Survey depth. The natural depth of the survey is set by the expected overheads, as discussed in section 1.7, leading to 5sigma point source limits Y=20.5, J=20.0, H=18.8, and K=18.4. Each of these results from a single pass at the minimum sensible integration time without becoming too inefficient, except for J that has two passes. Note also that we have defined a standard magnitude scale for the proposed new Y filter assuming zero colours for Vega. The increased J-depth follows from both the need for proper motions, and the need to measure red objects.
Proper motions. To determine proper motions a repeat visit in either J or K (see below) is required 2-3 years after the main JHK observation. It was originally envisaged that the repeat observation would be done in the K band. This would improve the overall K limit by a factor square root of 2, or 0.4 mag. However, the objects that have proper motion stars, in particular L and T dwarfs, are best measured in the J band. In particular, the best hope of recognising Population II brown dwarfs must be from their large proper motions. This suggests the second epoch survey should be done in the J band. The ultimate decision on this issue will be made when the J-K colours of L and T brown dwarfs are better constrained.
Choice of filters. Many of the most interesting objects will be very red, and the fainter ones will not be present in the SDSS. Three IR bands (JHK) are therefore certainly necessary for diagnostic purposes. Furthermore, to measure colours for the reddest objects J needs to be deeper than a single pass (see e.g. Figure 2.4 in the Appendix). On the other hand, the coolest T dwarfs are expected to be relatively blue, so that searches for these may have to limit the search to J<19.5.
In the previous section we made the case for the use of a new Y filter to look for high-z quasars. More detail is provided in the Appendix. Here we add the inverse point, that the Y filter will help to separate out T dwarfs from quasars, and indeed help in stellar classification in general.
Survey Area. The ambition of being a fundamental sky atlas argues for complete Northern Hemisphere coverage but that is not yet feasible. Large scale structure and galaxy and quasar population studies require a few thousand square degrees in order to build up large statistical samples. The key issue is the rare object searches - redshift seven quasars, solar neighbourhood substellar objects, halo brown dwarfs. We set survey size at 4000 sq.deg. to aim at approximately ten of each of these.