Bootstrapping photometry
When you are learning how to do photometry, how do you know you are getting 'the right answer'? There is no answer in the back of the book, but we can kind of make one.
If you have photometry on sources similar to yours (ideally in the same field, with the same instrument configuration) that someone else did, you can do photometry on the same sources as those other people did. You assume that they did it right (because they published it, or because they built the instrument, or whatever), and then you can try and match their values, within errors.
Imagine I give you a blank wooden stick and tell you to make a ruler out of it. Without any other information, you’d be stuck guessing the lengths, with no way to know if they are right. But what if I gave you two objects of known length (say, a brick known to be 6 inches long and a stick known to be 7 inches long), and then told you to make a ruler out of the blank wooden stick. Then you’d have a shot at making a reliable ruler, right? You could figure out from the two calibrator objects (the 6 and 7 inch things) how long 1, 2, 3, 4, 5, 6, 7 inches were, and from there you could work up pretty far on the ruler. It wouldn't be perfect (because you have to, say, guess what half the 6 in brick is to get 3 in, and guess what a third of the brick is to get 2 in, or find the 1 in difference between the 6 and 7in objects, then subtract 1 inch from the 3in measure, etc.), but in the end, you’d be reasonably close.
This is exactly what I'm describing with the photometry.
In the case of the PACS data we are working with, we are taking the 70 um source list from the PACS group as "truth." Those are our known measurements. If you measure the flux densities of the same sources, you can compare your measurements to theirs. If they are close, then we can trust our approach and continue to use it for the sources that are not in the 70 um catalog from the PACS group.
Try a variety of possible apertures and annuli and, for each one, compare it to the "truth" from the catalog. Pick the combination that matches their "truth" the best. Look at the scatter in those plots. Is all the scatter from them or from you? See if you can reduce your scatter by better selection of parameters. Once you match the "truth" catalog well enough ("well enough"), then you can venture out and do photometry in the same image for sources not in the PACS source list... and be reasonably certain you are getting the right answer.
