An automated method for planetary nebula detection with SIGNALS: first applications to NGC 4214 and NGC 4449

This paper presents an automated pipeline for detecting planetary nebulae using SIGNALS/SITELLE data, which achieves visual-level accuracy and is applied to NGC 4214 and NGC 4449 to identify new candidates, determine distances via the PN luminosity function, and calculate specific frequencies.

The Gist

Imagine you are a detective trying to find a specific type of rare, glowing firefly in a massive, chaotic forest. This forest is a galaxy, and the fireflies are Planetary Nebulae (PNe)—the beautiful, glowing shells of gas left behind by dying stars.

The problem? The forest is full of other bright things that look like fireflies from a distance: huge, messy clouds of gas (H II regions) and the explosive remnants of dead stars (Supernova Remnants). Finding the real fireflies among this noise is like trying to spot a single candle in a stadium full of fireworks.

This paper is about a team of astronomers who built a super-smart, automated robot detective to find these cosmic fireflies. Here is how they did it, explained simply:

1. The Tool: A "Super-Camera" with a Wide Lens

Usually, astronomers take pictures of galaxies with telescopes that have a narrow field of view (like looking through a straw). To see the whole galaxy, they have to take hundreds of tiny photos and stitch them together, which is slow and tedious.

For this project, they used a special instrument called SITELLE. Think of SITELLE not just as a camera, but as a 3D scanner.

• It takes a picture of the galaxy.
• But instead of just colors, it breaks every single pixel of that image into a rainbow (a spectrum).
• This creates a "data cube" where you can see what the light is made of, not just where it is.
• It has a huge lens (a wide field of view), allowing it to scan the entire galaxy in one go, rather than stitching together tiny puzzle pieces.

2. The Mission: The SIGNALS Survey

The team is part of a big project called SIGNALS. Their main goal is to study how stars are born in different environments. But because their "3D scanner" is so good at spotting glowing gas, they realized they could also use it to hunt for those dying-star fireflies (Planetary Nebulae).

They tested their new robot detective on two specific galaxies: NGC 4214 and NGC 4449. These are "dwarf irregular" galaxies—messy, chaotic, and full of star formation, making them the hardest places to find these fireflies.

3. How the Robot Detective Works

The team wrote a computer program (a pipeline) to do the heavy lifting. Here is the step-by-step process:

• Step 1: The "Flashlight" Search. The robot scans the galaxy looking for anything that glows brightly in a specific color of light (called [O III]). This is like turning on a flashlight that only makes fireflies glow.
• Step 2: The "Voice" Test (Spectroscopy). Just because something glows doesn't mean it's a firefly. A noisy cloud might glow too. The robot listens to the "voice" of the light. Every object has a unique chemical fingerprint. The robot checks if the light matches the specific signature of a planetary nebula and not a messy gas cloud or a supernova.
• Step 3: The "Shape" Test. Real fireflies (PNe) are tiny points of light, like a pinprick. The messy clouds are big and fuzzy. The robot checks the shape. If it's too big or stretched out, it's rejected.
• Step 4: The "Fake Firefly" Training. To make sure the robot isn't missing anything, the team created fake fireflies (mock data) and hid them inside the galaxy images. They ran the robot over these fake images to see how many it found. This told them exactly how good the robot is and where it might be blind (usually in the very bright, crowded center of the galaxy).

4. The Results: A Successful Hunt

The robot worked beautifully.

• In NGC 4214: They found 25 planetary nebulae. 6 of these were brand new discoveries that no one had seen before.
• In NGC 4449: They found 23 planetary nebulae. 13 of these were new discoveries!

They also used the brightness of these fireflies to calculate how far away the galaxies are. It's like knowing how bright a lightbulb is supposed to be; if it looks dim, you know it's far away. Their calculations matched up perfectly with previous measurements, proving their method is accurate.

5. Why This Matters

Before this, finding these objects required astronomers to stare at images for hours, squinting to guess what was what. It was slow and subjective (different people might guess differently).

This new method is:

• Fast: It automates the search.
• Objective: The computer follows strict rules, so everyone gets the same result.
• Scalable: They can now apply this to the rest of the 31 galaxies in their survey.

The Bottom Line:
The team built a high-tech, automated net to catch rare cosmic fireflies in a messy forest. They proved it works by catching dozens of them in two difficult galaxies, including many that were previously hidden. This opens the door to mapping the "graveyards" of stars across the entire universe, helping us understand how galaxies grow and change over time.

Technical Summary

1. Problem Statement

Planetary Nebulae (PNe) are crucial astrophysical tools for tracing galactic kinematics, measuring distances via the Planetary Nebula Luminosity Function (PNLF), and studying stellar populations. However, detecting PNe in star-forming galaxies is challenging due to:

• Contamination: PNe must be distinguished from bright, extended H II regions and Supernova Remnants (SNRs), which share similar emission lines (particularly [O III]$\lambda$5007).
• Survey Limitations: Traditional slitless spectroscopy or narrow-band imaging often lacks the spectral resolution or field of view (FOV) to efficiently survey large areas. Integral Field Spectrographs (IFS) offer a solution but often suffer from small FOVs, requiring complex mosaics.
• Manual Bottlenecks: Previous PN surveys using IFS data (e.g., from the SIGNALS survey) relied heavily on visual inspection of [O III] maps, which is time-consuming, subjective, and difficult to scale to the full sample of 31 galaxies.

The authors aim to develop a fully automated pipeline to detect PNe in Integral Field Spectrograph (IFS) data cubes, specifically utilizing the SIGNALS survey data obtained with SITELLE, and to apply this method to the dwarf irregular galaxies NGC 4214 and NGC 4449.

2. Methodology

The authors developed a multi-stage automated pipeline using data from the SITELLE imaging Fourier transform spectrometer (iFTS) on the Canada-France-Hawaii Telescope (CFHT).

A. Data Processing and Calibration

• Instrument: SITELLE provides a large FOV ($11' \times 11'$) with spectral resolutions of $R=1000$ (SN2 filter: 480–520 nm, covering [O III]) and $R=5000$ (SN3 filter: 651–685 nm, covering H$\alpha$, [N II], [S II]).
• Flux Calibration: Zero-points were re-calibrated using narrow-band HST images (F502N filter) to ensure accuracy, deriving correction factors of $1.00 \pm 0.08$ (NGC 4214) and $0.82 \pm 0.14$ (NGC 4449).
• Velocity Calibration: To correct for instrumental velocity offsets (up to 25 km s$^{-1}$), the authors mapped sky-line (OH) velocities in the SN3 filter. For SN2 (which lacks sky lines), they derived a correction factor by comparing SN2 and SN3 velocities of nearby sources.

B. The Automated Detection Pipeline

The pipeline consists of four main steps:

1. Source Identification: A "detection map" is generated by subtracting a local background spectrum (9$\times$9 pixel box) from the central spectrum (3$\times$3 pixel box) at each spaxel. Local maxima are identified using the DAOFIND algorithm. This method is superior to simple [O III] maps as it highlights compact point sources against extended emission.
2. Spectral Fitting: Emission lines are fitted using ORCS (a Python-based fitting tool). Initial velocity guesses are derived from the brightest sources to ensure convergence. Lines are fitted with a "sincgauss" profile (convolution of the instrument's sinc function and Gaussian Doppler broadening).
3. Contaminant Elimination (BPT Diagrams): Sources are classified using emission-line ratio diagnostic diagrams ([N II]/H$\alpha$ vs. [O III]/H$\beta$ and [S II]/H$\alpha$ vs. [O III]/H$\beta$).
   • Likely PNe: Fall strictly within the PN-only region defined by Kauffmann et al. (2003), Kewley et al. (2001), and Sabin et al. (2013).
   • Possible PNe: Fall within the PN region considering error bars, or have only [O III] detected (common for faint PNe).
   • Rejection: Sources falling in H II or SNR regions are discarded.
4. Morphological Filtering:
   • PSF Mapping: The Point Spread Function (PSF) distortion across the FOV is mapped using GAIA stars.
   • Roundness & Size: Candidates must be spatially unresolved (size $\le 1.3 \times$ PSF) and round (distortion parameter $D \le 0.3$) to distinguish them from extended H II regions.
   • Visual Inspection: A final, minimal visual check is performed only on candidates passing the automated criteria to confirm point-source nature.

C. Completeness and Recovery Tests

To quantify detection limits, the authors generated mock PN catalogues:

• Mock PNe were inserted into real data cubes with spatial distributions following the stellar continuum and magnitudes drawn from a theoretical PNLF.
• The pipeline was run on these mock cubes to determine recovery rates.
• Completeness Limits: Defined as the magnitude where recovery rate $\ge 50\%$.
  • NGC 4214: $m_{5007} = 25.5$
  • NGC 4449: $m_{5007} = 24.7$
• Results showed that recovery is significantly lower in galaxy centers due to bright contaminant emission.

3. Key Contributions

• Automated Pipeline: A robust, largely automated method for PN detection in IFS data that minimizes visual inspection while maintaining accuracy comparable to past visual methods.
• Mock Data Validation: A rigorous completeness analysis using mock PNe inserted into real data, accounting for spatial variations in PSF and background contamination.
• New PN Catalogues: The first automated PN catalogues for NGC 4214 and NGC 4449 derived from SIGNALS data, including new discoveries.
• Distance and Frequency Metrics: Calculation of PNLF distances and specific frequencies ($\alpha$) for these galaxies, including a newly defined $V$-band specific frequency ($\alpha_V$).

4. Results

NGC 4214

• Catalogue: Identified 25 PNe (11 bona fide, 14 possible).
  • Recovered 19 previously known candidates (from Dopita et al. 2010 and Vicens-Mouret et al. 2023).
  • Discovered 6 new PNe.
• Distance: PNLF fitting yields a distance of $3.09^{+0.25}_{-0.46}$ Mpc, consistent with previous estimates.
• Specific Frequency: Calculated $\alpha_{bol}$ and $\alpha_V$ parameters (Table 2).

NGC 4449

• Catalogue: Identified 23 PNe (9 bona fide, 14 possible).
  • Recovered 10 previously known candidates (from Annibali et al. 2017).
  • Discovered 13 new PNe.
• Distance: PNLF fitting yields a distance of $3.91^{+0.33}_{-0.52}$ Mpc, consistent with TRGB measurements.
• Completeness: The sample is complete only up to 0.5 mag from the bright-end cutoff due to central contamination.

General Findings

• Spatial Distribution: Recovered PNe are sparse in the bright central regions of both galaxies, confirming the necessity of completeness corrections.
• Velocity: Velocities were successfully calibrated, though sources with low H$\alpha$ S/N required SN2-based velocities with cross-matching corrections.
• $\alpha$ Parameter: The derived specific frequencies for these blue, star-forming galaxies are slightly lower than theoretical predictions for redder galaxies, highlighting the need for more data on late-type galaxies to resolve the tension between theory and observation regarding PN production rates.

5. Significance

• Scalability: This pipeline enables the efficient processing of the entire SIGNALS survey (31 galaxies), which was previously bottlenecked by manual inspection.
• Complementarity: The method complements narrow-band HST surveys by providing full spectral information (velocities and line ratios) over a much larger FOV, allowing for the study of PN populations in the outer halos of galaxies.
• Future Applications: The authors anticipate that this method will lead to hundreds of new PN discoveries in star-forming galaxies, significantly expanding the sample size for studying the PNLF, galactic distances, and the specific frequency of PNe across different galaxy morphologies and metallicities.
• Methodological Rigor: The use of mock data to define completeness limits and correct specific frequencies sets a new standard for PN surveys in complex, star-forming environments.