Skip to content. | Skip to navigation

Personal tools



  • To obtain a good flat field you should use sky or twilight flats. In addition to observations of the twilight sky, it is then also necessary to obtain dark frames with the same exposure time. The dark must then be subtracted from the sky frame before it is normalized to create the flat. Twilight flats should be sufficient to obtain 2% photometry. There may still be residual structure due to scattered light, however, that might depend on rotator angle (among other things).

Note: To obtain a dark frame place the Jc and H2 filters in the beam at the same time (their bandpasses are mutually exclusive).

  • Flats obtained with the flat field screen located in the secondary mirror support do not work as well as twilight flats. They are only sufficient for approximately 10% photometry. This appears to be due to scattered light. The result is a spurious structure that looks like a global tilt in the flat. Dome flats are therefore not recommended for flatfielding, although they may still be useful for creating a bad pixel mask.
  • If you want a dome flat nonetheless, you should obtain flats with both the lights on and the lights off in a given filter. This is particularly important in the redder filters, but also valuable for flats obtained during the day (when there is ambient light in the dome) and to remove structure in the dark.
  • The following table has guidelines on the exposure times and lamp settings for each filter. The final column (DARK) lists the number of counts in an exposure obtained with the flat field lamp off.


Filter Lamp Voltage Exptime ADU Dark
  Volts (seconds)   (ADU)
Yc 1.5 10 13,000 600
Jc 1.3 10 19,000 450
Hc 1.1 10 17,000 430
Ks 1.0 5 16,000 9000
H2 2.0 10 14,000 --
Brg 2.0 10 14,000 --


  • It is recommended that you use a loop of 10-20 for the lamps on and lamps off flatfield exposures.


Caveat 1: Just to reiterate again, since this is a change from the previous version of the manual: You are strongly discouraged from using dome flats for anything except bad pixel masks. Twilight flats work much better.

Caveat 2: The flatfield lamp is quite bright in the NIR and it is therefore quite easy to saturate the detector if the lamp is too bright. This may cause the data frames to appear to have no flux at all (heavily saturated) or unusual structure (lightly saturated). It is very important, particularly at Ks, to verify that the lamps are not too bright. One simple test is to take a slightly shorter exposure (3s instead of the recommended minimum of 5s) to insure that the mean in this shorter exposure is 60% of the mean in the 5s exposure.

Caveat 3: The flat fields obtained at Ks may not be very good. It is therefore advisable to obtain twilight (or simply sky) flats instead. We are currently developing the optimal strategy for Ks flats, although in this case it is important to obtain a dark frame in place of the lamps off image.

Document Actions