ALFOSC spectroscopic observations: observing and reduction strategy to best correct for fringes

Based on a discussion with Vallery Stanishev, we specify below some observing guidelines and points of notice regarding the removal of fringes in red ALFOSC spectra. As a rule of thumb, the fringe amplitude can be suppressed by a factor 10 by using standard data-reduction steps, provided that sufficient calibration data were obtained.

For spectra redwards of 6000 Å take 2 flats before and 2 after the science spectrum (with the telescope pointing to the target). These flats can be used to beat down the fringes. The flats have to be taken with the telescope pointing at the target in order to minimise flexure effects.

However, it is found that often quite large residual fringing remains after normal fringe correction. This is likely due to the fact that the calibration unit illuminates the ALFOSC optics differently than the telescope does.

One can try to shift the flat a bit (~+/-1 pixels along the dispersion axis, the shifts are not integer pixels, have in mind) before flat-fielding, and in many cases this can result in almost full removal of the fringes from the target spectrum, at the expense of slightly increased noise.

It should be kept in mind, however, that if a very blue object is observed, the second-order contamination will make removing the fringes almost impossible even if one shifts the flat. The reason is that the blue light does not fringe, but fringing is introduced when dividing by the red lamp flat.

As fringes only affect the red part of the spectrum, only the red part (>6000 Å) of the flats should be shifted: keep the blue part un-shifted. Flats for the blue can be taken in the afternoon.