Quality Parameters

[Tom Droege <tdroege2@earthlink.net>]

I am beginning to develop ways to tag the quality of a measurement.

I can see doing 4 things so far and might develop others in the future.  I 
would like to add two columns to the .cal files to indicate quality.  This 
is mostly for Michael's consideration, but others may want to comment.

The things I am working on:

Brightness

Look at the median level of the dark and flat corrected frames.  This 
measures moonlight, dawn, dusk,  haze and general cloudiness.  All make 
this number increase.  One might still get useful data for determining 
things like the period of a variable, but we might like to exclude this 
data from the good photometry.

Patchyness

Compare the median from the whole image to the median of sub images.  I am 
currently using 64 sub images and the process takes acceptable computation 
time.  This detects clouds, airplanes, and bright stars.  One can just look 
at sub images with low medians.  This then just detects clouds.  (If one 
patch is below the median, some other must be high.)

Gradient

Measure the median of a strip at the top and bottom of the image and take 
the difference.  This is a measure of what is going on in the cruddy sky 
around Chicago.  Dust and dirt in the sky make the air mass transparency 
change more than might be expected just from the amount of air that is in 
the viewing path.

Sky Clarity

Measure the radiation to free space in the direction that the telescope is 
pointing.  I am working on a device to do this which looks promising.  This 
is a possible way to detect "photometric" nights automatically.

--
Some of these change with the filter.  So one word is needed for each 
filter.  I would propose ORing the quantities together as Michael presently 
does for the flats.  One 15 bit word would give enough resolution to be 
useful for the 4 parameters above.

Can anyone suggest other things that should be measured?

The idea is that we will want to be able to sort out dat later to pick out 
things that are important for a particular measurement.

Tom Droege