Evaluating I band images: dependence on aperture radius

I have severe problems getting reasonable I band data for red giants. The data vary extremely on the number of pixels I select for the aperture radius.
Example:
The FWHM for the star is 2.86.
If I look at the star profile I see that ADU number is approaching zero only at about 11 pix. Taking 1.5*FWHM I still see ADU approx. 2000.
Looking at the" Effect of Signal…" I see that the Instrumental Magnitude (black) also going down all the way up to the 10.5px which is the maximum visible, not yet stable.

When I run the photometry I see a dramatic dependence of the calculated I magnitude on the aperture radius.
Example:
4.5 (1.5*FWHM): 4.638 (error 0.019 for 2 comp stars)
8: 4.886 (error 0.003)
12: 5.101 (error 0.006)
16: 5.233 (error 0.029)
18: 5.242 (error 0.027)
19: 5.237 (error 0.027)

Looks like convergence for 16 pix and above, way above the FWHM.

Now I could maybe reduce exposure time, but then all the comps stars are useless as they have I band manitudes about 5 mags below my target: The image shows some bright stars but they are no comp stars, also not when I add the APASS data. Probably most of the bright I stars are variable…My comps stars have S/N of 20-25, my target 250. Reducing exposure time is not a good idea I guess.

Therefore it is a bit like guessing which aperture to use. What are you guys doing?

Hi Matthias,

I’m no expert in this area as I only started doing photometry a few months ago, but how to judge the quality of results coming from different settings and options (aperture, choice of comp stars, etc.) seems to be a difficult question. Your experiment with the different aperture settings is interesting, I have not tried to do that in any systematic way as of yet.

So far in my analyses I have been going on the reported error value and the amount of discrepancy on the check star (between the measured and reference magnitudes) to judge what results are most accurate, though I’ve found that these often don’t agree. (That is, one set of aperture and comp stars might result in the lowest error value on the target but a different set results in the closest measured magnitude of the check star.) Also, I’m using a color sensor and measuring TG magnitudes using V reference mags, so that may introduce a whole other layer of difficulty beyond what you’re doing.

So I guess I don’t have any real answer here, but I’m curious about your question and hope others will join in and give their thoughts here as well.

Brian (SBQ)