Next: 4.2 Focusing the Telescope
Previous: 4 Observing Strategies

4.1 Calibration

• Bias frames. You may wish to take 10-15 of these at some convenient time when the dome is dark; they can be combined into a master bias frame by using zerocombine. In reality, none of these three chips have much bias structure (the hot columns cannot be effectively removed via biases), and so this is fairly pro forma.

• Dark frames. You may wish to take three frames of the same length of time as your longest exposure to evaluate how much dark current is contributing if you are attempting to do long-slit spectrophotometry without local sky subtraction. In practice, the dark current on all three of these chips is fairly minimal. Be sure to make these exposures with the dome dark.

• Wavelength calibration. All spectrographs have some minimal amount of flexure. If you are interested in radial velocities, then doing a new comparison exposure at each new position (and possibly bracketing) your exposures is a good idea. If you are satisfied with shifts of a pixel or two, then a single comparison exposure in the afternoon is good enough. There are a variety of comparison sources available at the 4-m, including a HeNeAr source and Th-A. The thorium-argon source should be used for higher dispersion applications. At the 2.1-m the only comparison source is a HeNeA source.

• Flat-field calibration. Flat-field calibration is needed to remove two instrumental effects:
  • Pixel-to-pixel sensitivity variations on the chip.
  • The ``slit function" in the spatial direction.

  Any line-free source will do to remove the first of these: either exposures of the large white spots on the side of the dome, or exposures with the internal quartz lamps. It is a little trickier to deal with the slit function, although a line-free source is not needed (i.e., twilight skys will work in principle) and yet removing this large-scale structure is crucial for good sky subtraction. Here are your options:

  • Exposure with the internal quartz lamp (projector flat) but correct the slit illumination by a twilight exposure.
  • Use the white spot (dome flat).

  Either of these two methods will work, but you have the following logistical considerations:

  • It takes about 10-15 minutes to do dome flat exposures. If you are using a slit, and not changing anything (such as the slit width), doing dome flat exposures during the afternoon should work fine. You may still want to obtain some twilight exposures to test how well the dome flats work, but our experience from direct imaging suggests that these should work to <1% across the length of the slit. (Note that at the 2-m that very bright twilight exposures will actually not mimic the dark sky illumination very well, which we attribute to the considerable amount of scattered light within the dome during twilight.)

  • Projector flats can be run in a few minutes without moving the telescope, and so are fast and convenient. However, the large scale stucture does not match the night sky very well. In the case of multi-slit masks, though, you may be best off by obtaining projector flats but not using these to remove the slit function: instead, assume that the slit function is constant over the short lengths of a particular slitlet. This reduction procedure is explained in the ``Multi-slits" manual (use apflatten rather than apnormalize.)

• Flux Calibration. Calibration to flux (and removal of atmospheric and instrumental wavelength effects) can be readily achieved by observing one or more spectrophotometric standards during the night. We recommend using the ``Kitt Peak Spectrophotometric Standards". These are 25 mostly line-free stars calibrated at 50Å intervals from 3200Å to 8100Å (Massey et al. 1988 ApJ 328, 315); 11 of these have been calibrated out to 1m (Massey et al. 1990, ApJ, 358, 344). Finding charts can be found in each of the domes, as well in the 1988 paper cited above. The spectrophotometric extinction for Kitt Peak is relatively well-known, and changes appear to be pretty gray. It is our experience that good results can be achieved by observing only a few standards, at airmasses comparable to your object using the same slit-width and observing technique used for your program objects. If you are using the Risley prisms at the 4-m to deal with atmospheric dispersion, you must also observe your standards through these prisms (even at zenith) in order to account for the wavelength-dependent transmission of the glass.