Twenty-four thousand hours of GREENBURST observations with the GBT

This paper presents the results of the first 24,186 hours of the GREENBURST commensal survey using the Green Bank Telescope, which discovered a new 2.2-second pulsar (PSR J0039+5407) and monitored three known repeating FRBs, while also highlighting the challenges of its single-beam design in distinguishing genuine transients from radio frequency interference.

The Gist

Imagine the Green Bank Telescope (GBT) as a giant, incredibly sensitive ear listening to the universe. Usually, this ear is busy listening to specific conversations for specific projects, like studying a particular star or galaxy. But what if, while listening to those specific conversations, the ear could also "eavesdrop" on the background noise to see if there are any other interesting sounds happening?

That is exactly what the GREENBURST project does. It's a "commensal" search, which is a fancy way of saying it piggybacks on other people's observations to hunt for Fast Radio Bursts (FRBs) and pulsars.

Here is a breakdown of their 24,000-hour journey (about 2.7 years of non-stop listening) in simple terms:

1. The Big Hunt: 24,000 Hours of Eavesdropping

The team listened to the sky for over 24,000 hours. To put that in perspective, if you listened to the radio 24 hours a day without sleeping, it would take you nearly three years to reach this amount of data.

During this time, they found:

• 50 Pulsars: These are like cosmic lighthouses—dead stars spinning so fast they beam radio waves at us like a lighthouse beam. They found 49 known ones and one brand new one they had never seen before.
• 3 Known FRBs: These are the "stars" of the show. FRBs are mysterious, super-bright flashes of radio energy from deep space. They found three that were already known to science, proving their system works.
• 1 Mystery Signal (GBP 220718): This is the "ghost in the machine." They found a signal that looked like an FRB, but they aren't sure what it is yet.

2. The New Discovery: PSR J0039+5407

The team found a new pulsar named PSR J0039+5407.

• The Analogy: Imagine a lighthouse that usually shines brightly, but then suddenly decides to take a nap for a few seconds, then wake up, then take another nap.
• The Reality: This pulsar spins very slowly (once every 2.2 seconds) and is about 2 million years old. But the weirdest thing is that it is a "napper." It is nulling, meaning it stops emitting radio waves about 70–80% of the time. It's like a lighthouse that is only on for 20% of the time and off for 80%. This makes it very hard to catch, but the team managed to figure out its rhythm.

3. The Mystery Signal: GBP 220718

This is the most confusing part of the paper.

• The Setup: They found a signal that looked like an FRB. It had a "Dispersion Measure" (a way of measuring how far the signal traveled through space), suggesting it came from a galaxy about 1 billion light-years away.
• The Twist: When they looked closer at the data, they realized the signal was very narrow-band (like a single, thin note on a piano) and had some strange features.
• The Investigation:
  • Is it a satellite? They checked for low-Earth orbit satellites (like Starlink) passing overhead. None were there.
  • Is it a glitch? They ran it through their "lie detector" tests (checking for radio interference). Usually, human-made radio signals look very "jagged" and unnatural. This signal looked surprisingly "smooth" and natural, which usually means it's from space.
  • The Problem: However, when they looked at the data surrounding the main pulse, they found other tiny pulses that didn't seem to travel through space at all (they had zero "distance" markers). This is like hearing a voice from a distant galaxy, but then hearing a whisper right next to your ear that sounds exactly the same.
• The Verdict: They can't decide yet. It might be a brand new type of cosmic signal, or it might be a very clever piece of Earth-based interference that fooled their filters. They need more eyes (more telescopes) to look at it.

4. The Challenge: The "Single Ear" Problem

The paper highlights a major difficulty: The Green Bank Telescope only has one beam (one ear).

• The Metaphor: Imagine you are in a crowded room trying to hear a whisper. If you have two ears, you can tell if the sound is coming from the left or right, or if it's just a cough from the person next to you. But if you only have one ear, it's much harder to tell if a sound is a real whisper from the back of the room or just someone tapping a microphone nearby.
• Because they only have one beam, it is very hard to distinguish between a real cosmic signal and "Radio Frequency Interference" (RFI)—which is just human-made noise like cell phones or satellites.

5. The Future

The team is now using what they learned to:

• Look for more "napping" pulsars.
• Scan nearby galaxies for more FRBs.
• Improve their filters to catch signals that last longer or look different than before.

In a nutshell: The GREENBURST project spent years listening to the universe while doing other work. They found a new, sleepy pulsar, confirmed some known cosmic flashes, and stumbled upon a mystery signal that might be a new type of alien whisper or a very tricky Earthly prank. They are still trying to solve the riddle!

Technical Summary

1. Problem Statement

Fast Radio Bursts (FRBs) are transient radio pulses of unknown origin that serve as probes for fundamental physics and the large-scale structure of the Universe. While dedicated surveys (e.g., CHIME/FRB, ASKAP) are primary drivers of FRB discovery, many radio observatories operate in "commensal" modes, where data collected for other scientific projects can be re-analyzed for transient signals.
The challenge lies in maximizing the scientific return of large facilities like the Green Bank Telescope (GBT) without dedicating exclusive time to FRB searches. Furthermore, distinguishing genuine astrophysical transients from Radio Frequency Interference (RFI) is increasingly difficult due to the proliferation of Low Earth Orbit (LEO) satellites and terrestrial communication systems. The paper addresses the need to validate a real-time, commensal FRB detection pipeline (GREENBURST) and assess its efficacy in discovering new sources and characterizing known ones.

2. Methodology

Observational Setup:

• Instrument: The Green Bank Telescope (GBT) L-band feed (1.4 GHz).
• Backend: The SERENDIP VI backend architecture, sharing a copy of the signal stream.
• Data Specs: 4096 frequency channels covering 960 MHz (centered at 1440 MHz), sampled at 256 $\mu$s with 8-bit precision.
• Duration: 24,186 hours of observation between March 14, 2019, and July 31, 2024.

Data Pipeline & RFI Mitigation:
The pipeline (summarized in Fig. 1) processes data through several stages:

1. Data Ingestion: UDP packets are buffered and written to filterbank format.
2. RFI Excision (Key Innovation): The authors implemented advanced normality tests to distinguish Gaussian astrophysical noise from non-Gaussian anthropogenic signals.
   • Statistical Moments: Utilization of Spectral Kurtosis (4th moment) and Skewness (3rd moment).
   • Outlier Detection: Application of Jarque–Bera and D'Agostino's $K^2$ statistics. Outliers are flagged using the Inter-Quartile Range (IQR) to identify samples 4 standard deviations from the median.
   • Multi-scale Analysis: Analysis of 64-sample (16 ms) and 4096-sample (1 s) blocks, plus a 16384-sample (4.2 s) cadence check for non-linear signal chain saturation.
   • Subtraction: A weighted zero-DM subtraction technique (Lazarus et al. 2015) removes broadband RFI by accounting for non-uniform receiver sensitivity.
3. Search & Classification:
   • Heimdall: Searches for dispersed pulses with Dispersion Measures (DM) between 10–10,000 pc cm$^{-3}$.
   • FETCH: A deep-learning model (Fast Extragalactic Transient Candidate Hunter) autonomously classifies candidates as astrophysical or RFI.
   • Human Review: Daily manual inspection of astrophysical candidates via a dedicated Slack channel.

Follow-up Strategy:
Candidates are followed up using the Canadian Hydrogen Intensity Mapping Experiment (CHIME) and the Lovell Telescope to refine timing solutions and verify repeatability.

3. Key Contributions

• Pipeline Validation: Demonstrated the robustness of the GREENBURST pipeline over a massive dataset (24k+ hours), successfully integrating deep learning (FETCH) with statistical RFI mitigation.
• Discovery of PSR J0039+5407: Identified a previously unknown Galactic pulsar with a 2.2-second period, characterized by extreme nulling behavior.
• Case Study on Ambiguous Signals: Provided a rigorous technical case study of "GBP 220718," a pulse that mimicked FRB properties but exhibited characteristics of terrestrial interference, highlighting the limitations of single-beam observations.
• Infrastructure Development: Detailed the evolution of the data analysis pipeline since its initial commissioning, specifically the implementation of higher-order statistical tests for RFI removal.

4. Results

A. Pulsar Detections:

• Total: 50 pulsars detected (49 previously known, 1 new).
• New Discovery (PSR J0039+5407):
  • Period: 2.2 s.
  • DM: ~73 pc cm$^{-3}$.
  • Distance: ~3.0 kpc (NE2025 model).
  • Characteristics: High nulling fraction ($NF \approx 73\text{--}80\%$), characteristic age of 2.3 Myr, and magnetic field strength of $6.0 \times 10^8$ T.
  • Significance: The high nulling fraction correlates with its long period and age, fitting theoretical models of pulsar emission states.

B. FRB Detections:

• Known Sources: Three repeating FRBs (FRB 20121102A, FRB 20190520B, FRB 20200120E) were detected. These served as validation for the pipeline's ability to make serendipitous detections during other GBT projects.
• GBP 220718 (The Anomaly):
  • Initial Detection: A 16 ms pulse with DM = 145.5 pc cm$^{-3}$ and S/N = 10.
  • Follow-up: No repetitions were found in 8 hours of Lovell Telescope observations or 150 hours of CHIME/FRB monitoring.
  • Re-analysis: Offline analysis revealed additional pulses in the same narrow frequency band. Crucially, many of these secondary pulses showed zero DM (consistent with terrestrial origin) and complex spectral structures not typical of astrophysical FRBs.
  • Statistical Analysis: Unlike typical RFI, GBP 220718 did not show high Kurtosis or Skew values, allowing it to pass initial filters. This suggests it may be a "stealthy" terrestrial source or a unique astrophysical phenomenon.
  • Conclusion: The origin remains ambiguous; it is likely terrestrial but cannot be definitively ruled out as celestial without multi-beam confirmation.

C. Sky Coverage:

• The survey covered significant portions of the sky, with the L-band receiver in focus 72% of the time.

5. Significance

• Efficiency of Commensal Science: The paper proves that piggybacking on standard GBT operations is a highly effective strategy for transient astronomy, yielding significant discoveries (new pulsars, FRB monitoring) without dedicated telescope time.
• Advancement in RFI Rejection: The implementation of multi-moment normality tests (Kurtosis + Skew) and weighted zero-DM subtraction sets a new standard for filtering complex RFI environments, particularly in the era of LEO satellite constellations.
• Characterization of Nulling: The detailed study of PSR J0039+5407 provides valuable data on the relationship between pulsar age, period, and nulling behavior, contributing to pulsar emission physics.
• Cautionary Tale for Single-Beam Arrays: The GBP 220718 case study underscores a critical limitation: single-beam telescopes struggle to distinguish between narrow-band terrestrial interference and genuine FRBs when the interference does not exhibit standard non-Gaussian statistical moments. This highlights the necessity of interferometric (multi-beam) follow-up for ambiguous detections.
• Future Outlook: The authors plan to extend the search to longer pulse widths to detect slow transients and continue monitoring nearby galaxies, leveraging the improved pipeline sensitivity.