feedback requested: change photometric calibration strategy?

[mommerm...@gmail.com]

At the moment, the pipeline will match field sources with a catalog and basically use 50% of those sources that it could match to derive a zeropoint for the respective image. In theory, pp_calibrate tries to minimize the reduced Chi2 in the magnitude residuals (blue line in the calibration plots), but since the reduced Chi2 is usually monotonically decreasing with the number of calibration stars used, it basically always stops at the 50% limit.

I observed that in many cases, a reduced Chi2 of unity can actually be reached with a decent number of stars. Technically, this is the best fit and probably the criterion that the pipeline should use. However, this might affect (reduce) the zeropoint uncertainties, which is a good thing, but I would like to hear your feedback first. Is this something you would like to have implemented? If Chi2red=1 cannot be reached, it can still use the 50% limit. Any thoughts?

M