🔭 astrophysics

Down going muon rate monitoring in the ANTARES detector

This paper outlines effective criteria for grouping compatible data runs in the ANTARES underwater neutrino detector based on the number of active photomultiplier tubes, ensuring accurate muon rate monitoring across varying environmental conditions and detector configurations.

Original authors: K. Gracheva, M. Anghinolfi, V. Kulikovskiy, E. Shirokov, Y. Yakovenko

Published 2026-05-21

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: K. Gracheva, M. Anghinolfi, V. Kulikovskiy, E. Shirokov, Y. Yakovenko

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 ANTARES detector as a giant, deep-sea camera array sitting 2.5 kilometers underwater off the coast of France. Its job is to catch "ghost particles" called neutrinos. Since neutrinos are hard to catch, the scientists look for a side effect: when a neutrino hits water, it creates a fast-moving particle (a muon) that leaves a trail of light, like a boat leaving a wake.

The detector uses hundreds of "eyes" (called Photomultiplier tubes or PMTs) to see these light trails. The goal is to count how many of these muon trails appear every minute. In a perfect world, this number should be steady, like a metronome ticking at the same speed forever.

The Problem: The Ocean is Noisy

In reality, the ocean is messy. The paper explains that the number of muons the detector "sees" fluctuates wildly—sometimes by a factor of four! Why?

The "Eyes" Blink: Sometimes, the underwater cameras (PMTs) get overwhelmed by natural ocean noise, like glowing plankton (bioluminescence) or electrical glitches. When this happens, the camera shuts off temporarily to protect itself or because its memory buffer gets full.
The Configuration Changes: The detector isn't always set up the same way. Sometimes they use different "filters" (triggers) to decide what data to save, and sometimes they have fewer working cameras than other times.

If you just added up all the muon counts from different days without checking these conditions, your data would be a jumbled mess, like trying to compare the speed of a car on a highway with a car stuck in traffic.

The Solution: Counting the "Active Eyes"

The authors of this paper came up with a clever way to fix this. They realized that the most important factor isn't just the time of day or the weather, but how many cameras were actually working and sending data during a specific run.

They invented a simple rule:

The "Active Eye" Count: They count how many cameras are "awake" (sending more than 1,000 hits) in a given time slice.
The Relationship: They found that the number of muons detected is directly linked to this count. If you have fewer active cameras, you see fewer muons. If you have more, you see more. It's a predictable relationship, almost like a mathematical formula.

The Analogy: The Party Guest List

Think of the detector as a party where you are trying to count how many people (muons) walk through the door.

The Cameras are the bouncers at the door.
The Problem: Sometimes bouncers get tired, go on break, or get distracted by a fire drill (bioluminescence). If you have 10 bouncers, you count 100 people. If only 5 bouncers are working, you might only count 50 people, even if the same number of people actually walked through.
The Fix: Instead of just counting the total people, the scientists say, "Let's group our data based on how many bouncers were working."
- Group A: Days with 50 working bouncers.
- Group B: Days with 60 working bouncers.

By grouping the data this way, they can mathematically "normalize" the results. They can say, "Okay, on the day with 50 bouncers, we saw 50 people. If we had 60 bouncers, we would have seen 60. Let's adjust the numbers so they all look like they happened with 60 bouncers."

The "Galactic Center" Twist

The paper also notes a fun quirk: sometimes they use a special filter to look specifically at the center of our galaxy. Because the Earth rotates, the detector only "sees" the galactic center for part of the day. This causes the muon count to rise and fall in a daily rhythm (like a tide), just because the detector is pointing in the right direction for a few hours. They had to account for this "sidereal day" rhythm to make sure they weren't misinterpreting the data.

The Result: Cleaning the Data

Using this method, the scientists took the messy data from 2009 and sorted it into 9 distinct groups (or "clusters").

They checked how far each day's data was from the "perfect line" of expectation.
If a day's data was too far off (more than 4 standard deviations, or a statistical outlier), they threw it out as "Bad Run."
This only removed about 5.7% of the data, meaning they kept the vast majority of their hard work but ensured it was all comparable.

Conclusion

In short, this paper is a recipe for data quality control. It tells the ANTARES team: "Don't just dump all your muon counts together. First, check how many cameras were working, group your days by that number, and then you can safely add them all up to get a clear, accurate picture of the universe." This allows them to combine years of data to find rare, high-energy neutrinos that would otherwise be hidden in the noise.

Technical Summary: Down Going Muon Rate Monitoring in the ANTARES Detector

Problem Statement
The ANTARES neutrino telescope, located 2,475 meters deep in the Mediterranean Sea, reconstructs muon tracks produced by neutrino interactions using an array of Photomultiplier Tubes (PMTs). While an ideal detector would exhibit a constant muon rate, the actual observed rate of down-going atmospheric muons fluctuates significantly (up to a factor of four) across different data runs. These fluctuations arise from varying environmental conditions (such as bioluminescence), changing detector configurations (specifically the number of active PMTs and sectors), and different trigger conditions. To achieve sufficient statistical power for physics analysis, data from various periods must be summed; however, this requires rigorous criteria to ensure that only compatible runs—those with consistent detection efficiencies—are grouped together.

Methodology
The paper proposes a method to normalize and group data runs based on the "effective number of active PMTs" ($APMT$) rather than raw time or simple sector counts. The methodology proceeds as follows:

Definition of Active PMTs: A PMT is considered "active" within a specific time slice (0.105 s) if its hit count ( $h_i$ ) exceeds 1,000. This threshold accounts for data loss due to High Rate Veto (HRV) conditions or buffer overflows (Xoff) caused by bioluminescence bursts, where PMTs stop sending data to shore despite being physically present.
Modeling the Rate Dependence: The authors analyze the distribution of hits ( $n_{hit}$ ) used in muon reconstruction, applying a cut of $n_{hit} \geq 6$ . They observe that the number of reconstructed muons follows an exponential distribution relative to the number of hits. By modeling the total number of reconstructed muons ( $\mu_N$ ) as a function of active PMTs ( $N$ ), they derive a relationship showing that the muon rate depends exponentially on the number of active PMTs.
Cluster Identification: The data is plotted as muon rate versus $APMT$. This reveals distinct clusters corresponding to different detector configurations (e.g., runs with 49 active sectors vs. 54 active sectors). Within these clusters, the muon rate aligns linearly with the number of active PMTs on a logarithmic scale.
Selection Criteria: To define compatible subsets of runs, the authors propose a procedure:
- Calculate $APMT$ for each run.
- Plot the reconstructed muon rate per minute against $APMT$.
- Separate data into clusters based on the number of active sectors.
- Interpolate a line for each cluster and calculate the distance of each point from this line.
- Accept runs falling within a specific cut (typically $4\sigma$ ) of the distribution of these distances.

Key Results

Correlation Verification: The study confirms a strong correlation between the muon rate and the number of active PMTs. For instance, in a configuration with 54 active sectors, the rate for 590 active PMTs is successfully predicted to be approximately 177 (normalized units) relative to a rate of 200 for 640 active PMTs, using the derived exponential model.
Trigger Impact: The analysis shows that while trigger conditions (such as the Galactic Centre directional trigger) introduce modulation (e.g., a 15% efficiency enhancement when the Galactic Centre is visible), they do not fundamentally alter the clustering based on $APMT$. However, the Galactic Centre trigger does introduce a larger scatter in the data points.
Data Selection Efficiency: Applying this method to 2009 data, the authors identified 9 distinct groups of runs where data could be safely summed. The procedure successfully excluded "bad runs," with the exclusion rate reaching a maximum of 5.7% of the total dataset.

Significance and Claims
The paper claims to provide an "effective method" for data quality control in the ANTARES experiment. Its primary significance lies in offering a quantitative, physics-based criterion to link measured muon rates to detector configurations. By normalizing runs based on the effective number of active PMTs, the collaboration can reliably sum data from different periods to improve statistics without introducing biases from varying environmental or operational conditions. The authors present this not as a theoretical proposal but as a practical procedure already adopted to process the 2009 dataset, ensuring the consistency required for subsequent neutrino physics analyses.