HDF Scheduling Details

The initial scheduling was completely automatic, based on the algorithm described below. Further hand editing was to put in DARK and scattered-light calibrations, and to allow more time for guide star acquisitions.

A few statistics:

			10-sigma limits (20 pix area)
Filter Nexp    hours    counts/s     AB mag
F300W 100      48.93     0.0148      27.47
F450W  62      36.52     0.0232      28.12
F606W  77      34.94     0.0370      28.71
F814W  49      34.86     0.0299      28.02

The initial scheduling algorithm

Pass 1 schedules long bright & dark windows
    Bright: minimum background > 7.5e-7 erg cm-2 s-1 A-1 arcsec-2 (= 22.4 ABMAG in F814W)
    Dark:   maximum background < 4.7e-7 (22.95 ABMAG in F814W)
    Long:   > 2100 sec

    Bright windows all go to F300W
    Dark windows are parceled equally to F450W, F606W, and F814W.

    Most of the observing time was scheduled in this pass.

Pass 2 schedules gray periods which are either
    Darkish:   maximum background < 6.9e-7 (22.5 ABMAG in F814W), with t>2100 sec
    Gray:      mean F814W background <= 6.9e-7 (22.5 ABMAG in F814W) with t>500 sec

    In practice, only a small fraction of these windows have mean
    background more than 1.2 times the minimum, so these are
    essentially dark time, except that they are more prone to any
    uncertainties in the scattered light model.

    In scheduling these periods, there is some attempt to give F450W preference
    for long observing times. At the lowest background levels, the required
    integration time for sky noise to equal read noise is 1740, 390, and 780 sec
    for F450W, F606W, and F814W, respectively. So F606W gets all the observing
    windows shorter than 500 sec, F814W gets all the windows with 500 < t < 1800 sec,
    and the longer windows are split between F450W, F606W, and F814W, but with
    more weight given to choosing F450W.


Pass 3 schedules the remaining periods with t>300 sec
    Periods with mean background in F814W > 1.4e-6 are given to F300W
    Periods with mean background in F814W < 6.0e-7 are given to
        F606W if 300 < t < 1000 sec
        F814W if 1000 < t < 1500 sec
        F450W if t > 1500 sec
    Periods with mean background b in F814W 6.0e-7 < b < 1.4e-6  are given to
        F300W if t < 1000 sec
        F814W if 1000 < t < 1500 sec
        F450W if t > 1500 sec
 
    In practice 87% of these windows have mean background more than 1.5 times the
    minimum, so these really are not adding much to our S/N in the red filters.