Originally I thought the context of this discussion was exclusively focused on FWHM where it easy to discriminate the difference between good and average data. However when it comes to gradients from the moon or other sources it does add complexity.
From a data rejection perspective, its imperative for the algorithm to determine what is outlier data i.e. not normal. Combining several average subs with a gradient will see the information you want to reject be viewed as normal. Care should be taken to exclude or reduce the quantity of average subs but having a couple in place should do no harm. Equally, depending on the gradient it is likely not too difficult to clean up post processing.
|