The PSF smoothing effect on concentration-related parameters of high-redshift galaxies in HST and JWST

Aims: We performed a comprehensive investigation of the PSF smoothing effect on the measurement of concentration-related parameters (C, Gini, and M₂₀) of high-redshift galaxies in the HST and JWST surveys.
Methods: Our sample contains massive galaxies (10^9.5 M_⊙ ≤ M_* ≤ 10^11.5 M_⊙) from the CANDELS/EGS survey (at redshift 0 < z < 2), and the CEERS survey (at redshift 1 < z < 3). The non-parametric concentration-related parameters (R₂₀, R₈₀, C, Gini, and M₂₀) and the model-dependent parameters (n and R_e) of these galaxies were derived from Statmorph and GALFIT, respectively. The best-fit Sérsic index (n) derived from image modelling is generally robust against the PSF smoothing effect and can be used to describe the intrinsic light distribution of galaxies. On the other hand, the concentration-related parameters are significantly affected by the PSF smoothing effect since they are directly calculated from the pixels of galaxy images. We tried to evaluate the PSF smoothing effect by comparing the concentration-related parameters to the Sérsic index in both observations and mock images.
Results: We find that the concentration index is generally underestimated, especially for smaller galaxies with a higher Sérsic index (eventually converging to the concentration index of the PSF). However, galaxies with a lower Sérsic index (n ≤ 1) or larger relative size (R_e/FWHM > 3) are less affected by the PSF smoothing effect. Tests with idealised mock images reveal that overestimating the measured R₂₀/R_e ratio leads to underestimating the concentration index C. Another commonly used concentration index C₅₉, derived from R₅₀ and R₉₀ values, is less affected by the PSF. The Gini coefficient and the absolute M₂₀ statistic also show a similar behaviour as the concentration index. Caution should be taken for the possible correction of the concentration-related parameters, where both the relative size and the Sérsic index of the galaxy are important. We also generated high-redshift artificial images from the low-redshift HST observations and confirm that the traditional correction method that simply adds a single term to the non-parametric indicators of galaxies at higher redshifts is unable to reliably recover the true distribution of the structural parameters. Compared to the HST images, the PSF smoothing is much less severe for images in the CEERS survey (for the short-wavelength filters) due to the much higher spatial resolution. In fact, it is better to use the Sérsic index rather than the non-parametric morphology indicators to trace the light concentration for galaxies at high redshifts. From the single Sérsic modelling of the HST and JWST images, we also confirm that galaxies at higher redshifts are more compact with smaller R_e. The low-mass galaxies are more disc-like (n ∼ 1) compared to the high-mass galaxies that are more spheroid dominated (n ∼ 3).