IF-MOS trade-off study: Conclusions

The comparison has been carried out, for a resolution of R = 150, by considering observations of unclustered faint galaxies - guided by the HDF-N/S data - to a given s/n per object.

The basic conclusions can be stated as follows:

0. The MOS addresses a (much) larger FOV than the IFS-LR.
1. Due to overlapping spectra and edge effects, the MOS can only observe, at a single pointing, between ~70% (at K_AB = 20±0.5) and ~30% (at K_AB = 27±0.5) of sources in the field. These losses depend on assumptions about how the sky background is determined.
2. The MOS can still address about 10 times the number of sources at K_AB = 20 at a given pointing, but this factor decreases to ~ 5 at the faintest magnitudes.
3. For an MOS working at a fixed spectral resolution (R = 150), the IFS is faster in reaching a given s/n per object (not per pixel) by a factor of between 3 and 4.
4. The precise factors in the above two points depend on the actual choices for spatial pixel size. The IFS is faster due to the slit losses of MOS, which are related to the spatial pixel size, and to the way in which the spatial pixels of both spectrographs are projected onto the detector.
5. The overall speed ratio of the MOS/IFS-LR in 'redshift-engine' mode - using galaxy sizes and surface densities from the HDF and Keck data - varies from around 2 at K_AB = 20±0.5 to around unity at K_AB = 26±0.5. Although many factors contribute to the ratio, the variation with magnitude is produced by crowding, slit loss and object/pixel matching effects. Given all the uncertainties and assumptions in this comparison, the two concepts appear to have similar speeds for faint galaxy surveys.
6. The calibration procedures for an IFS are expected to be significantly simpler than for a MOS. There are fewer changing parameters, and the background subtraction should be easier. For the IFS, object spectrophotometry is insensitive to pointing/tracking. The MOS parameters we have used here, however, give a higher spatial resolution. Also, the MOS can easily exclude (most of) the light from bright objects in the field.
7. In general, operational considerations favour the IFS concept because of its mechanical simplicity and lack of a pre-imaging requirement. The MOS provides opportunities for being operated in a very flexible way to optimise the use of observing time and balance s/n for a range of source brightnesses. However, it still has the limitation imposed by the crowding factors. For deep surveys, the MOS is probably most efficiently used for observing just a narrow range of magnitude in each exposure. The IFS works always in a cumulative manner, ie, it exposes on everything in the FOV.
8. This benchmark comparison is based on an observing programme which most favours the MOS. For clustered sources and single large objects ('astrophysics mode'), the IFS becomes relatively more favoured.
9. The IFS offers greater opportunity for serendipitous discovery, particularly of emission line sources not visible in broadband imaging.