Following on from the Data Summary post, we have reviewed all the good data and built an understanding of what the flux-ratios are between B and the other filters. We then use that ratio to test all observations to see if the FW was behaving as expected or showed signs of malfunction. The malfunctions often mean that all fluxes are the same, consistent with one single filter being set instead of the sequence of filters the script asks for.
Focusing then on only those nights where the FW was well behaved we extracted all total fluxes (total bias-corrected image counts divided by nominal exposure time), limited the data to airmasses less than 4, removed the lunar eclipse, and then plotted all fluxes against lunar phase:
(sorry about the strange cropping!) In different colors we see the extinction-corrected fluxes as a function of lunar phase. We only show data between 30 and 90 degrees (90=half Moon) since this is all we can acquire in co-add mode. The blue,green,red and orange symbols correspond to B,V,VE2, and VE1+IRCUT (very similar filters), respectively. The curves are 4th order polynomia fitted with a robust code that omits outliers.
We see some scatter around the lines – some of it (e.g. near 50-55 degrees is probably clouds. Some of it, in IRCUT/VE1 near 90 degrees may be ‘no filter was inserted at all’, or shutter exposure time was longer than requested.
From the fits we can generate flux-tables for use in identifying the remaining data: There are many more images available but they seem to be with unknown filters because the filter acquired and requested were simply not the same. These images may still be of use in DS/BS and albedo analysis – we just need to figure out what filter was used!
Apart from causing many lost images (i.e. we get a bias frame instead) the shutter semes to mostly work as expected when it opens at all. The Filter Wheel, however, selects random filters (as far as we presently know), or no filter at all, when it does not work.
The present work enables a data-selection filter for use in post-observational processing.
Note that the present data are not relevant for DS/BS studies as scattered light has not yet been removed. First we have to identify the images that can be further analysed!
I did it by eye! Looked at histograms and noticed that some sets of data have constant ratios.
Great work Peter! Splendid to resolve this troublesome issue! How did you define exposures where you were confident of which filter was in the beam? – Chris