...
Both phi and FWHM depend on the "airmass" a of the observation. Looking straight up corresponds to a=1. Looking at an angle of 60 degrees from the zenith has a=2. The atmosphere attenuates the light by an amount that depends on the airmass as well as the passband. So the SNR as a function of airmass and exposure time is a sensible merit function that we seek to optimize, summed over all fields and over the 10 year duration of the survey. This is subject to some science-driven constraints, such as the desire to achieve a uniform survey-integrated SNR across the survey area.
...