How large are the systematic errors in the parameter estimates?
Just as reassuring, as shown in Figures 3 and 4, there appears to be no significant systematic deviation between the GALFIT-measured and known-input parameters down to galaxies fainter than three or more magnitudes below the sky surface brightness. Around this regime “stealth” neighboring sources (faint or low surface brightness galaxies) in the simulations can explain some of the scatter behavior. Stealth sources can’t be removed or dealt with because SExtractor does not recognize their existence, yet they make up roughly 40% of all sources in the simulations. Their influence is more noticeable when the primary science targets are comparably diffuse or faint. At low surface brightnesses, the issue of sky determination is another which can sensitively impact the fitting quality. Accuracy of the sky determination depends on factors like the Poisson noise, flatfielding errors, the number of nearby stealth sources, and the fitted image size (which is tied to SExtractor estimate of the g
