Measurement of the Atmospheric ${\nu }_e$ Spectrum with IceCube

We present a measurement of the atmospheric ${\nu }_e$ spectrum at energies between 0.1 and 100 TeV using data from the first year of the complete IceCube detector. Atmospheric ${\nu }_e$ originate mainly from the decays of kaons produced in cosmic-ray air showers. This analysis selects 1078 fully contained events in 332 days of live time, and then identifies those consistent with particle showers. A likelihood analysis with improved event selection extends our previous measurement of the conventional ${\nu }_e$ fluxes to higher energies. The data constrain the conventional ${\nu }_e$ flux to be $1.3_{-0.3}^{+0.4}$ times a baseline prediction from a Honda’s calculation, including the knee of the cosmic-ray spectrum. A fit to the kaon contribution ($\xi$) to the neutrino flux finds a kaon component that is $\xi =1.3_{-0.4}^{+0.5}$ times the baseline value. The fitted/measured prompt neutrino flux from charmed hadron decays strongly depends on the assumed astrophysical flux and shape. If the astrophysical component follows a power law, the result for the prompt flux is $0.0_{-0.0}^{+3.0}$ times a calculated flux based on the work by Enberg, Reno, and Sarcevic.

Figure 1

Four representative variables in the BDT (see description in the text). The dotted red lines are for CR muon background simulation, while the blue lines are for the atmospheric ${\nu }_e$ simulations. Real data are shown with black circles. The ${\nu }_e$ signal is scaled up by 1000 to show the separation.

Figure 2

BDT score distribution at Level 3. Real data are shown with points while the sum of all Monte Carlo predictions is shown as a solid green line. Atmospheric neutrino predictions are shown with the cyan dotted line (${\nu }_{\mu }$) and the blue solid line (${\nu }_e$). The CR muon background simulation is shown with a red line. The sideband for the final level CR muon estimation is indicated with the magenta vertical band.

Figure 3

Two-dimensional distributions of a reconstructed visible energy as a function of the BDT score are shown for the real data (left), CR muon background (middle), and atmospheric neutrinos (right). The right side of the black vertical line indicates the final event sample. The $z$-axis is the number of events in 332.3 days.

Figure 4

The ${\nu }_e$ effective volume is plotted as a function of energy at different cut levels. The effective volume is the detector volume multiplied by the ratio of the number of selected events to the number of generated events over $4\pi$ steradian. No self-veto correction is applied to the number of events. The relative efficiency change at high energies is mainly due to containment requirements while, at low energies, the change is driven by the strict selections needed to reject backgrounds.

Figure 5

Cascade reconstruction uncertainties as a function of true ${\nu }_e$ energy using ${\nu }_e$ CC interactions. The upper panel shows the energy statistical (only) resolution obtained by comparing the visible energy with reconstructed energy. The lower panel is for zenith angle resolution in degrees. Solid blue lines represent the iterative reconstruction results. Systematic uncertainties (gray bands) add an additional 12% for energy uncertainty and 2 degrees for zenith angle uncertainty.

Figure 6

Reconstructed energy distribution for the baseline best fit. Note that the prompt component is fit to zero.

Figure 7

PID-BDT output distribution for the baseline best fit. The conventional ${\nu }_{\mu }$ CC as a hybrid component and the conventional cascade component, including ${\nu }_e$ and ${\nu }_{\mu }$ CC, are plotted separately. Events with high PID-BDT scores are cascades. The prompt component is fit to zero.

Figure 8

Reconstructed zenith angle for the baseline best fit. The prompt component is fit to zero.

Figure 9

Effects of systematic variations in final level ${\nu }_e$ samples. The upper panel shows the DOM efficiency impact on the reconstructed energy compared to the nominal prediction. The lower panel shows the ice systematic impact due to an increased scattering on the reconstructed zenith angle.

Figure 10

Best fit contours for the conventional flux at 68% C.L. The baseline fit with the astrophysical spectral index $\gamma$ free is shown with a solid blue line. An alternative fit with the index fixed at $\gamma =2.0$ is shown as a dotted line.

Figure 11

For a given astrophysical spectral index ($x$-axis) in the upper panel, the best fit prompt flux (blue line) and its errors (band at 68% C.L.) from the profile likelihood scan are obtained. The bottom panel shows the range of allowed region of the index parameter from the full fit.

Figure 12

The atmospheric ${\nu }_e$ flux result (shown as red filled triangles). Markers indicate the IceCube measurements of the atmospheric neutrino flux while lines show the theoretical models. The black circles and the blue band come from the through-going upward ${\nu }_{\mu }$ analyses [3, 4]. The open triangles show the ${\nu }_e$ measurement with the IceCube-DeepCore data set [2]. The magenta band shows the modified ERS prediction.