There is a fundamental choice in photometric "blank sky" surveys of distant galaxies: whether to build up deep exposures in a small sky area or to take relatively shallow exposures in a much wider area. Other than that, the only remaining variable of significance is the choice of filters, which is dependent on the scientific program, but in the optical region the constraints of a wide band pass and availability usually leave relativity few choices. For a fixed observing time, shallow surveys maximize the total number of galaxies (and other objects) obtained and are particularly suitable for finding relatively rare objects. Probably the ideal overall strategy is a predominantly shallow survey with some areas selected for a deeper imaging. The later could be chosen on the basis of either containing "interesting objects" or perhaps an area where many redshifts are available. The discussion below promotes wide angle sky surveys in their relation to Megacam, without seeking to balance them against other possibilities.
The opportunity of a truly wide area sky survey is clear. There is essentially no competition. The impressively successful APM survey does not reach the depths of a modern CCD camera. Similarly, the relatively short exposures (a little under 1 minute) of the 2.5 meter Sloan survey will cover a vast sky area, but not reach to the depth that is spectroscopically accessible with larger telescopes. In fact the Sloan exemplifies the argument: telescopes of varying apertures have a natural range of efficiency. It would be quite unwise to compete in the Sloan's regime, roughly 22nd magnitude and brighter, with CFHT. The very large telescopes equiped with large fields, such as Subaru, will be able to quickly go much deeper than CFHT, but in much smaller fields. That leaves CFHT an enormous middle ground of great scientific opportunity.
The regime extending to about 24th magnitude, which can be done in a single Megacam exposure of 15-20 minutes, is a particularly interesting one. The distance modulus to z = 1 is 43.1 magnitudes (for q0 = 0.1 and 42.7 mag for q0 = 0.5) which means that galaxies a magnitude or so below the characteristic M* are readily visible. It also happens to be the magnitude range which allows the study of halo stars in our own galaxy, including picking up about one disk white dwarf per square degree at the absolute magnitude (M ~ +17 or 18) that is beyond the expected peak in the white dwarf luminosity function for the full Hubble age. Furthermore, this is the range where many very distant quasars are expected to be found. These distant QSO's are key objects for an understanding of the conditions of the universe at redshifts beyond 5, where the distance modulus is 47.9 mag.
Of particular interest in a wide angle survey will be large rich clusters of galaxies near redshift one. These objects can be used to measure a variety of cosmological parameters (Mass, baryon, , H0 for variable lensed sources, the amplitude of the mass density fluctuation spectrum, and t0 from the galaxy ages) and are exceptionally important in the understanding of galaxy evolution. Objects which are efficient to followup with spectroscopic surveys are the exceptionally rich clusters with velocity dispersions greater than 800 km/sec. Smaller clusters are more abundant, but will lead to only 10-30 member galaxies brighter than about 1 mag below M*. Such small clusters will lead to large errors on the quantities of interest. The sky density of exceptionally rich clusters is already fairly well understood, and is likely to be about one cluster every 2 to 5 square degrees. From CFHT photometric data alone, it will be possible to derive quite fairly accurate redshifts and an indicator of cluster mass sufficiently accurate for some cosmological parameter estimation and to study galaxy evolution. These clusters will be of immense interest to X-ray satellites (in particular XMM and AXAF, the latter being a pointed survey) and Sunyaev-Zeldovich measurements using radio interferometers. These can accurately establish the gas mass and the abundance of various elements in the gas. In as much as this gas is representative, it is a useful indicator of the chemical evolution of the universe. Having a good understanding of the statistical properties of these clusters is going to be an important part of the Cosmic Background Radiation studies, since they add a significant foreground signal very near to the angular scale where the temperature fluctuations on the sky are expected to show the maximum from primary anisotropies.
Megacam, as a wide field camera, should be exploited to do the widest possible sky surveys, although it is not the intent of this discussion to suggest that either deep surveys or targeted surveys are of less merit. It would be entirely feasible for Megacam to cover as much as 1000 square degrees in a shallow survey mode. A two colour survey would allow considerable standalone science based on the "red sequence" galaxy redshifts. Furthermore, it would be a unique resource for followup spectroscopic and proper motion surveys.
The two questions posed at the Megacam meeting were the desirability of optical UV response and an ADC. The latter is quite undesirable from the point of view of this program. Image information is a vital aspect of faint galaxy identification and scientific analysis. Extracting faint galaxy morphological information is already difficult enough from ground that one absolutely does not want to introduce any further PSF variability into the data. UV imaging would be valuable but probably would not even be part of a first cut of the programs outlined here.