Mid-infrared Variability-based AGN Selection using the Multi-epoch Photometric Data from WISE

This study demonstrates that combining mid-infrared variability metrics from WISE multi-epoch data effectively identifies active galactic nuclei, recovering about 28% of optically selected AGNs and revealing candidates in optically inactive hosts while showing minimal contamination from transient events.

The Gist

Imagine the universe as a giant, bustling city. In the center of many of these city-blocks (galaxies) sits a massive, invisible monster: a Supermassive Black Hole. Sometimes, this monster is sleeping quietly; other times, it's awake and eating voraciously, gobbling up gas and dust. When it eats, it glows incredibly bright, especially in invisible "heat" light called Mid-Infrared (MIR). Astronomers call these active monsters AGNs (Active Galactic Nuclei).

The big challenge? Some of these monsters are wearing heavy "masks" (dust clouds) that hide them from our eyes, and some are just too quiet to be seen with standard telescopes.

This paper is like a new detective story. The authors, led by Shinyu Kim and Minjin Kim, tried to find these hidden monsters not by looking at their static "face" (like a photo), but by watching them wiggle and change over time.

Here is the story of how they did it, explained simply:

1. The Old Way vs. The New Way

• The Old Way (Optical Spectroscopy): Imagine trying to identify a person in a crowd by asking them to shout a specific phrase. If they shout, you know who they are. But if they are wearing a gas mask (dust) or whispering (low energy), you miss them completely. This is how most astronomers find AGNs—by looking at their light spectrum. It misses the quiet or hidden ones.
• The New Way (MIR Variability): Instead of asking them to shout, the astronomers decided to watch the crowd for 10 years and see who is moving. The idea is that active black holes are restless; their brightness flickers and changes like a candle in a draft. Inactive galaxies are like sleeping rocks—they don't change much.

2. The Detective Tools (The "Three Clues")

The team used data from the WISE and NEOWISE space telescopes, which have been taking pictures of the sky for about 14 years. They looked at two specific colors of infrared light (W1 and W2). To decide if a galaxy was a "monster" (AGN), they used three clues:

1. The "Is it Wiggling?" Test ($P_{var}$): They calculated the odds that the changes in brightness were real and not just camera noise. If the odds were over 99%, it passed the first test.
2. The "Synchronized Dance" Test ($r$): This is the clever part. If a black hole is flickering, it should flicker in both colors of light at the same time, like a dancer moving both arms in sync. If the light flickers in one color but not the other, it's probably just a glitch or a random error. They required a strong "sync" score (correlation) to pass.
3. The "Red Coat" Test ($P_{AGN}$): Active black holes usually look very "red" in infrared (like wearing a red coat). They checked how often the galaxy wore this "red coat." However, they found that relying on this alone was a mistake (see below).

3. The Big Discovery

When they applied these rules to galaxies they already knew were AGNs (the "known suspects"), they found:

• The "Wiggling + Sync" method successfully identified about 28% of the known AGNs.
• The "Red Coat" method (just looking at color) was a disaster. It only found 9% of them! Why? Because some AGNs are so dim or hidden by their host galaxy's starlight that they don't look "red" enough. Relying on color alone is like trying to find a spy only by looking for people wearing red hats; you miss all the spies in blue hats.

The Best Strategy: The authors concluded that the best way to find these monsters is to look for the synchronized wiggling in the infrared light, ignoring the color for a moment.

4. Finding Monsters in Quiet Neighborhoods

The most exciting part? They applied this "wiggling" method to galaxies that looked completely normal and inactive (no shouting, no masks).

• They found that 3% of "Star-Forming" galaxies (galaxies busy making new stars) were actually hiding AGNs.
• They found 12% of "Composite" galaxies (a mix of stars and AGN) were hiding AGNs.
• Even in "Normal" galaxies, they found a tiny fraction (0.4%) of hidden monsters.

The Analogy: It's like walking through a quiet library. You assume everyone is just reading. But by listening for the rustling of pages (variability), you realize that a few people are actually whispering secrets (AGN activity) that you couldn't hear before.

5. The "Sleeping" Monsters (LINERs)

They also looked at a specific group called LINERs. These are galaxies that look like they have weak AGNs.

• The Result: Almost none of them showed the "wiggling" behavior.
• The Meaning: This suggests these monsters are either very weak or, more likely, they have lost their "dusty coat" (torus). Without the dust, they don't glow in infrared the way big, active monsters do. It's like a dragon that has lost its fire; it's still a dragon, but it doesn't breathe fire anymore.

6. The "Co-Evolution" Connection

The study found a fascinating link: The more "star-forming" a galaxy is (the more babies it's having), the more likely it is to have a wiggling AGN.

• The Metaphor: It's like a family dinner. The more food the family eats (star formation), the more the head of the house (the black hole) eats too. They grow and evolve together. This supports the idea that galaxies and their central black holes are best friends who grow up side-by-side.

7. False Alarms? (Supernovae and TDEs)

Could the "wiggling" be caused by something else, like a star exploding (Supernova) or a star getting eaten (Tidal Disruption Event)?

• The Check: The team looked at the light curves of 1,000 random galaxies. They found a few that looked like sudden explosions (transients), but they were rare (less than a few percent).
• The Verdict: The "wiggling" is almost certainly the black hole, not a random explosion.

Summary

This paper teaches us that movement is a better sign of life than color. By watching galaxies dance over a decade, astronomers can find hidden black holes that were previously invisible. It's a powerful new tool that helps us understand that black holes and their host galaxies are deeply connected, growing and changing together throughout the history of the universe.

Technical Summary

Here is a detailed technical summary of the paper "Mid-infrared Variability-based AGN Selection using the Multi-epoch Photometric Data from WISE."

1. Problem Statement

Identifying Active Galactic Nuclei (AGNs) is crucial for understanding the co-evolution of supermassive black holes (SMBHs) and their host galaxies. However, traditional selection methods face significant limitations:

• Optical Spectroscopy: Biased against low-luminosity and highly obscured AGNs (Type 2), which are key to understanding the early Universe and obscured populations.
• Mid-Infrared (MIR) Colors: While effective for dust-obscured AGNs, color-based selection (e.g., the "AGN wedge") suffers from contamination by star-forming (SF) regions and host galaxy starlight, reducing sample purity and completeness.
• X-ray/Radio: X-ray selection is limited by sensitivity to nearby sources, while radio selection is biased toward radio-loud sources (a minority of the population).

There is a lack of systematic validation for MIR variability as a selection method. While optical variability is well-studied, MIR variability properties are less explored, and its efficiency in identifying AGNs across diverse galaxy types (including optically inactive hosts) has not been rigorously quantified with large samples.

2. Methodology

The study utilizes a diverse sample of $\sim1.3$ million galaxies from the MPA-JHU catalog (derived from SDSS DR8), classified into AGNs, Star-Forming (SF) galaxies, Composites, LINERs, and Normal galaxies based on optical emission line ratios (BPT diagram).

Data Sources:

• MIR Photometry: Multi-epoch data from the WISE and NEOWISE missions (W1: 3.4$\mu$m, W2: 4.6$\mu$m, W3: 12$\mu$m) covering a $\sim14$-year baseline.
• Preprocessing:
  • Cross-matching with a 1-arcsec radius.
  • Binning data into 90-day intervals to reduce noise.
  • Removing outliers (high $\chi^2$, large spatial offsets) and applying zero-point corrections for detector temperature drifts.
  • Discarding AllWISE data due to a systematic offset with NEOWISE, relying solely on NEOWISE for the final analysis.
  • Final usable sample: $\sim0.9$ million sources ($\sim59\%$ of the initial sample).

Variability Metrics:
The authors define three independent parameters to identify AGNs:

1. $P_{var}$ (Likelihood of Variability): The probability that observed variance is intrinsic rather than random noise, calculated via the reduced chi-squared statistic ($\chi^2$). A stringent threshold of $P_{var} > 0.99$ is adopted for both W1 and W2 bands.
2. Correlation Coefficient ($r$): The Pearson correlation coefficient between W1 and W2 light curves. Since intrinsic AGN variability should affect both bands simultaneously, a threshold of $r > 0.75$ (with $p < 0.05$) is used to filter out uncorrelated noise/outliers.
3. $P_{AGN}$ (MIR Color Fraction): The fraction of epochs where the W1-W2 color exceeds 0.8 mag. This is used as a supplementary diagnostic but found to be too restrictive for the primary selection.

Selection Groups:

• Group 1: $P_{var} > 0.99$ (W1 & W2).
• Group 2: $P_{var} > 0.99$ (W1 & W2) AND $r > 0.75$.
• Group 3: Group 2 criteria AND $P_{AGN, 0.8} > 0.5$.

3. Key Contributions

• Systematic Validation: This is one of the first studies to systematically validate MIR variability-based AGN selection against a large, spectroscopically classified sample of both active and inactive galaxies.
• Optimized Criteria: The authors establish that combining variability probability ($P_{var}$) with inter-band correlation ($r$) is superior to using either metric alone or relying on MIR colors.
• Detection in Inactive Hosts: The method successfully identifies AGN candidates in galaxies classified as "Normal" or "Star-Forming" by optical spectra, suggesting these hosts harbor obscured or low-luminosity AGNs missed by traditional methods.
• Physical Insights: The study links MIR variability fractions to physical properties (redshift, black hole mass, Eddington ratio, and star formation rate), providing new constraints on AGN torus physics and co-evolution.

4. Key Results

• Recovery Efficiency:
  • Group 1 recovers $\sim34.2\%$ of optically selected AGNs but has lower purity due to potential instrumental artifacts.
  • Group 2 (the recommended method) recovers $\sim28.2\%$ of optically selected AGNs with higher reliability.
  • Group 3 (adding color cuts) drastically reduces recovery to $\sim9.3\%$, confirming that color-based selection misses many AGNs, particularly those where host galaxy light dominates the MIR flux.
• AGN Candidates in Inactive Galaxies:
  • Composites: $\sim11.8\%$ identified as AGN candidates.
  • Star-Forming Galaxies: $\sim3.3\%$ identified.
  • Normal Galaxies: $\sim1.4\%$ identified (drops to $0.14%$when restricted to$z < 0.4$ to match redshift distributions).
  • LINERs: Only $\sim1.6\%$ (Group 2) identified, significantly lower than other AGN types.
• Dependence on Physical Properties:
  • Luminosity/Redshift: The variability fraction ($f_{var}$) increases with AGN brightness ([O III] luminosity) and redshift. The redshift trend is attributed to Malmquist bias and the shifting of WISE filters closer to the peak of hot dust emission ($\sim2-3\mu$m) at higher $z$.
  • Black Hole Mass: Variable sources are more common in higher mass black holes, likely reflecting the correlation between BH mass and AGN luminosity.
  • Eddington Ratio: $f_{var}$ peaks at moderate Eddington ratios and declines at low ratios, consistent with the theory that the dusty torus fails to form at very low accretion rates.
  • Star Formation: $f_{var}$ correlates positively with the deviation from the Star Formation Main Sequence ($\Delta_{SFMS}$), suggesting coordinated evolution between SMBH growth and star formation.
• LINER Anomaly: Classical LINERs show extremely low MIR variability and lie outside the AGN wedge. This supports the hypothesis that they host low-luminosity AGNs lacking a dusty torus (due to insufficient outflow to sustain it) or have non-AGN origins.
• Contamination: Transient events (Supernovae, TDEs) were visually inspected in 1,000 variable SF galaxies. Only 11 transient-like light curves were found ($<1\%$), indicating minimal contamination from non-AGN transients.

5. Significance

This study demonstrates that MIR variability is a robust, complementary tool for AGN selection, particularly for:

1. Uncovering Obscured/Weak AGNs: It detects AGN activity in galaxies where optical emission lines are weak or overwhelmed by star formation.
2. Overcoming Color Bias: It avoids the incompleteness of MIR color selection, which fails for AGNs with significant host galaxy contamination.
3. Evolutionary Studies: The correlation between MIR variability and star formation activity provides strong evidence for the co-evolution of SMBHs and their host galaxies.
4. Torus Physics: The distinct behavior of LINERs (low variability, no torus signature) offers observational support for disk-wind models of torus formation that predict the torus disappears at low accretion rates.

The authors conclude that while the method does not recover 100% of AGNs (completeness is $\sim28\%$ for optically selected AGNs), it offers a unique and necessary perspective for building a complete census of the AGN population, especially in the local Universe where multi-epoch MIR data is becoming increasingly available (e.g., for future missions like SPHEREx).