Beyond UVJ: Color Selection of Galaxies in the JWST Era

TL;DR

The paper introduces a new rest-frame color-color selection method, $(ugi)_s$, that improves the identification of quiescent galaxies at high redshift in the JWST era by reducing contamination compared to the traditional $UVJ$ diagram.

Contribution

A novel $(ugi)_s$ color selection method that outperforms $UVJ$ in purity and reduces contamination in high-redshift galaxy classification.

Findings

$(ugi)_s$ reduces contamination in quiescent galaxy samples by nearly a factor of two.

$(ugi)_s$ maintains high completeness (~85-90%) at $z=3-3.5$.

$(ugi)_s$ achieves a true-to-false-positive ratio greater than 2.2 at $z brace 3.5-6$.

Abstract

We present a new rest-frame color-color selection method using "synthetic $u_s-g_s$ and $g_s-i_s$'', $(ugi)_s$ colors to identify star-forming and quiescent galaxies. Our method is similar to the widely-used $U-V$ versus $V-J$ ($UVJ$) diagram. However, $UVJ$ suffers known systematics. Spectroscopic campaigns have shown that $UVJ$-selected quiescent samples at $z \gtrsim 3$ include $\sim 10-30\%$ contamination from galaxies with dust-obscured star formation and strong emission lines. Moreover, at $z>3$, $UVJ$ colors are extrapolated because the rest-frame J-band shifts beyond the coverage of the deepest bandpasses at $< 5~\mu m$ (typically $Spitzer$/IRAC 4.5 $\mu m$ or future $JWST$/NIRCam observations). We demonstrate that $(ugi)_s$ offers improvements to $UVJ$ at $z>3$, and can be applied to galaxies in the $JWST$ era. We apply $(ugi)_s$ selection to galaxies at $0.5<z<6$ from the (observed) 3D-HST and UltraVISTA catalogs, and to the (simulated) JAGUAR catalogs. We show that extrapolation can affect $(V-J)_0$ color by up to 1 magnitude, but changes $(u_s-g_s)_0$ color by $\leq$ 0.2 mag, even at $z\simeq 6$. While $(ugi)_s$-selected quiescent samples are comparable to $UVJ$ in completeness (both achieve $\sim$85-90% at $z=3-3.5$), $(ugi)_s$ reduces contamination in quiescent samples by nearly a factor of two, from $\simeq$35% to $\simeq$17% at $z=3$, and from $\simeq $60% to $\simeq $33% at $z=6$. This leads to improvements in the true-to-false-positive ratio (TP/FP), where we find TP/FP $\gtrsim$ 2.2 for $(ugi)_s$ at $z \simeq 3.5 - 6$, compared to TP/FP $<$ 1 for $UVJ$-selected samples. This indicates that contaminants will outnumber true quiescent galaxies in $UVJ$ at these redshifts, while $(ugi)_s$ will provide higher-fidelity samples.