The dark history for the north is relatively complex and is marked by events that do not always correlate band to band. I've broken each band down into the most reasonable periods defined by hardware events that typically cause an offset in the average frame value.
Examination of the individual nightly darks differenced with the period average show a fairly strong relationship between the average offset from the average and the steepness of the ramp seen in the quadrants. Fortunately coadd/photometric quality is relatively insensitive to errors in the darks to the level of hundreds of DN (During OPS we tested a scan using both good darks and darks taken with a half open shutter. The coadds were virtually indistinguishable, differing only slighlty in the noise patterns at a fraction of the 1 sigma level).
If approved I will deliver these canonical darks to the pipeline. Further subdivision is
unlikely to yield substantially better results, though if there is a strong desire, we could combine
some of these into longer periods. The frame median plots on the top page give the clearest indication
of the discontinuities at these breaks.
The stretch is -500 to 500 DN for averages, 0 to 300 DN for standard deviations.
The stretch is -400 to 400 DN for averages, 0 to 300 DN for standard deviations.
The stretch is -200 to 500 DN for averages, 0 to 200 DN for standard deviations.
