

5. Set the focal point of the system. Minor adjustments through the night are ok in order to allow for temperature changes of your optical tube assembly. Do not make changes more than a mm or so. You’ll have to take new flats if you do make larger changes.

6. Set the CCD camera’s angle to the system. Leave it here all night.

7. Point your telescope at the blue sky towards the western side of the meridian. Avoid areas of sky where there are bright stars (which will not be visible yet, as the sun is still up).

8. Take a 1 second integration.

9. Once it downloads, use MaxIm DL’s Information Window in Area Mode to inspect the average pixel count of the image. If it is too bright, some pixels will be saturated, and you will have to wait until the sun sets some more. If you have an image that reads about 20-50% of the full well count, then proceed immediately to the take a series of flats.

a. Generally the Sun is at a point in the west where its light might just be still touching the top of the treetops on the eastern horizon. Stars are not visible to the eye, nor generally to the camera yet.

b. My full well count with an SBIG camera is 65535, so I aim to get flats with an area average of 20000.

c. You can use MaxIm DL’s image series command to take a set of flats with any given filter. Repeat all the steps above as needed until you have flats for what you need.

10. You can use these flats for as long as you wish, but for precision work, flats are taken every night and sometimes in the morning after your imaging is complete. If you are not after precision work, then taking flats once a week is enough. Some would say that’s sacrilege!

A helpful hints:

If you want to start taking flats earlier, just to give yourself some time, cut out sheets of frosted mylar (used in silkscreening) to cover the objective of the telescope. Use 5 to 10 sheets of this milky white plastic material to basically dim the incoming sky brightness to the optics.

You can take flats while aiming at evenly illuminated clouds. This is ok!

I have gotten away with as few as 6 averaged flats. For truly accurate work, I have gotten up to 40 averaged flats.

Here is a flat. Look how ugly it can be! The donuts are dust. The edge darkening in the corners is caused by vignetting.


About johnb

- Director of Grainger Observatory, Phillips Exeter Academy. - Variable-star-crazed astronomer, but have done research in other areas. - Drummer, archer, pilot, chef, friend, pet owner, husband, father, Train-nut.
This entry was posted in Astronomy. Bookmark the permalink.