This guide deals with the practice of astronomical imaging. Most of these considerations enter into the use of photography as well as CCDs or other electronic devices.
Preparation: before observing, one must settle on choices of detector and filter for a given project, guided by availability, sensitivity, pixel (or grain) size, dynamic range, and spectral response. There are common photographic emulsion-filter combinations that match standard optical passbands reasonably well:
|Plate-filter combinations to match popular photometric passbands|
|Band||λeff,Å||Δ λ, Å||Plate||Filter|
|B||4330||950||103aO||2mm GG13 (IIIaJ also)|
|U||3540||600||103aO||2mm UG2 (or CuSO4)|
Similarly, most CCD systems have filters that can be transformed to, for example, the UBVRI photometric system closely. The transformations can be defined more accurately in this case. For some work, such as narrow-band mapping, the exact photometric transformation is secondary to spectral selectivity. Note that when used with an image intensifier, the response of the intensifier's first photocathode (rather than the final detector) determines the passband.
Most astronomically useful filters are either dyed glass or gelatin filters (Schott and Kodak Wratten, respectively) or custom-made narrow-band interference filters. Schott glass filters usually fall into UG, BG, GG, OG, or RG families, plus a number indicating either the filter type or a cutoff wavelength. The second letter apparently means "glass", while the first denotes a spectral range: Ultraviolet, Blue, Green, Orange, or Red. BG filters block near-UV light, with a number giving the half-transmission wavelength in 100-Angstrom units. For example, a BG-38 blocks light shortward of 3800 Å. A GG filter acts similarly blocking blue light, as a GG 495 transmits little shortward of 4950 Å. UG filters block redward of their listed cutoff, and RG and OG transmit only red light. An older nomenclature gives only catalog numbers (i.e. GG13 is similar to GG495). HST instruments have given rise to a convention in which a filter is designated by the letter F, the center wavelength in nm (or tens of nm for some instruments in the near-IR, in a slight breakdown of convention), and a designation as to whether its passbands is Wide, Medium, or Narrow - hence F555W, F606W, F410M. In all cases, the final response function is the product of those of the detector, filter, optics, and (for Earthbound instruments) the atmospheric transmission (which depends on air mass and thus zenith distance, and can change the effective wavelength of blue and near-UV observations significantly).
One should also prepare an observing list with some care - accurate coordinates, finding charts unless the identification is unambiguous (like 1st-magnitude stars) or the telescope sets very accurately. For many programs, these include standard stars or other reference objects. It may be necessary to precess these coordinates to the epoch of observation. Also, to avoid unpleasant surprises, be sure your targets will be above the horizon (preferably by 30° or more) and far enough from the Sun to be observable in a dark sky. Comet observers especially have trouble with these.
Focal positions: imaging is frequently done at the prime or Cassegrain focus, but for such applications as close double stars and planetary mapping, a larger image scale is needed. One can use a Barlow lens, use an eyepiece or microscope objective to reimage the focal plane at a larger scale, or use such a lens plus a camera lens for so-called afocal imaging. The magnification for eyepiece projection is given by the ratio of focal plane - eyepiece and eyepiece - emulsion distances, and for afocal setups by the ratio of camera lens and eyepiece focal lengths. In both these cases, for small telescopes a serious problem with vibration from the camera shutter can appear; the projection setup offers a long unsupported lever arm for vibrations. Sometimes the best solution is the "hat trick" - using a piece of cardboard in front of the objective instead of the camera shutter to limit exposure time, opening and closing the shutter only when the objective is blocked.
Focussing: this can be more trouble than one would suspect. For photographs, three techniques are available for various optical arrangements. If a 35mm camera is attached to a telescope (for planetary photography, for example) one may use the SLR focussing screen, most likely the clear region since split-screen areas black out at long focal ratios. For use with plates at larger telescopes, a knife-edge arrangement may be used. This is a Foucault mirror test in reverse, with the knife edge mechanically held in a duplicate plateholder at the exact location of the emulsion. To focus, one views from behind the knife edge looking at an out-of-focus star image. If a star is moved across the knife-edge location, the star image will black out from one side to the other unless the knife-edge is in the focal plane, in which case there will be no preferred direction and each annulus of the mirror blacks out simultaneously. Finally, for many telescopes with repeatable and metered focus settings, one must take a focus plate, looking at the same star shifted between exposures with different focal settings (usually with a double shift at the end to avoid any ambiguity!). After development and inspection, the telescope may be set to the desired focus. This technique must often be used with CCDs as well, unless the chip can readout rapidly enough for simple on-the-fly focussing based on repated images displayed in sequence. Some implementations use the CCD's register shifting to avoid physical motion of the telescope.
The sensitivity of focus depends on the atmospheric seeing and the f/ratio reaching the detector or emulsion. For a geometrically perfect image, a point source as measured a distance d from perfect focus at an effective focal ratio f will be an image of the telescope aperture (entrance pupil, secondary obstructions and all) with diameter d/f. Thus short-focus systems are less forgiving. In practice there is often a seeing component that makes reaching exact focus difficult (but somewhat pointless). For small systems like Schmidt cameras, it may be possible to make a system so independent of temperature fluctuations that a single focal position will work. Note, however, that the focus will be shifted when a filter is inserted or changed. The focus will be lengthened by the filter's optical thickness (physical thickness times refractive index minus 1 in the long-focus limit). With interference filters, there is the further complication that when placed in a converging beam, the passband broadens and shifts blueward compared to its specified properties in parallel light; the wavelength shift is proportional to 1/f 2.
Making the exposure: this is often as simple as opening the shutter and checking your watch to close it. Sometimes precise timing, as to start or durection, is needed. The easiest time coordination uses standard radio stations (WWV) which broadcast time signals at standard frequencies - 5, 10, 20 MHz, for example. Electronic receivers can turn these into trigger signals, or use them to correct electronic timers. For some applications, it is important to know the duration or start and stop times to a fraction of a second, but not necessarily to control them to this level. For short exposures with focal-plane shutters, the entire field may not be equally exposed; many camera systems have minimum useful exposure times for a given level of accuracy and uniformity.
During the exposure, guiding is often necessary. Few telescopes can track well enough during long exposures to take quality images free of trailing. Even mechanically excellent drive mechanisms can fall victim to atmospheric refraction or slight mislignment of components. One may guide using a separate co-aligned telescope, an eyepiece or TV camera viewing a small portion of the field beyond the area of the camera, or a more elaborate autoguiding system. With some systems, there is advantage to preselecting guide stars (from the SAO or HST-GSC catalogs for example), if you would otherwise spend valuable time hunting for them. At some observatories, this can be done automatically.
Visual guiding uses a special eyepiece with a crosshair in the focal plane, illuminated faintly by a source out of the line of sight. Spider web is an excellent appropriately thin wire, and purely optical patterns can also be used. Sometimes the eyepiece should be defocussed slightly if there is risk of the star being hidden by the crosshairs. A very smooth control of telescope motion is a must, mechanical or electrically driven.
If guiding is impossible, or multiple exposures are desirable to increase dynamic range or overcome cosmetic defects, it is popular to use a shift-and-add technique. Multiple exposures are made with the telescope slightly shifted, then co-added digitally in proper register rejecting any features not fixed on the sky. Photographs can be stacked in the darkroom in an analogous procedure (though this can be very delicate).
One may need to interrupt an exposure, to avoid excess skylight during cloudy intervals or skip times of especially poor seeing. Interrupted exposures are common in rapid-guiding systems, to improve the net image quality at the expense of a smaller duty cycle.
Special considerations for photography: there are some hints derived from the delicate nature of the process. There is often a critical exposure time to show some feature of interest; plates don't follow Poisson statistics. An extra 10% in exposure time may give a disproportionate improvement not found in digital devices. By the same token, sky photographs should almost always be exposed long enough for the sky background to reach some density near 0.5-1; otherwise features near the faint limit will still be on the toe of the characteristic curve and not well shown. Batches of plates can be processed together; it's a good idea to do one right away during the night to guard against disasters. A single test plate can prevent a wasted night.
2006 © 2000-2006