🔭 astrophysics

DETECT: A Pipeline to Quantify Detection Thresholds in Rubin for Nearby Targets Embedded in Bright Host Galaxies

The paper introduces DETECT, a robust pipeline that quantifies detection thresholds and suppresses false positives for pre-supernova variability in bright host galaxies by performing source injection, image subtraction, and forced photometry, as validated on Rubin LSST simulated and real data.

Original authors: Tobias Géron, Maria R. Drout, W. V. Jacobson-Galán, C. D. Kilpatrick

Published 2026-03-24

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Tobias Géron, Maria R. Drout, W. V. Jacobson-Galán, C. D. Kilpatrick

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 you are a detective trying to spot a tiny, flickering firefly in the middle of a massive, blazing bonfire. That is essentially the challenge astronomers face when trying to study the "pre-explosion" moments of dying stars (supernovae).

Here is the story of the paper, broken down into simple concepts and everyday analogies.

The Problem: The Firefly in the Bonfire

For a long time, scientists thought massive stars were quiet and calm right before they exploded. But recently, we've realized that some stars throw a "pre-party"—they have bright outbursts or flares days or weeks before the big explosion.

To catch these pre-party moments, astronomers are using a new, incredibly powerful camera system called Rubin LSST. It's like upgrading from a bicycle headlight to a stadium floodlight. It can see incredibly faint things.

However, there's a catch.
Many of these dying stars live inside bright, nearby galaxies.

The Analogy: Imagine trying to hear a whisper (the star's pre-explosion flare) while standing next to a screaming crowd (the bright host galaxy).
The Glitch: Even the best computer programs designed to subtract the "screaming crowd" from the audio often fail. They get confused by the noise and start "hearing" things that aren't there (false alarms) or missing the whisper entirely.

The Solution: DETECT

The authors built a new tool called DETECT (Detection Efficiency and Threshold Estimation for Characterization of Transients). Think of DETECT not as a better microphone, but as a calibration lab.

Instead of just listening to the audio, DETECT runs a series of "stress tests" to figure out exactly how loud a whisper needs to be to be heard over the noise at that specific spot.

How DETECT works (The "Fake Firefly" Test):

Pick a Spot: It finds a spot in the galaxy that looks just like where the star is (same brightness, same background noise).
Inject a Fake: It digitally paints a "fake firefly" (a fake star) of a known brightness onto that spot.
Run the Subtraction: It runs the standard computer program to subtract the background galaxy.
Check the Result: Did the computer see the fake firefly?
- If the fake firefly was bright, yes.
- If it was dim, maybe not.
Repeat: It does this hundreds of times with different brightness levels to create a "confidence curve."

This tells the astronomers: "Okay, at this specific location in this specific galaxy, we can only trust a detection if the signal is this bright. Anything dimmer is just noise."

The Results: What Did They Find?

The team tested DETECT using simulated data (a practice run) and then applied it to real data from the Rubin Observatory's first release (DP1).

1. The "False Alarm" Zone:
They found that the standard computer programs were very bad at distinguishing real signals from noise when the signal was "medium" loud.

The Metaphor: If the signal-to-noise ratio is between 5 and 10, the standard program is like a paranoid security guard who thinks every shadow is a burglar.
The Fix: DETECT showed that about 36% of the "detections" flagged by the standard system were actually just noise (false positives). However, if the signal was very strong (above 10), the standard system was usually right.

2. The "Deep Limits":
DETECT didn't just find fake signals; it also gave better "upper limits."

The Metaphor: If you don't see a firefly, you want to know how small a firefly could be and still be invisible. Standard tools might say, "We didn't see anything, so it must be smaller than a candle." DETECT says, "Actually, because of the noise here, we didn't see anything, so it must be smaller than a glow-worm."
This is crucial because it helps scientists rule out theories. If a theory says the star must have been this bright, but DETECT says "we would have seen it if it were that bright," and we didn't, then that theory is wrong.

3. Real-World Success:
They applied this to 15 real supernova candidates.

Some turned out to be real pre-explosion flares (great!).
Some were just noise (DETECT saved them from a false alarm).
Some were weird, short-lived events (like a flare from a nearby star or a nuclear explosion in a galaxy center) that the standard system missed or misidentified.

Why Does This Matter?

The Rubin Observatory is about to start its main survey, which will scan the entire sky for years. It will find thousands of supernovae.

If we use the standard tools, we might get confused by the bright galaxies and think we see pre-explosion flares that aren't there, or miss the real ones. DETECT is the quality control manager. It ensures that when we say, "We found a star that flared before it exploded," we are 100% sure it's real and not just an artifact of a bright background.

In a nutshell:
DETECT is a smart tool that helps astronomers stop guessing and start knowing exactly how faint a signal they can trust when looking at the most chaotic, noisy parts of the universe. It turns "I think I saw something" into "I know exactly what I can and cannot see."

1. Problem Statement

The study addresses a critical challenge in time-domain astronomy: the reliable detection of pre-supernova (pre-SN) variability and transient outbursts in targets embedded within bright, nearby host galaxies.

The Core Issue: Standard image subtraction pipelines (like those used by the Rubin Observatory's LSST) often struggle when a transient is located in a bright galaxy. The subtraction process can leave residual artifacts, leading to:
- False Positives: The default Signal-to-Noise Ratio (SNR) threshold of $>5$ frequently flags background noise or subtraction residuals as detections in bright galaxies.
- Inaccurate Upper Limits: Without robust characterization, the non-detection of a transient cannot be confidently translated into a physical upper limit on precursor emission.
Context: As the Rubin LSST survey begins, it will probe much deeper ( $r \approx 24.6$ mag) than current surveys (ATLAS, ZTF). While this allows for the detection of faint precursor emission, the proximity of these targets to Earth means they often reside in bright galaxies where standard pipelines fail.

2. Methodology: The DETECT Pipeline

The authors developed DETECT (Detection Efficiency and Threshold Estimation for Characterization of Transients) to provide robust, location-specific detection thresholds. Unlike standard pipelines that rely on a global threshold, DETECT performs a localized, empirical analysis.

Workflow Steps:

Injection Location Identification: The pipeline identifies regions within the host galaxy that have background flux similar to the target's location. These "injection locations" must be spatially separated to avoid Point Spread Function (PSF) overlap.
Source Injection: Synthetic point sources of known magnitudes are injected into these locations on the science images (not the template).
Image Subtraction & Photometry: The Rubin science pipelines perform image subtraction and forced photometry on the difference images.
Background Correction: To account for subtraction artifacts, the pipeline measures the flux at injection locations before injecting fake sources. This "background contamination" flux is subtracted from the measured flux of injected sources to determine true detectability.
Recovery Curve Modeling: The fraction of recovered sources is plotted against magnitude. A piecewise linear model is fitted to this data to determine specific detection thresholds (e.g., $m_{50\%}$ and $m_{80\%}$ ).
Validation: The pipeline compares the target's measured magnitude against these tailored thresholds to determine if a detection is real or a false positive.

3. Key Contributions

Novel Pipeline: Introduction of DETECT, a tool specifically designed to quantify detection efficiency and calculate reliable upper limits for transients in complex, high-background environments.
Empirical Thresholding: Moving away from static, global SNR thresholds ( $>5$ ) to dynamic, location-specific thresholds derived from synthetic source injection.
Background Contamination Correction: A specific algorithm to estimate and subtract flux artifacts caused by imperfect image subtraction in bright galaxies, preventing false positives.
Open Source: The code is publicly available on GitHub, facilitating community adoption.

4. Results

The authors validated DETECT using Rubin Data Preview 0 (DP0) (simulated data) and applied it to Rubin Data Preview 1 (DP1) (real data).

Validation (DP0 Simulations):

Host Galaxy Brightness Impact: In faint galaxies ( $m_{gal} \gtrsim 16$ mag), the 80% detection threshold behaves as expected ( $\sim 23-24$ mag). However, in bright galaxies ( $m_{gal} \lesssim 13$ mag), the threshold degrades significantly (to $\sim 21$ mag), and the default $SNR > 5$ pipeline incorrectly flags extremely faint sources ( $m \sim 27$ mag) as detections.
Recovery of Real Events: Using a simulated injection of SN 2009ip's light curve into a bright host at 130 Mpc, DETECT successfully recovered the precursor outburst, demonstrating that Rubin LSST + DETECT can detect such events at distances where ATLAS would fail.

Application (DP1 Real Data):

Sample: Analyzed 15 transients in DP1 fields.
False Positive Reduction:
- Of the epochs originally flagged as detections by the default Rubin $SNR > 5$ threshold, ~36% were reclassified as non-detections after DETECT analysis.
- SNR Correlation: Most false positives occurred when the SNR was between 5 and 10. No false positives were found when SNR $> 10$ .
Case Studies:
- SN 2025brs (Type Ia): Located in a bright galaxy. Default pipelines flagged 8 epochs as detections. DETECT confirmed zero significant detections, providing reliable upper limits 7 magnitudes deeper than ATLAS.
- AT 2024ahzi: DETECT identified reliable pre-discovery detections (2 days prior) that were missed by standard analysis, though the nature (precursor vs. early rise) remains under investigation.
- AT 2024aigj: A very short-timescale event (disappearing in 43 mins) was identified, potentially an M-dwarf flare, highlighting the pipeline's sensitivity to transient phenomena.
Upper Limits: DETECT-derived upper limits are typically 0.4 mag shallower (more conservative) than the default Rubin 5 $\sigma$ limits, but they are statistically robust and free from the false positives plaguing the default pipeline in bright galaxies.

5. Significance and Future Outlook

Pre-SN Science: DETECT is crucial for the upcoming Rubin LSST main survey. It enables the confident study of precursor emission, which is essential for understanding the mass-loss mechanisms and progenitor systems of Type IIn and "IIn-like" supernovae.
General Applicability: While developed for pre-SN variability, the pipeline is broadly applicable to any scenario requiring reliable upper limits in uncertain detection environments (e.g., nuclear transients, tidal disruption events).
Template Quality Dependence: The authors note that DETECT's performance is tied to the quality of the template images. As the LSST survey accumulates more data and builds deeper, higher-quality templates, the detection thresholds will improve, and the false positive rate will likely decrease.
Operational Recommendation: The paper advises that for targets in bright galaxies, relying solely on the default $SNR > 5$ threshold is insufficient. DETECT (or similar injection-based methods) should be used to validate detections and establish physical upper limits.

In summary, DETECT provides the necessary statistical rigor to separate real astrophysical signals from pipeline artifacts in the high-contrast environments that will be common in the era of the Rubin Observatory, ensuring that the discovery of faint pre-supernova activity is both reliable and physically meaningful.