Low-energy atmospheric neutrino flux calculation with accelerator-data-driven tuning

This paper presents an improved low-energy atmospheric neutrino flux calculation for Super-Kamiokande analysis that incorporates accelerator-data-driven hadron interaction tuning, resulting in a 5–10% lower flux prediction with reduced and more directly evaluable uncertainties (7–9% below 1 GeV and 5–7% between 1–10 GeV) compared to conventional methods.

The Gist

Imagine the Earth is constantly being pelted by invisible rain made of tiny, high-speed particles called cosmic rays. When these rays hit our atmosphere, they crash into air molecules and create a massive, chaotic chain reaction—a "shower" of new particles. Some of these particles decay and turn into neutrinos, ghost-like particles that can pass through the entire Earth without stopping.

Scientists at the Super-Kamiokande detector in Japan study these neutrinos to learn about the universe. But to understand what they see, they need to know exactly how many neutrinos should be arriving. This is like trying to solve a mystery: if you expect 100 guests at a party but only 80 show up, you need to know if your invitation list was wrong, or if the guests got lost.

The Old Way: Guessing with "Atmospheric Muons"

For years, scientists tried to predict this neutrino "guest list" using a method called muon tuning.

• The Analogy: Imagine you are trying to guess how many people are walking into a building, but you can't see the front door. Instead, you only have a camera in the basement that sees people walking down the stairs. You try to guess the total number of people entering based on how many are leaving the basement.
• The Problem: This works okay for the middle of the day (medium energies), but it's terrible for the early morning or late night (low and high energies). At low energies, the "people" (muons) often stop or decay before they even reach the basement, so your camera sees nothing. At high energies, different types of particles (like kaons) start showing up, which your basement camera doesn't account for. This left a big "fog" of uncertainty around the low-energy neutrinos.

The New Way: The "Accelerator Data" Tune-Up

In this paper, the researchers decided to stop guessing based on the basement camera and instead go to the source. They used data from particle accelerators (giant machines that smash particles together in controlled environments).

• The Analogy: Instead of guessing how many people enter a building by watching the basement, the scientists went to the front door and counted the actual people walking in. They took a massive dataset of exactly what happens when particles crash into air nuclei in a lab.
• The Process: They built a computer simulation of the atmosphere. Then, they took their "lab data" and compared it to their "atmosphere simulation."
  • If the simulation said, "We expect 10 pions here," but the lab data said, "Actually, we only get 9," they applied a weight (a correction factor) to the simulation.
  • Think of it like a recipe. If a recipe says "add 1 cup of sugar," but you know from tasting that it's too sweet, you adjust the recipe to "add 0.9 cups." They did this for millions of different particle collisions to perfect their recipe for neutrino production.

What Did They Find?

1. The Numbers Dropped: With this new, more accurate "recipe," they found that the predicted number of low-energy neutrinos is actually 5% to 10% lower than they thought before.
2. The Uncertainty Shrank: This is the big win. Before, they were very unsure about the low-energy neutrinos (the "fog"). Now, by using the direct lab data, they can say with much more confidence that their prediction is within 7% to 9% of the truth. It's like going from "I think there are about 100 people, give or take 30" to "I think there are 95 people, give or take 5."
3. Consistency: Even though the numbers changed, the new prediction still fits with what they saw in the detector. It just means their previous estimate was slightly too optimistic.

Why Does This Matter?

Getting the low-energy neutrino count right is crucial for some of the biggest mysteries in physics:

• Dark Matter: Scientists are looking for dark matter, but low-energy neutrinos create a "fog" that looks like dark matter. If we don't know exactly how many neutrinos are there, we might mistake them for dark matter.
• Supernovas: There is a faint background glow of neutrinos from ancient exploding stars. To see them, we need to know exactly how much "noise" (background neutrinos) is in the room.
• The Universe's Secrets: Tiny differences in how neutrinos behave (oscillations) could explain why the universe is made of matter instead of antimatter.

The Bottom Line

The researchers took a complex, messy problem (predicting cosmic particle showers) and cleaned it up by using direct, high-quality measurements from particle accelerators. They replaced a shaky guess with a solid measurement. While they found that there are slightly fewer low-energy neutrinos than expected, the most important result is that we now know the numbers much more precisely, clearing the fog and allowing us to look deeper into the secrets of the universe.

Technical Summary

Here is a detailed technical summary of the paper "Low-energy atmospheric neutrino flux calculation with accelerator-data-driven tuning" by Sato et al.

1. Problem Statement

Accurate prediction of atmospheric neutrino flux is critical for neutrino oscillation experiments (e.g., Super-Kamiokande) and searches for rare phenomena like the Diffuse Supernova Neutrino Background (DSNB) and "neutrino fog" in dark matter searches.

• The Core Issue: The primary source of uncertainty in flux prediction is the modeling of hadron interactions within air showers.
• Limitations of Previous Methods: The conventional approach ("$\mu$-tuning") adjusted hadronic interaction models based on atmospheric muon flux observations. While effective for the $1 < E_\nu < 10$GeV region (reducing uncertainty to$\sim7%$), it failed in the **low-energy region ($E_\nu < 1$ GeV)**.
  • Reason: Low-energy muons often decay before reaching ground-level detectors, providing insufficient data to constrain the models.
  • Consequence: This left a large uncertainty in the $0.1$–$1$ GeV range, which is a critical background for DSNB and dark matter searches, and a region sensitive to CP violation and mass hierarchy effects.

2. Methodology

The authors developed a new accelerator-data-driven tuning method to replace or supplement the muon-based tuning.

A. Simulation Framework

• Model: A full 3D Monte Carlo (MC) simulation of air showers tracking all particles.
• Physics Inputs: Primary cosmic rays (based on satellite/balloon data), geomagnetic field (IGRF), air density (NRLMSISE-00), and hadronic interaction generators (JAM for $E < 31$ GeV, DPMJET-III for higher energies).
• Mechanism: The simulation uses pre-tabulated inclusive multiplicity ($f_{mul}$) and momentum distributions ($f_{p-\theta}$) to speed up calculations.

B. The Tuning Strategy

Instead of modifying the interaction model parameters directly, the authors applied event weights to the simulation chain based on accelerator data.

1. Data Selection: Utilized inclusive differential cross-section data from fixed-target accelerator experiments (HARP, BNL E910, NA61/SHINE, NA49, NA56/SPY, NA20) covering beam momenta from 3 to 450 GeV/c.
2. Weight Definition: A weight $w$ was defined as the ratio of the measured differential cross-section to the simulated one:
   $$w = \frac{[E \frac{d^3\sigma}{dp^3}]_{\text{Data}}}{[\sigma_{prod} E \frac{d^3n}{dp^3}]_{\text{MC}}}$$
   This weight corrects the simulation to match experimental data for specific kinematic variables: incident momentum ($p_{in}$), Feynman-$x$ ($x_F$), and transverse momentum ($p_T$).
3. Parameterization:
   • The accelerator data were grouped into six momentum sections (3–5, 5–8, 8–12.3, 12–17.5, 17.5–31, >31 GeV/c).
   • A fitting function ($f_{p+A}$) was constructed for each section, combining:
     • $f_{p+p}$: BMPT parameterization for $p+p$ collisions (symmetric in $x_F$).
     • $f_{nucl}$: A power-law function describing the nuclear mass number ($A$) dependence.
     • $f_{inc}$: A logarithmic function for incident momentum dependence.
   • Isospin relations and quark counting were used to derive cross-sections for unmeasured particle combinations (e.g., neutron-induced interactions).
4. Application: During the MC run, every hadronic interaction vertex in the air shower chain was assigned a weight based on the kinematics of that specific vertex. The total event weight is the product of all vertex weights.

3. Key Contributions

• Novel Tuning Approach: First implementation of a hadron interaction tuning for atmospheric neutrinos driven directly by accelerator cross-section data rather than atmospheric muon observations.
• Phase Space Coverage Analysis: A rigorous evaluation of how well the selected accelerator data cover the phase space relevant to neutrino production. The study identified that while nucleon-induced pion production is well-covered for $1 < E_\nu < 10$GeV, low-energy neutrino production ($E_\nu < 1$GeV) relies heavily on large-angle scattering (negative$x_F$), which was historically poorly covered but is partially addressed by the HARP central TPC data.
• Systematic Uncertainty Quantification: A comprehensive framework for evaluating systematic uncertainties arising from:
  • Measurement errors in accelerator data.
  • Gaps in phase space coverage.
  • Fitting residuals ($\chi^2$).
  • Extrapolation to high energies (Feynman scaling assumptions).
  • Target nucleus ($A$) dependence extrapolation.

4. Results

• Flux Magnitude: The new tuning resulted in a neutrino flux that is 5%–10% smaller than the previous prediction (Ref. [4]) in the low-energy region. However, this new flux remains consistent with the previous prediction within its uncertainty bounds.
• Flux Ratios: The flavor ratios $(\nu_\mu + \bar{\nu}_\mu)/(\nu_e + \bar{\nu}_e)$ and particle/antiparticle ratios ($\bar{\nu}/\nu$) remained consistent with previous predictions (within a few percent).
• Uncertainty Reduction:
  • $E_\nu < 1$ GeV: The flux uncertainty was evaluated to be 7%–9%. This is a significant improvement over the conventional $\mu$-tuning, which could only provide a conservative estimate in this region due to the lack of low-energy muon data.
  • **$1 < E_\nu < 10$GeV:** The uncertainty was evaluated at **5$\sim7%$.
• Dominant Uncertainty Sources:
  • In the low-energy region ($<1$ GeV), the dominant uncertainty comes from the measurement errors of the accelerator data (specifically from HARP and E910).
  • In the high-energy region ($>10$ GeV), the uncertainty increases because the accelerator data used only extend up to 450 GeV/c, requiring extrapolation based on Feynman scaling.

5. Significance

• Improved Physics Reach: By reducing the uncertainty in the $0.1$–$1$ GeV range, this work significantly lowers the background noise for Diffuse Supernova Neutrino Background (DSNB) searches and Direct Dark Matter searches (neutrino fog).
• Oscillation Studies: The refined flux prediction enhances the precision of atmospheric neutrino oscillation analyses, particularly for measuring the CP-violating phase and determining the neutrino mass hierarchy, which have specific signatures in the 0.2–10 GeV range.
• Methodological Advancement: This study demonstrates that accelerator data can effectively constrain atmospheric neutrino flux models, offering a more direct and physically grounded approach than indirect muon tuning.
• Future Outlook: The authors suggest that the uncertainty can be further reduced by obtaining more precise accelerator data at beam momenta below 20 GeV/c and by extending the tuning scope to higher energies (>450 GeV/c) to better constrain the high-energy flux. The method is intended to complement, not replace, muon tuning in future combined analyses.