🔭 astrophysics

Identifying and Distinguishing Quenching Galaxies with Spatially Resolved Star Formation in Mock CASTOR and NGRST Observations

This paper presents synthetic observations from IllustrisTNG simulations for the CASTOR and NGRST missions, demonstrating that spatially resolved photometric analysis combined with machine learning can effectively distinguish quenching galaxies from star-forming ones and differentiate between inside-out and outside-in quenching mechanisms out to intermediate and high redshifts.

Original authors: Cameron Lawlor-Forsyth, Michael L. Balogh, Sean L. McGee, Gregory H. Rudnick

Published 2026-07-20

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Cameron Lawlor-Forsyth, Michael L. Balogh, Sean L. McGee, Gregory H. Rudnick

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe as a giant, bustling city where galaxies are the neighborhoods. For billions of years, these neighborhoods have been busy construction sites, constantly building new stars like skyscrapers rising from the ground. But sometimes, the construction stops. The cranes go silent, the blueprints are rolled up, and the neighborhood goes quiet. Astronomers call this "quenching." It's one of the most important mysteries in the cosmos: why do some galaxies keep building stars forever, while others suddenly shut down?

To solve this, scientists look at how the construction stops. Does the building stop in the middle of the city and spread outward, leaving the center dark while the suburbs stay bright? Or does it start at the edges and creep inward, like a fog rolling over a town? Figuring out which pattern a galaxy follows is like finding the fingerprint of the event that killed its star formation. It could be a violent crash with another galaxy, a lack of fuel, or a powerful feedback loop from a black hole. But to see these patterns, we need telescopes that can see the "young" stars (which glow in ultraviolet light) and the "old" stars (which glow in red and infrared) with incredible clarity.

This is where a new generation of space telescopes comes in. We have the Nancy Grace Roman Space Telescope (NGRST), which is like a giant, high-definition camera for the red and infrared universe, and a proposed mission called CASTOR, which is a specialized ultraviolet camera designed to catch the light of young, hot stars. Together, they promise to fill a "gap" in our vision, allowing us to see the full story of how galaxies age. But before these telescopes launch, scientists need to know: will the pictures they take be clear enough to spot these subtle patterns?

In this paper, a team of researchers decided to find out by building a "virtual universe" inside their computers. They didn't wait for the telescopes to launch; instead, they used a massive simulation called IllustrisTNG, which acts like a digital physics lab where galaxies are born, live, and die. They took galaxies from this simulation that were in the middle of "quenching" (shutting down their star formation) and created fake, or "mock," images of them. These images were designed to look exactly like what the CASTOR and NGRST telescopes would see, complete with all the fuzziness, dust, and static noise that real space observations have.

The researchers then acted like detectives, using special software to analyze these fake images. They tried to measure specific "morphological metrics"—basically, shape-shifting clues that tell a story. For example, they looked at how concentrated the star formation was in the center versus the edges, and where the star formation suddenly stopped. They wanted to see if they could tell the difference between a galaxy that was dying from the inside out (like a heart attack) versus one dying from the outside in (like a frost creeping in).

The results were promising. In their simulated world, the team found that they could successfully distinguish between normal, healthy star-forming galaxies and those that were quenching. Even better, they could tell the difference between the different "causes" of the shutdown. They could separate the "inside-out" galaxies from the "outside-in" ones with high accuracy, even when the images were blurry and noisy. They even used machine learning to help classify these galaxies, finding that they could estimate how far along a galaxy was in its "death spiral."

The authors suggest that when the real CASTOR and NGRST telescopes start their surveys, they will be able to find and classify thousands of these quenching galaxies out to intermediate distances in the universe. This means we won't just know that galaxies are stopping their star formation; we will finally be able to see the specific steps of how they do it, turning a blurry mystery into a clear, step-by-step story of cosmic evolution.

Technical Summary: Identifying and Distinguishing Quenching Galaxies with Spatially Resolved Star Formation in Mock CASTOR and NGRST Observations

Problem Statement
Understanding the mechanisms that suppress star formation (quenching) in galaxies is central to galaxy evolution. While various internal (e.g., AGN feedback) and external (e.g., ram pressure stripping, strangulation) mechanisms have been proposed, distinguishing between them observationally is challenging. Previous work (Lawlor-Forsyth et al. 2026, hereafter Paper I) demonstrated that simulated galaxies undergoing quenching exhibit distinct spatial signatures: "inside-out" quenching (suppression starting in the center) and "outside-in" quenching (suppression starting in the outskirts). However, Paper I relied on idealized simulation data without accounting for observational limitations such as dust, noise, and instrumental resolution. This paper addresses the feasibility of applying the morphological metrics developed in Paper I to real-world data from upcoming large-scale surveys, specifically the Cosmological Advanced Survey Telescope for Optical and UV Research (CASTOR) and the Nancy Grace Roman Space Telescope (NGRST).

Methodology
The authors generated a comprehensive set of mock observations to test the recovery of spatially resolved star formation metrics.

Simulation Sample: The study utilizes the TNG50 run of the IllustrisTNG suite, selecting 361 quenched galaxies (stellar mass $M_* \geq 10^{9.5} M_\odot$ ) and their associated quenching episodes. This yields a sample of 5,365 quenching galaxy snapshots and a mass-matched control sample of 64,128 normal star-forming galaxies.
Mock Observation Creation:
- Stellar Populations: Synthetic images were generated using the galaxev stellar population synthesis library (Bruzual & Charlot 2003) applied to simulation stellar particles. The model assumes a Chabrier initial mass function and Padova 1994 tracks.
- Dust Modeling: A spatially varying foreground dust screen was applied. The attenuation profile ( $A_V$ ) was parameterized as a radially declining exponential dependent on galaxy stellar mass and offset from the star-forming main sequence ( $\Delta_{MS}$ ), calibrated against MaNGA observations (Greener et al. 2020).
- Instrumental Effects: Images were convolved with a Point Spread Function (PSF) of 0.15" FWHM (matching CASTOR's expected quality) and noise was added. Noise sources included background (zodiacal light, earthshine), dark current, and read noise, calibrated for both CASTOR (UV/blue-optical) and NGRST (red-optical/near-IR) survey strategies (Wide, Deep, and Ultradeep tiers).
- Redshift: All mock observations were placed at a characteristic redshift of $z=0.5$ .
Analysis Pipeline:
- SED Fitting: Photometry was extracted in 20 concentric circular annuli (out to $5 R_e$ ). Spectral Energy Distributions (SEDs) were fitted using FAST++ with a flexible hybrid star formation history (delayed- $\tau$ model with a suppression/burst coefficient $R$ for the last 100 Myr).
- Metric Recovery: The study recovered stellar mass ( $M_*$ ) and Star Formation Rate (SFR) radial profiles to compute specific SFR (sSFR).
- Morphological Metrics: Four metrics were adapted for observational data:
  - $C_{SF}$ : Concentration of star formation (ratio of SFR within 1 kpc to total SFR).
  - $R_{SF}$ : Ratio of the effective radius of the star-forming disk to the stellar mass disk.
  - $R_{inner}$ and $R_{outer}$ : Radii marking sharp truncations in the sSFR profile (inside-out vs. outside-in signatures).
- Classification: Machine learning techniques were employed to classify galaxies based on these metrics and to estimate the progress of the quenching episode.

Key Results

Recovery of Physical Parameters: The FAST++ pipeline successfully recovered spatially resolved stellar mass and SFR. For stellar mass, the median difference was 0.19 dex with a scatter of 0.23 dex. For SFR, the median difference was -0.13 dex with a scatter of 0.46 dex (improving to 0.31 dex scatter for SFR $\geq 0.1 M_\odot$ yr $^{-1}$ ). Metallicity recovery was less precise due to limited grid spacing in the models, but dust screen strength was well recovered (median difference -0.08 mag).
Metric Recovery: The observational proxies for the morphological metrics ( $C_{SF}$ , $R_{SF}$ , $R_{inner}$ , $R_{outer}$ ) showed strong correlation with the ground-truth simulation values, allowing for the distinction between different quenching populations.
Classification Accuracy: The study confirmed that quenching galaxies can be distinguished from a mass-matched control sample of normal star-forming galaxies. Furthermore, the "inside-out" and "outside-in" quenching mechanisms could be separated with reliable accuracy using machine learning.
Quenching Progress: The metrics allowed for the estimation of the progress through the quenching episode for different populations.
Survey Predictions: Based on the proposed CASTOR and NGRST survey depths (Wide, Deep, Ultradeep), the authors predict that these facilities will enable the classification of thousands of quenching galaxies out to intermediate redshifts, with even higher yields at higher redshifts.

Significance and Claims
The paper claims that the morphological metrics developed in Paper I are robust enough to be applied to real observational data from upcoming space telescopes. By demonstrating that CASTOR (providing UV/blue-optical data to trace young stars) and NGRST (providing red-optical/IR data to trace stellar mass) can jointly resolve the spatial signatures of quenching, the study provides a pathway to:

Distinguish between different quenching mechanisms (inside-out vs. outside-in) in the real universe.
Estimate the evolutionary stage of quenching galaxies.
Predict the abundance of these populations in future large-scale surveys.

The authors emphasize that while radiative transfer calculations were considered, a simplified dust modeling approach was sufficient for the purposes of recovering SFR and mass distributions, validating the use of these mock observations for feasibility studies. The work serves as a preparatory framework for utilizing the "ultraviolet gap" filling capabilities of CASTOR in conjunction with NGRST to advance the understanding of galaxy evolution.

More like this