The Super-K Zenith Dip: How One Plot Changed Neutrino Physics

In June 1998, at the Neutrino conference in Takayama, Japan, the Super-Kamiokande collaboration presented a single plot that changed neutrino physics. The plot showed muon-neutrino event counts as a function of the zenith angle of the arriving neutrino. The rate from above was as expected. The rate from below — neutrinos that had traversed the entire Earth — was suppressed by approximately a factor of two.

This was, beyond reasonable doubt, the signature of muon-neutrino oscillation into some other flavor. The oscillation length was inferred to be of order Earth’s diameter at the typical atmospheric-neutrino energy. The mass-squared difference between the relevant mass eigenstates was approximately $2.5\times 10^{-3}$ eV². Neutrinos, contrary to the original Standard Model assumption of exact massless-ness, have mass.

The 1998 paper that followed this presentation (Fukuda et al., Physical Review Letters 81, 1562) is the foundational publication of modern neutrino physics. Combined with SNO’s later confirmation through solar neutrinos (2001), the discovery led to the 2015 Nobel Prize for Takaaki Kajita and Arthur McDonald.

This post is about the zenith dip — what it shows, why it was decisive, and how a single plot can reshape a field.

What atmospheric neutrinos are

Cosmic-ray protons hit the upper atmosphere, producing showers of secondary particles. The dominant secondary mesons (pions and kaons) decay rapidly into muons and neutrinos: ${\pi }^{\pm }\to {\mu }^{\pm }+{\nu }_{\mu }$

The muons themselves often decay before reaching the ground: ${\mu }^{\pm }\to e^{\pm }+{\nu }_e+{\nu }_{\mu }\text{(or anti-versions)}$

The naive prediction: each cosmic-ray-induced shower produces approximately twice as many muon neutrinos as electron neutrinos (because the pion produces a ${\nu }_{\mu }$ directly and the muon decay produces another ${\nu }_{\mu }$ plus a ${\nu }_e$, giving a 2:1 muon-to-electron ratio).

This “atmospheric ν_μ/ν_e ratio of ~2” is a robust prediction of cosmic-ray-induced neutrino fluxes that depends only weakly on the underlying flux model. By the 1980s, observations from IMB, Kamiokande, and Soudan had begun to find a deficit of muon neutrinos relative to electron neutrinos — the atmospheric neutrino anomaly. The observed ratio was about 1, not 2.

But the deficit alone was not conclusive. It could be due to:

• Mis-estimated atmospheric flux (not enough kaons, too many pions, etc.)
• Detector systematics in distinguishing muon-like from electron-like events
• Statistical fluctuations
• Genuine particle-physics effects (oscillation)

Why the zenith angle matters

The crucial test is the zenith-angle dependence. Neutrinos arriving at a detector from above have travelled only a short distance through the atmosphere — typically 10-20 km. Neutrinos arriving from below have travelled through the entire Earth — up to 12,800 km of rock and iron.

If the muon-neutrino deficit is due to atmospheric flux mismodeling, it should affect downward and upward neutrinos similarly (since the production occurs symmetrically around the Earth). If the deficit is due to detector systematics, it should also be approximately symmetric.

But if the deficit is due to oscillation over a baseline-dependent path, the upward neutrinos should be more suppressed than the downward ones. The oscillation length sets the relevant scale: neutrinos that have travelled less than one oscillation length show no effect; those that have travelled multiple oscillation lengths show full average suppression.

For atmospheric muon neutrinos with energies around 1-10 GeV, the oscillation length corresponding to $\Delta m_{\text{atm}}^2\sim 2\times 10^{-3}$ eV² is approximately 1000 km. So:

• Downward neutrinos (path ~10 km): $L/L_{\text{osc}}\sim 0.01$, no effect
• Horizontal neutrinos (path ~500 km): $L/L_{\text{osc}}\sim 0.5$, partial suppression
• Upward neutrinos (path ~10,000 km): $L/L_{\text{osc}}\sim 10$, full averaged suppression

The expected survival probability: $P({\nu }_{\mu }\to {\nu }_{\mu })=1-{\sin }^2(2{\theta }_{23}){\sin }^2\left(\frac{1.27\Delta m^{2}L}{E}\right)$

For maximal mixing (${\sin }^{2}2{\theta }_{23}\to 1$) and large $L/E$, the survival probability averages to 1/2.

The 1998 plot

Super-Kamiokande, with its 50-kiloton mass and high statistics, produced the definitive zenith-angle measurement in 1998. The data sample included approximately 4,300 fully-contained muon-neutrino events accumulated over 535 days of running.

The plot showed the data points clustered along the oscillation prediction, not along the no-oscillation flat line. The chi-squared comparison strongly favored the oscillation hypothesis. The fitted parameters were:

• ${\sin }^2(2{\theta }_{23})>0.82$ (consistent with maximal mixing)
• $\Delta m_{23}^2\approx 5\times 10^{-3}$ eV² (subsequent measurements have refined this to $\sim 2.5\times 10^{-3}$ eV²)

The statistical significance of the oscillation interpretation was greater than 6$\sigma$ — a discovery in any reasonable sense.

Why no other interpretation worked

Several alternative explanations were considered and ruled out:

Atmospheric flux mismodeling. The flux models had been refined over years and uncertainties were typically 10-20% in absolute terms. But the zenith-angle dependence of the flux was much better understood (a few percent), since it depends mostly on the atmospheric depth profile. The observed up/down asymmetry was much larger than any plausible flux-model uncertainty.

Detector systematics. Super-K’s particle identification (muon-like vs. electron-like) was studied with calibration data and showed no significant zenith-angle dependence. The total event rate as a function of zenith angle was reproduced to high precision in Monte Carlo for the no-oscillation case (when applied to electron-like events, which were not expected to oscillate at the same rate).

Statistical fluctuation. The data sample of ~4,300 events was large enough that the observed asymmetry corresponded to many standard deviations.

Decay or decoherence. Some alternative theoretical mechanisms — neutrino decay, decoherence between mass states, etc. — were considered. They produce different functional forms of the L/E dependence than oscillation. Subsequent analyses by Super-K and others confirmed the oscillation L/E pattern over a wide range, ruling out the alternatives at high significance.

The robustness of the oscillation interpretation grew steadily through 2000-2003 as additional data and analyses accumulated. By the time of SNO’s complementary solar measurement in 2001, the oscillation framework was effectively established.

What the parameters mean

The 1998 measurement gave:

• A mass-squared difference of approximately $2.5\times 10^{-3}$ eV² (modern value)
• A mixing angle near maximal: ${\sin }^2(2{\theta }_{23})\approx 1$

The mixing angle being near maximal is striking. If the leptonic flavor structure had no special pattern, we might expect the mixing to be small (the corresponding quark sector mixing is small — order of degrees). Instead, the atmospheric mixing is essentially as large as it can be — ~45°.

This near-maximal mixing has been a puzzle for theory. The PMNS matrix has structure unlike the CKM matrix, with two of its three angles being large. Various models have been proposed to explain this (discrete flavor symmetries, anarchy in the high-scale Yukawa structure, special boundary conditions). None has emerged as a clear consensus.

The 1998 paper and its impact

The 1998 paper has been cited more than 5,000 times. It is the most-cited paper in modern neutrino physics. Its key result — the existence of neutrino oscillation — opened the door to:

• Subsequent precision measurement of the oscillation parameters (T2K, NOvA, MINOS)
• The hunt for the third mixing angle ${\theta }_{13}$ (Daya Bay, RENO, Double Chooz)
• The CP-violation programme (T2K, NOvA, eventually DUNE)
• The search for sterile neutrinos at LSND-favored parameters (PROSPECT, STEREO, MicroBooNE)
• The mass-ordering question (JUNO, NOvA, DUNE)

Each of these subprogrammes traces back to the 1998 atmospheric discovery. Without the zenith dip, none of them would exist.

A single plot, a paradigm shift

The 1998 zenith plot is one of the rare cases in modern physics where a single graph fundamentally changed our understanding of fundamental particles. Before the plot: neutrinos were (in the standard textbook picture) massless, three-flavor, with no mixing. After the plot: neutrinos have mass, three flavors are mixed in a non-trivial way, and the Standard Model needs modification.

The plot’s power came from its specificity. Earlier hints of the atmospheric anomaly (the deficit) were ambiguous because the deficit alone could be explained in many ways. The zenith-dependent pattern was specific enough that only oscillation could fit. The data forced a specific theoretical conclusion.

This is, in some respects, a model for how particle physics makes progress. Identifying observables that distinguish hypotheses is more important than accumulating raw statistics on observables that don’t. The zenith dependence was a distinguishing observable. It separated the oscillation hypothesis from all the alternative explanations.

Twenty-eight years later, the discovery still feels recent. Neutrino oscillation is now textbook physics, but the 1998 plot remains as compelling visually as it was in the original paper. The data points, the curves, the specific zenith dependence — they tell a story about how the physical world works.

In an era when discoveries often come from massive collaborations and extensive analyses, the 1998 result is a reminder that one well-chosen plot can carry the weight of a paradigm. Whatever else has changed in physics methodology in the past three decades, the power of a clean experimental signature has not.

The Super-K zenith dip — the up-versus-down asymmetry of muon-neutrino events — is among the most consequential plots in modern physics. It marks the moment when the Standard Model’s massless-neutrino assumption became untenable, and the post-Standard-Model neutrino programme began.