RICE limits on the diffuse ultrahigh energy neutrino flux

We present new limits on ultrahigh energy neutrino fluxes above ${10}^{17}\mathrm{eV}$ based on data collected by the Radio Ice Cherenkov Experiment (RICE) at the South Pole from 1999–2005. We discuss estimation of backgrounds, calibration and data analysis algorithms (both online and offline), procedures used for the dedicated neutrino search, and refinements in our Monte Carlo (MC) simulation, including recent in situ measurements of the complex ice dielectric constant. An enlarged data set and a more detailed study of hadronic showers results in a sensitivity improvement of more than 1 order of magnitude compared to our previously published results. Examination of the full RICE data set yields zero acceptable neutrino candidates, resulting in 95% confidence-level model-dependent limits on the flux $E_{\nu }^{2}d\varphi /d{E}_{\nu }<{10}^{-6}\mathrm{GeV}/(\mathrm{c}{\mathrm{m}}^2\mathrm{s}\mathrm{sr})$ in the energy range ${10}^{17}<E_{\nu }<{10}^{20}\mathrm{eV}$. The new RICE results rule out the most intense flux model projections at 95% confidence level.

Figure 1

Reconstructed xy-vertices for transmitter data taken with transmitter located above the indicated AMANDA hole, compared with a small sample of general triggers. True source surface coordinates are as indicated in Table .

Figure 2

Reconstruction of $z$-coordinate of a transmitter as it is lowered into RICE hole B4. For analytic vertex reconstruction, we assume a constant value of index-of-refraction $n=1$.

Figure 3

Recorded rms voltage, for three channels, as a function of time of day, for data taken between 1/4/04 and 6/4/04.

Figure 4

Simulation of $x$-coordinate of reconstructed vertex ($v_x$) vs $y$-coordinate ($v_y$) of reconstructed vertex for “noise” hits generated in Monte Carlo simulations.

Figure 5

August 2000 data, $v_x$, vs $v_y$ vertex distribution, for events having low hit multiplicity ($N_{\mathrm{hit}}=4$) and large time-between-successive-triggers. Events are required to pass initial time-over-threshold requirements.

Figure 6

Real and Imaginary parts of RICE complex effective height (in units of meters), measured for dipole antennas in air.

Figure 7

Locations of all simulated neutrino interactions (points) vs neutrino interactions producing triggers, for $E_{\mathrm{Shower}}=1000\mathrm{PeV}$, separately for electromagnetic (asterisks) vs hadronic showers (crosses), as described in text.

Figure 8

Locations of all simulated neutrino interactions (points) vs neutrino interactions producing triggers, for $E_{\mathrm{Shower}}={10}^5\mathrm{PeV}$, separately for electromagnetic (asterisks) vs hadronic showers (crosses), as described in text.

Figure 9

Scatter plot of ${\chi }^2$ for fits that constrain the origin of the Cherenkov cone to the vertex determined using grid-based vertexing algorithm, compared to the vertex found using the analytic “4-hit” vertex algorithm, for events passing initial time-over-threshold and minimum hit-multiplicity requirements. Recorded voltages are matched against those expected for a neutrino-induced shower generating a Cherenkov cone, with the apex constrained to a given vertex. The cone width is set to the value expected at the peak of the RICE bandpass ($\sim 300\mathrm{MHz}$) We require that either Cherenkov fit satisfy the condition ${\chi }^2/\mathrm{\text{d}}.\mathrm{\text{o}}.\mathrm{\text{f}}.<50$. The more distant the reconstructed vertex, the more difficult it is to discern the curvature of the Cherenkov wavefront, and the smaller the ${\chi }^2$; in such a case, all receivers lie in the same swath of Cherenkov cone.

Figure 10

Time residual vs time-between-successive triggers for MC neutrino simulations, general data triggers, and thermal noise fluctuations. In the absence of an intervening general data trigger, the 10-minute interval between successive forced unbiased events is evident in the figure.

Figure 11

Event surviving up to hand-scan, but rejected on the basis of early hit in channel 11.

Figure 12

Illustration of ray tracing, and possible refractive effects (ray 2), as described in text. Dotted lines indicate possible rays emanating from a neutrino interaction point.

Figure 13

Comparison of test profile for index-of-refraction at South Pole with the current RICE default. Former is used to assess systematic error due to $n(z)$ uncertainties.

Figure 14

Comparison of relative effective volume (EM showers), obtained without ray tracing and “worst-case” (“Test $n(z)$” in previous Figure) variation in index-of-refraction profile, as a function of depth. Differences are due to fraction of volume lying in the “shadow” region.

Figure 15

Hit-time differences with (“$+$”) and without (“$\times$”) ray-tracing corrections for channels 0 and 1 (selected at random).

Figure 16

Effective volume dependence on assumed form of transfer function. Shown is the ratio of effective volumes (note log $y$-scale) calculated using an impedance-matching model for evolving the transfer function in air to ice, as done in the current RICE Monte Carlo simulation, relative to a simple scaling $T_{\mathrm{ice}}^{\prime }(\omega )=n{T}_{\mathrm{air}}(n\omega )$. For $E>1000\mathrm{PeV}$, this ratio plateaus at a value close to unity.

Figure 17

Sample channel gains (2002 data). Smaller calculated gain at low frequencies is due to explicit filtering of low-frequency backgrounds due to the AMANDA experiment, not corrected for in this calculation.

Figure 18

Current effective volume for electromagnetic showers (including LPM effects), and also hadronic showers, with $\pm 1\mathrm{\text{−}}\sigma$ systematic errors, as indicated.

Figure 19

(Top) Upper bounds on total (all flavor) neutrino fluxes for AGN models of PR [73] and MB [74], GZK [60] neutrino models of ESS [61], PJ [62], MSS [64], and KKSS [63], and the topological defect model of PS [75], due to all flavor NC+CC interactions, based on 1999–2005 RICE livetime of about 20500 hrs. Dashed curves are for model fluxes and the thick curves are the corresponding bounds. The energy range covered by a bound represents the central 80% of the event rate. (Bottom) Bounds on diffuse neutrino fluxes from GRBs derived from RICE data. The bounds are for the internal shock [76], afterglow-ISM [77], and afterglow-wind [78] neutrino flux models assuming an isoflavor mixture at the detector; we use updated results [79] for the fluxes. Systematic errors have not been folded into calculation of upper limits.

Figure 20

Left scale: Product of $V_{\mathrm{eff}}\times \mathrm{\text{Livetime}}$, as a function of energy. Given an input flux ($N/({\mathrm{cm}}^2-\sec -\mathrm{sr})$) and a scattering cross section (and inelasticity) for any process ($\nu +N\to \mathrm{\text{electromagnetic shower}}$, $\nu +N\to \mathrm{\text{hadronic shower}}$, $\nu +N\to \mu \mathrm{\text{−}}\mathrm{\text{black}}\mathrm{\text{−}}\mathrm{\text{hole}}\to \mathrm{\text{hadronic shower}}$, etc.), one can derive limits on putative cosmic ray fluxes. The RICE-specific event reconstruction efficiency has already been folded into these curves, and can therefore be directly convolved with a given input flux to yield an expected number of events. Right scale: Exposure ($\mathcal{A}$); see text for details.

Figure 21

Model-independent limits for the RICE experiment superimposed on model-dependent limits. See Appendix  for discussion.