Mon, 10/10/2016 - 23:15
I am wondering which would be better. Over or under sampling. At my observatory I have large swings in seeing over the night. When seeing is average my sampling is in the 2-3 range, however I get large periods of really good seeing when I'm undersampling, This seems to be worse during the summer than the winter. So my question is would over sampling be better than under sampling?
Thanks
Jim
Jim,
If you have not already, you might want to read pages 17-18 in the CCD manual:that cover sampling.
https://www.aavso.org/ccd-photometry-gude
One of the quickest fix's for under sampling is to simply de-focus your image and spread the light out over more pixels.
IMO, over sampling is a better option than under sampling, if you are unable to achieve the approximate number of 2-3 pixels per FWHM.
You will increase the degree of uncertainty (another topic) in relationship to the amount of over sampling. On the the other hand it has been my personal experience and experience with a number of observers whom I have mentored that under sampling can often result in suspicious results.
Ok, as a practical matter I would avoid unde rsampling if it were to get much worse than ~ 1.8 pixels per FWHM and I usually quit observing once the over sampling gets around 5.5 pixels per FWHM; and I have gone all the way out to 7 pixels/FWHM if the data was important and the seeing, while poor, at least remained constant.
This is a good topic and glad to see it come up from time to time. I am sure more information will be provided here by some other observers on this topic, also.
Tim Crawford, CTX
Tim,
Thanks for the info. I thought it would be better to be a little over sampling than under sampling. I have changed to a different focal reducer to test . That is if the clouds ever go away!
Jim
In the context of digital signal processing, oversampling is always better than undersampling. An oversampled signal can be decimated to reach the Nyquist rate, but there is nothing that can be done to construct the signal if one undersamples.