Magnification bias reveals severe contamination in Hubble Frontier Field photo-z catalogs

TL;DR

This study uses magnification bias to statistically estimate contamination levels in Hubble Frontier Field photometric redshift catalogs, revealing that over half of the high-z galaxy candidates may be low-z contaminants, impacting the interpretation of galaxy luminosity functions.

Contribution

First application of magnification bias to assess contamination in HFF photometric catalogs, highlighting significant low-z contamination in high-z galaxy samples.

Findings

Approximately 56% of z~4-5.5 candidates are likely low-z contaminants.

Magnification bias can effectively estimate contamination levels in lensing field catalogs.

Current selection methods may not sufficiently exclude low-z interlopers.

Abstract

Gravitational lensing by massive galaxy clusters enables faint distant galaxies to be more abundantly detected than in blank fields, thereby allowing one to construct galaxy luminosity functions (LFs) to an unprecedented depth at high redshifts. Intriguingly, photometric redshift catalogs (e.g. Shipley et al. (2018)) constructed from the Hubble Frontier Fields survey display an excess of z$\gtrsim$4 galaxies in the cluster lensing fields and are not seen in accompanying blank parallel fields. The observed excess, while maybe a gift of gravitational lensing, could also be from misidentified low-z contaminants having similar spectral energy distributions as high-z galaxies. In the latter case, the contaminants may result in nonphysical turn-ups in UV LFs and/or wash out faint end turnovers predicted by contender cosmological models to $\Lambda$CDM. Here, we employ the concept of magnification bias to perform the first statistical estimation of contamination levels in HFF lensing field photometric redshift catalogs. To our great worry, while we were able to reproduce a lower-z lensed sample, it was found $\sim56\%$ of $3.5 < z_{phot} < 5.5$ samples are likely low-z contaminants! Widely adopted Lyman Break Galaxy-like selection rules in literature may give a 'cleaner' sample magnification bias-wise but we warn readers the resulting sample would also be less complete. Individual mitigation of the contaminants is arguably the best way for the investigation of faint high-z Universe, and this may be made possible with JWST observations.