On The Robustness of z=0-1 Galaxy Size Measurements Through Model and Non-Parametric Fits

We present the size-stellar mass relations of nearby (z=0.01-0.02) SDSS galaxies, for samples selected by color, morphology, Sersic index n, and specific star formation rate. Several commonly-employed size measurement techniques are used, including single Sersic fits, two-component Sersic models and a non-parametric method. Through simple simulations we show that the non-parametric and two-component Sersic methods provide the most robust effective radius measurements, while those based on single Sersic profiles are often overestimates, especially for massive red/early-type galaxies. Using our robust sizes, we show that for all sub-samples, the mass-size relations are shallow at low stellar masses and steepen above ~3-4 x 10^10\Msun. The mass-size relations for galaxies classified as late-type, low-n, and star-forming are consistent with each other, while blue galaxies follow a somewhat steeper relation. The mass-size relations of early-type, high-n, red, and quiescent galaxies all agree with each other but are somewhat steeper at the high-mass end than previous results. To test potential systematics at high redshift, we artificially redshifted our sample (including surface brightness dimming and degraded resolution) to z=1 and re-fit the galaxies using single Sersic profiles. The sizes of these galaxies before and after redshifting are consistent, and we conclude that systematic effects in sizes and the size-mass relation at z ~ 1 are negligible. Interestingly, since the poorer physical resolution at high redshift washes out bright galaxy substructures, single-Sersic fitting appears to provide more reliable and unbiased effective radius measurements at high-z than for nearby, well-resolved galaxies.