Rick Kessler (U. Chicago) has run some simulations of the statistical accuracy we might expect to obtain in measuring the atmospheric extinction using a multiple-bandpass filter on the SNIFS imager.

Figure 1 shows the RMS extinction determination assuming 50 stars per field measured with signal-to-noise of 100; the cases for 5 or 7 sub-filters are presented. These calculations assume that the spectral slope (color) between adjacent bandpasses has been measured.

The upper half of Figure 2 shows the atmospheric components that went into the model, while the lower half shows the assumed filter bandpasses relative to the atmospheric components. Rayleigh extinction is the solid black curve, aerosals are represented by the dashed black line, H2O is in blue, ozone is cyan, and O2 is in red. No attempt has been made to measure the O2 absorption bands at 6900A and 7600A using this technique; they are shown to indicate regions the filters must avoid in order to obtain a clean signal for the other components.