Kinematic reconstruction of atmospheric neutrino events in a large water Cherenkov detector with proton identification

M. Fechner et al. (Super-Kamiokande Collaboration)

Phys. Rev. D 79, 112010 – Published 22 June 2009

Abstract

We report the development of a proton identification method for the Super-Kamiokande (SK) detector. This new tool is applied to the search for events with a single proton track, a high purity neutral current sample of interest for sterile neutrino searches. After selection using a neural network, we observe 38 events in the combined SK-I and SK-II data corresponding to 22 85.1 days of exposure, with an estimated signal-to-background ratio of 1.6 to 1. Proton identification was also applied to a direct search for charged-current quasielastic (CCQE) events, obtaining a high precision sample of fully kinematically reconstructed atmospheric neutrinos, which has not been previously reported in water Cherenkov detectors. The CCQE fraction of this sample is 55%, and its neutrino (as opposed to antineutrino) fraction is $91.7 \pm 3 %$ . We selected $78 μ$ -like and 47 $e$ -like events in the SK-I and SK-II data set. With this data, a clear zenith angle distortion of the neutrino direction itself is reported in a sub-GeV sample of $μ$ neutrinos where the lepton angular correlation to the incoming neutrino is weak. Our fit to $ν_{μ} \to ν_{τ}$ oscillations using the neutrino $\frac{L}{E}$ distribution of the CCQE sample alone yields a wide acceptance region compatible with our previous results and excludes the no-oscillation hypothesis at 3-sigma.

15 More

Received 13 January 2009

DOI:https://doi.org/10.1103/PhysRevD.79.112010

Authors & Affiliations

Click to Expand

Article Text (Subscription Required)

Click to Expand

References (Subscription Required)

Click to Expand

Issue

Vol. 79, Iss. 11 — 1 June 2009

Reuse & Permissions

Access Options

Author publication services for translation and copyediting assistance advertisement

Images

Figure 1
Event displays of a Monte-Carlo NC elastic event, with proton momentum of $1490 MeV / c$ (left), and a Monte-Carlo $300 MeV / c$ muon. The proton stopped early, causing a thin ring pattern on the wall. The muon ring is thicker than most proton rings with similar opening angles.
Reuse & Permissions
Figure 2
Probability of hadronic interactions in the water as a function of proton momentum. The clear region shows the fraction of protons that do not interact hadronically in the water. The crosshatched region shows events whose interactions produce only subthreshold secondaries. The region with a vertical hatch pattern corresponds to production of above-threshold charged secondaries, with no $π^{0}$ . The horizontally hatched region shows the amount of $π^{0}$ production.
Reuse & Permissions
Figure 3
$\frac{R_{tot}}{R_{tot}^{\max }}$ as a function of $\frac{L}{L_{\max }}$ for Monte-Carlo batches at different momenta and fixed path lengths in the detector. The solid (red) line shows the best fit to the functional form $a (\exp (- b \frac{L}{L_{\max }}) - 1)$ , with $a = - 1.29$ and $b = 1.5$ . The function allows one to calculate the expected amount of light for any proton pattern.
Reuse & Permissions
Figure 4
Bias (left) and root-mean-square (right) of proton momentum reconstruction, using monochromatic Monte-Carlo.
Reuse & Permissions
Figure 5
Fitted track length for Monte-Carlo protons at 1.3 and $2.0 GeV / c$ . At $2.0 GeV / c$ the protons never reach their maximum path length.
Reuse & Permissions
Figure 6
Average fraction of the maximum path length in water, traveled in water as a function of proton momentum (estimated with NC elastic Monte-Carlo events). The effect of hadronic interactions is clear.
Reuse & Permissions
Figure 7
Visible energy distribution for data and MC single-ring FC events for SK-I and SK-II is shown with no spallation cuts applied (crosses), after the spallation cut (open circles), and after the sparse ring cut (triangles). The corresponding MC distributions are shown (solid line: no cuts; dash-dotted line: only the spallation inefficiency is applied; thick dashed line: the sparse ring cut and spallation inefficiency are applied). Spallation events are present only in the data.
Reuse & Permissions
Figure 8
Distribution of the visible energy, Cherenkov opening angle, and proton pattern likelihood for single-ring events after spallation and sparse ring removal, for SK-I (top row) and SK-II (bottom row). The solid and dashed lines show Monte-Carlo expectation from NEUT and NUANCE, with oscillation reweighting, but no correction for the absolute flux normalization. The hatched histogram shows the fraction of signal events according to NEUT.
Reuse & Permissions
Figure 9
Discriminating variables used as input for the neural network. The expectations from NEUT (solid line) and NUANCE (dashed line) are shown. The hatched histogram shows the fraction of signal events according to NEUT.
Reuse & Permissions
Figure 10
Neural network response for data and MC (solid line: NEUT, dashed line: NUANCE). Left: SK-I, middle: SK-II, right: combined. The MC were oscillated with ( $Δ m^{2} = 2.5 \times 10^{- 3} {eV}^{2}$ , ${sin }^{2} 2 θ = 1.0$ ).
Reuse & Permissions
Figure 11
Top: SK-I neural network response for signal and background, separated according to the main interaction modes. For signal events, the main mode is NC elastic on proton, followed by NC single-pion production where the pion is absorbed. The rest is mostly NC elastic on neutrons where the neutron produced a proton. Background mostly comes from low energy CCQE muon events, and pion tracks from single-pion interactions.
Reuse & Permissions
Figure 12
Up-down asymmetries with statistical errors. In each plot the present measurement is the left-most point, labeled “data.” The middle points labeled Neut and Nuance show the MC expectation including 2-flavor $ν_{μ} \to ν_{τ}$ oscillation, with MC statistical error only. The right-most point labeled “sterile” shows the expectation of the Neut MC assuming oscillations are only $ν_{μ} \to ν_{sterile}$ . The error bars for simulated events correspond to Monte-Carlo statistics.
Reuse & Permissions
Figure 13
Event displays of Monte-Carlo CCQE events, where both the muon (large ring) and the proton (smaller ring) are visible. In the left figure, both tracks were found by the standard fitter (the thin lines show the fit results). In the figure on the right only the lepton (a muon in this example) was fitted; the proton is visible as a smaller ring next to the muon.
Reuse & Permissions
Figure 14
Distribution of analysis variables with cuts for two-ring CCQE event selection in SK-I. The top two plots are two-dimensional $L$ , $P$ distributions for data and MC, shown with the charged pion removal cut; the arrows show the selected domain. In the bottom three plots, the solid and dashed lines show Monte-Carlo expectation from NEUT and NUANCE, with oscillation reweighting, but no correction for the absolute flux normalization. The hatched regions show the contribution from CCQE events.
Reuse & Permissions
Figure 15
Distribution of analysis variables with cuts for two-ring CCQE event selection in SK-II. The top two plots are two-dimensional $L$ , $P$ distributions for data and MC, shown with the charged pion removal cut; the arrows show the selected domain. In the bottom three plots, the solid and dashed lines show Monte-Carlo expectation from NEUT and NUANCE, with oscillation reweighting, but no correction for the absolute flux normalization. The hatched region shows the contribution from CCQE events.
Reuse & Permissions
Figure 16
Distribution of analysis variables with cuts for single-ring CCQE event selection in SK-I. The top two plots are two-dimensional $L$ , $P$ distributions for data and MC, shown with the charged pion removal cut; the arrow shows the selected domain. In the bottom three plots, the solid and dashed lines show Monte-Carlo expectation from NEUT and NUANCE, with oscillation reweighting, but no correction for the absolute flux normalization. The hatched region shows the contribution from CCQE events.
Reuse & Permissions
Figure 17
Distribution of analysis variables with cuts for single-ring CCQE events selection in SK-II. The top two plots are two-dimensional $L$ , $P$ distributions for data and MC, shown with the charged pion removal cut; the arrow shows the selected domain. In the bottom three plots, the solid and dashed lines show Monte-Carlo expectation from NEUT and NUANCE, with oscillation reweighting, but no correction for the absolute flux normalization. The hatched region shows the contribution from CCQE events.
Reuse & Permissions
Figure 18
Reconstructed energy spectra of the CCQE enriched sample, $e$ -like events are on the left and $μ$ -like events on the right. The hatched region shows the CCQE fraction from NEUT. The solid histograms correspond to our nonoscillated Monte-Carlo simulation, while the dashed histograms show the prediction with oscillations assuming $Δ m^{2} = 2.5 \times 10^{- 3} {eV}^{2}$ and ${sin }^{2} 2 θ = 1$ .
Reuse & Permissions
Figure 19
Reconstructed neutrino zenith angle of the CCQE enriched sample, $e$ -like events are on the left and $μ$ -like events on the right. The box histograms correspond to our nonoscillated Monte-Carlo simulation, while the dashed histograms show the prediction with oscillations assuming $Δ m^{2} = 2.5 \times 10^{- 3} {eV}^{2}$ and ${sin }^{2} 2 θ = 1$ .
Reuse & Permissions
Figure 20
The lepton zenith angle of the CCQE enriched sample, $e$ -like events are on the left and $μ$ -like events on the right. The box histograms correspond to our nonoscillated Monte-Carlo simulation, while the dashed histograms show the prediction with oscillations assuming $Δ m^{2} = 2.5 \times 10^{- 3} {eV}^{2}$ and ${sin }^{2} 2 θ = 1$ .
Reuse & Permissions
Figure 21
Distribution of $L / E$ for the combined CCQE enriched $μ$ -like (left) and $e$ -like (right) samples. The continuous lines show the expectation from NEUT without $ν$ oscillations, while the dashed lines show the expectation with $ν_{μ} \to ν_{τ}$ oscillations with $Δ m^{2} = 2.5 \times 10^{- 3} {eV}^{2}$ and ${sin }^{2} 2 θ = 1$ .
Reuse & Permissions
Figure 22
Allowed regions at 68.3% (dotted line), 90% (dashed line) and 99% (solid line) C.L. for $ν_{μ} \to ν_{τ}$ oscillations from the CCQE kinematically reconstructed samples. The star shows the positions of the best fit from [2], while the triangle shows the position of the best fit for this sample in the physical region. These allowed regions are consistent with previously published results and exclude the no-oscillation hypothesis at greater than 3-sigma.
Reuse & Permissions

Physical Review D

covering particles, fields, gravitation, and cosmology