arXiv:astro-ph/0512265v2 29 Mar 2006
Constraints on Cosmic Neutrino Fluxes from the ANITA Experiment
S. W. Barwick,1J. J. Beatty,2D. Z. Besson,3W. R. Binns,4B. Cai,5J. M. Clem,6A. Connolly,7D. F. Cowen,8
P. F. Dowkontt,4M. A. DuVernois,5P. A. Evenson,6D. Goldstein,1P. W. Gorham,9C. L. Hebert,9M. H. Israel,4
J. G. Learned,9K. M. Liewer,10J. T. Link,9S. Matsuno,9P. Mioˇ cinovi´ c,9J. Nam,1C. J. Naudet,10R. Nichol,2
K. Palladino,2M. Rosen,9D. Saltzberg,7D. Seckel,6A. Silvestri,1B. T. Stokes,9G. S. Varner,9and F. Wu1
1Department of Physics and Astronomy, University of California at Irvine, Irvine, California
2Department of Physics, Ohio State University, Columbus, Ohio
3Department of Physics and Astronomy, University of Kansas, Lawrence, Kansas
4Department of Physics, Washington University in St. Louis, St. Louis, Missouri
5School of Physics and Astronomy, University of Minnesota, Minneapolis, Minnesota
6Bartol Research Institute, University of Delaware, Newark, Delaware
7Department of Physics and Astronomy, University of California at Los Angeles, Los Angeles, California
8Department of Astronomy and Astrophysics, Pennsylvania State University, University Park, Pennsylvania
9Department of Physics and Astronomy, University of Hawaii at Manoa, Honolulu, Hawaii
10Jet Propulsion Laboratory, Pasadena, California
We report new limits on cosmic neutrino fluxes from the test flight of the Antarctic Impulsive Transient
Antenna (ANITA)experiment, which completed an 18.4day flight of aprototype long-duration balloon payload,
called ANITA-lite, in early 2004. We search for impulsive events that could be associated with ultra-high energy
neutrino interactions in the ice, and derive limits that constrain several models for ultra-high energy neutrino
fluxes and rule out the long-standing Z-burst model. Weset a 90% CL integral flux limit on a pure E−2spectrum
for the energy range 1018.5eV≤ Eν≤ 1023.5eV at E2
νF ≤ 1.6×10−6GeV cm−2s−1sr−1.
Cosmic rays of energy above 3×1019eV are almost cer-
tain to be of extragalactic origin. Their gyroradius far ex-
ceeds that required for magnetic confinement in our galaxy.
At this energy pion photo-production losses on the cosmic
microwave background radiation (CMBR) via the Greisen-
Zatsepin-Kuzmin (GZK ) process limit their propagation
distances to the local supercluster, of order 40 Mpc or less.
However, the neutrinos that result from this process  would
be observable out to the edge of the visible universe. Recent
studies make compelling arguments that input from neutrino
observationswill be necessaryto resolvetheultra-highenergy
cosmic ray (UHECR) problem . Neutrinos are coupled to
the highest energycosmic rays bothas a direct byproduct,and
perhapsas a potentialsourceofthem. Straightforwardreason-
ing indicates there is a required cosmogenic neutrino flux 
with a broad peak in the energy range of 1017−19eV. First,
Lorentz invariance allows transformation of the cross section
for photo-pion production at center-of-mass energies of order
1 GeV, the ∆+-resonance energy, up to GZK energies, a boost
of order 1011. Second, precision measurements of the CMBR
establish its flux density for all cosmic epochs and thus de-
termine the number density of boosted targets for the photo-
pion productionprocess. Third, we applythe standardcosmo-
logical postulate that the cosmic-ray sources are not uniquely
overdense (and hidden) in our local supercluster compared to
the cosmic distribution. Finally, evidence from composition
studies indicates that the UHECRs are hadronic, and thus un-
able to evade interaction with the CMBR, even if they are
as heavy as iron . We conclude that any localized source
of UHECR at any epoch is surrounded by a “GZK horizon”
beyond which the opacity of the CMBR to photo-pion inter-
actions is sufficient to completely attenuate the charged pro-
genitors, yielding pion secondaries which decay to neutrinos.
The intensity of all of these GZK neutrino spheres sums to
a quasi-isotropic cosmogenic neutrino flux, unless any of the
assumptions above is strongly violated, which would in turn
constitute a serious departure from Standard Model physics.
Neutrinos may not only be cosmogenic byproducts, but
could also be closely associated with sources of the UHECR,
though this possibility is far more speculative. If there are
large fluxes of neutrinos at energies of order 1022−23eV, they
can annihilate with Big-Bang relic cosmic backgroundneutri-
nos (Tν∼ 1.9K) in our own Galactic halo via the interaction
ν¯ ν → Z0, the Z-burst process [10, 11, 12, 13]. Decays of the
neutral weak vector boson Z0then yield UHECRs, overcom-
ing the GZK cutoff because of the nearby production. More-
over, Topological Defect (TD) models  postulate a flux of
super-heavy (1024eV) relic particles that decay in our current
epoch and within the Earth’s GZK sphere, yielding both neu-
trinos and UHECR hadrons in the process. Variant versions
of such models, including hypothetical mirror-matter ,
can evade standard bounds to TD models; such variants cur-
rently have the weakest experimental constraints. Limits of
the fluxes of ultra-high energy (UHE) neutrinos can constrain
or eliminate such models as sources for the UHECR. Both
of these classes of neutrino models predict fluxes well above
the current predictions for cosmogenic GZK neutrinos. In all
models, theneutrinofluxesin the 1018−20eV energyrangeare
well below whatcan be observedwith a cubickilometertarget
volume, so detection methods must use larger scales.
The ANITA mission is now completing construction for
a first launch as a long duration balloon payload in 2006.
The mission has a primary design goal of detecting EeV
cosmogenic neutrinos, or providing a compelling flux limit.
ANITA will detect neutrino interactions through coherent ra-
dio Cherenkov emission from neutrino-induced electromag-
lite prototype flew as a piggyback instrument aboard the
Trans-IronGalactic Element Recorder (TIGER) payload. The
payload launched Dec. 18, 2003, and was aloft for 18.4 days,
tica. The payload landed on the ice sheet several hundred km
from Mawson Station (Australia) at an elevation of 2500 m.
ANITA-lite investigated possible backgroundsto neutrino de-
usedbythe full-scaleANITA.Thepayloadoperationwas suc-
cessful, and we have searched for neutrino-induced cascades
among the impulsive events measured. While ANITA-lite did
not have adequate directional resolution to establish with cer-
tainty that the origin of any event was within the ice, the data
quality was sufficient to distinguish events that were consis-
tent with cascades, and exclude events which were not, thus
enabling us to establish flux limits in the absence of candidate
ANITAexploitsa propertyofEMcascadesthat hasbecome
known as the Askaryan effect . During the development of
the EM cascade, selective electron scattering processes lead
to a negative charge asymmetry, inducing strong coherent ra-
dio Cherenkov radiation in the form of impulses with unique
broadband spectral and polarization properties. When a high
energy neutrino showers in the ice sheet, which has a radio
attenuation length Lα≥ 1 km , the resulting impulses can
easily propagate up through the surface to the balloon pay-
load. From balloon altitudes of 37 km, the horizonis at nearly
700 km distance, giving a synoptic view of ∼ 1.5 M km2of
ice, or ∼ 2 M km3volume to a depth of ≃ Lα. ANITA will
consist of a 2π array of dual- polarization antennas designed
to monitor this entire ice target. ANITA-lite flew only two
first-generation ANITA antennas, with a field-of-view cover-
ing about 12% of the 1.5 M km2ice sheet area within its hori-
zonat anytime, butthe≈170,000km2areaofice inviewstill
represents an enormous monitored volume for the uppermost
kilometer of ice to which we were primarily sensitive. This
leads to the strongest current limit on neutrino fluxes within
its energy regime.
The ANITA-lite antennas are dual-linear-polarizationverti-
cal (V) and horizontal (H) quad-ridged horn antennas, sen-
sitive over 230-1200 MHz, over which their angular re-
sponse remains single-mode, with a nearly constant effective
directivity-gainof about9-11dBi. Theantennabeamis some-
37◦in E-plane and H-plane, respectively. The antenna bore-
sights were offset from one another by 1 m lateral separation,
22.5◦in azimuth,and were canted10◦downwardin elevation.
The antennas were designed to retain an off-axis directivity of
≥ 6 dBi at the angles corresponding to the boresights of the
adjacent antennas; thus each antenna retains good sensitivity
to trigger on events that are centered on its nearest neighbor’s
field-of-view. The combined field-of-view of the two anten-
nas taken in coincidence is of order 45◦in azimuth for typical
events, but can be significantly larger for strong impulses.
A block-diagram of the ANITA-lite antenna, trigger, and
GPS time & position
Flight computer &
slow−ADC & DAC
Digitizer & Trigger
FIG. 1: The ANITA-lite system block diagram.
data acquisition system is shown in Fig. 1. The H- and V-
polarizationantennavoltagesarefirst filteredto limit the pass-
bandto 0.2–1.1GHz. A notchfilter (not shown)removesa re-
gion around 400 MHz used by payload telemetry. The signals
are amplified by low-noise amplifiers (LNAs) with approxi-
mately 100 K noise figure, for a net gain of ∼ 62 dB. The
resulting signals, with thermal-noise levels corresponding to
∼ 35 mV rms, are split equally between the digitizers and the
trigger coincidence section.
Coherent Cherenkov emission from showers in solid media
is 100% linearly polarized , and Antarctic ice does not pro-
duce significant depolarizationover the propagationdistances
(∼ 1 km) required for detection of neutrino interactions .
ANITA-lite takes advantage of this characteristic by requir-
ing that any trigger have roughly equal amplitude in left- and
right-circular polarizations (LCP & RCP). This favors sig-
nals with a high degree of linear polarization, and provides
of order a factor of two improvement in rejecting circularly-
polarized backgrounds. The conversion from the H- and V-
polarizations received from the antenna into LCP and RCP is
done by broadband 90◦hybrid-modecombiners.
The trigger system is critical to the sensitivity of a radio
impulse detection system. It initiates digitization of antenna
waveforms based on correlated pulse amplitudes among the
different antenna channels. For ANITA-lite, the trigger re-
quired a 1- to 3-fold coincidence among the four independent
channels (two antennas and two polarizations), where each
channel was required to exceed a power threshold during a
due to ideal thermal noise is nearly Gaussian, and ANITA-
lite was operated with an average threshold corresponding to
4.3σV, whereσV=?k?Tsys?Z∆ν forbandpass-averagedsys-
tem temperature values of ?Tsys? ≈ 700 K during the flight.
Herek is Boltzmann’sconstant, Z =50Ω, and∆ν=800MHz
400500600 700800 900
FIG. 2: (Color online) Frequency dependence of the excess effective
antenna temperature ∆T when pointing to the Sun and the Galactic
Center . The top band is a model of the expected ∆T, with a
width equal to the systematic uncertainties. The lower two bands
give contributions due to galactic and solar emissions, respectively.
The antenna frequency response is folded into the model.
is the system bandwidth.
Calibration of the system gain, timing, and noise temper-
ature was performed by several means. A calibrated noise
diode was coupled to the system between the antenna and the
first bandpass filter. Also, during the first day of the flight,
a pulse generator and transmitter antenna at the launch site
(Williams field, near McMurdo Station) illuminated the pay-
load with pulses synchronized to GPS signals. An onboard
GPS signal synchronously triggered the ANITA-lite system.
These pulses were recorded successfully by the system out to
several hundredkm distance. Timing analysis of these signals
indicates that pulse phase could be estimated to a precision of
150 ps for ≥ 4σ signal-to-noise ratio. Finally, the response
of the antennas to broadband noise from both the Sun and the
Galactic center and plane was determinedby differentialmea-
surements using data when the payload (which rotated slowly
during the flight) was toward or away from a given source.
The results of this analysis are shown in Fig. 2, showing the
spectral response functionwith the variouscontributionsfrom
astronomical sources. The ambient RF noise levels at balloon
float altitudes were found to be consistent with thermal noise
due to the ice at Tef f∼ 250 K and our receiver noise temper-
ature of 300−500 K, which included contributions from the
cables, LNA, connectors, filters, and power limiters. Other
than our own ground calibration signals, we also detected no
sources of impulsive noise that could be established to be ex-
ternal to our own payload. Several types of triggers were in-
vestigatedfor correlationsto knownAntarctic stations, andno
such correlations were found.
ANITA-lite recorded ∼ 113,000 events at an average live-
time of 40% . Of these events, ∼ 87,500 are 3-fold-
coincident triggers considered for data analysis. The remain-
ification. Two independent analyses were performed within
collaboration, both searching for narrow Askaryan-like im-
pulses, in which almost all signal power is delivered within
10 ns about peak voltage, time-coincident in at least two
of four channels. Analysis A primarily relied on matched-
FIG. 3: (Color online) Limits on various models for neutrino fluxes
at EeV to ZeV energies. The limits are: AMANDA cascades ,
RICE , the current work, GLUE , the FORTE satellite ,
and projected sensitivity for the full ANITA. Models shown are
Topological Defects for two values of the X-particle mass , a
TD model involving mirror matter , a range of models for
GZK cosmogenic neutrinos [4, 21, 22], and several models for Z-
bursts [11, 20]. In the Z-burst models plotted as points, the flux is a
narrow spectral feature in energy, and the error-bars shown indicate
the range possible for the central energy and peak flux values.
filtering the data with the expected signal shape and requir-
ing the filtered data to show better signal-to-noise ratio than
the unfiltered data. Analysis B primarily relied on reject-
ing events which show high level of cross-correlation with
known payload-induced noise events. This approach very ef-
ficiently removes the very common repetitive payload noise
events. These constituted about 90% of triggers, with the re-
None of these resembled the expected neutrino signals. Anal-
ysis A determined the signal passing efficiency by tightening
the cuts until the last background noise event was removed,
and found 53% of the simulated signal still passing the cuts.
Analysis B blinded 80% of the data, optimized the cuts with
the other 20% using the modelrejection factor technique,
and found 65% signal efficiency. No data events pass either
of the analyses. In both analyses, the systematic uncertainty
in passing rates was estimated at ∼ 20%.
To estimate the effective neutrino aperture and exposure
for ANITA-lite, two different and relatively mature simula-
tion codes for the full ANITA instrument were modified to
account for the ANITA-lite configuration. These simulations
account for propagationof neutrinos throughearth crust mod-
els, for the various interaction types and neutrino flavors, for
inelasticity, and both hadronic and electromagnetic interac-
tions (including LPM effects ). The shower radio emis-
4 Download full-text
sion is estimated via standard parameterizations which have
been validated at accelerators [7, 26]. Propagation through
the ice uses a frequency- and temperature-dependent model
for Lα, based on data measured at the South Pole . Sur-
face refraction is accounted for using a combination of ray-
and physical-optics. Refracted emission is propagated geo-
metrically to the payload where a detailed instrument model,
based on lab measurements of the spare flight system, is ap-
plied to determine whether a detection occurs.
Based on the treatment described in Refs. [28, 31], the re-
sulting model-independent 90% CL limit on neutrino fluxes
with Standard Model cross-sections  is shown in Fig. 3.
ANITA-lite approaches the highest energy cosmogenic neu-
trino flux model , and now appears to have entirely ex-
cluded the Z-burst model [8, 11, 13] at a level required to
account for the fluxes of the highest energy cosmic rays, as
represented by the two crosses in the figure, with vertical
and horizontal bars indicating the range of allowed model
parameters for this case. Prior limits from the GLUE and
FORTE experiments had constrained most but not all of this
range. Our limits rule out all of the remaining range for two
of the highest standard topological defect models, shown in
Fig. 3, both of which were constrained already by other ex-
periments. We also provide the first experimental limits on
the highest mirror-matter TD model . Table I shows the
expected event totals and limits for several of these models.
The ANITA-lite 90% CL integral flux limit on a pure E−2
spectrum for the energy range 1018.5eV ≤ Eν≤ 1023.5eV is
νF ≤ 1.6×10−6GeV cm−2s−1sr−1.
TABLE I: Expected numbers of events from few UHE neutrino mod-
els, and confidence level of exclusion by ANITA-lite observations.
Model & references
Baseline GZK models [4, 21, 22]
Strong source evolution GZK models [4, 20, 22]
GZK Models that saturate all bounds [20, 22]:
Yoshida et al. 1997, MX= 1016GeV 
Yoshida et al. 1997, MX= 1015GeV 
Berezinsky 2005, Mirror Necklaces 
Fodor et al. (2002): Halo background 
Fodor et al. (2002): Extragalactic background 
Kalashev et al. (2002) 
0.48 to 0.60 38 to 45
Although designed primarily as an engineering test,
ANITA-lite has set the best current limits on neutrino fluxes
above 1019.5eV, improving constraints by more than an or-
der of magnitude over the GLUE results . This demon-
strates the power of the radio Cherenkov technique applied to
the balloon-based observations of the Antarctic ice. Simula-
tions for ANITA shown in Fig. 3, indicate totals of order 5-50
events for the GZK model range shown for 50 days of flight
time, sufficient to detect these model fluxes for the first time.
This work has been supported by NASA. We thank the
Columbia Scientific Balloon Facility and the National Sci-
ence Foundation for their excellent support of the Antarctic
 K. Greisen, Phys. Rev. Lett. 16 (1966) 748; G. T. Zatsepin and
V. A. Kuz’min, JETP Letters 4 (1966) 78.
 V. S. Berezinsky and G. T. Zatsepin, Phys. Lett. 28B (1969)
423; Sov. J. Nucl. Phys. 11 (1970) 111; F. W. Stecker, Astro-
phys. Space Sci. 20 (1973) 47; Ap. J. 238 (1979) 919.
 D. Seckel and T. Stanev, Phys. Rev. Lett. 95 (2005) 141101.
 R. Engel, D. Seckel, and T. Stanev, Phys. Rev. D64 (2001)
 R. Abbasi et al., Astrophys. J. 622 (2005) 910
 G. A. Askaryan, JETP 14 (1962) 441; JETP 21 (1965) 658.
 D. Saltzberg et al., Phys. Rev. Lett. 86 (2001) 2802;
P. W. Gorham et al., Phys. Rev. E62 (2000) 8590.
 B. Eberle et al., Phys. Rev. D70 (2004) 023007
 S. Yoshida et al., Ap. J. 479 (1997) 547.
 O. E. Kalashev et al., Phys. Rev. D65 (2002) 103003.
 Z. Fodor, S. D. Katz, and A. Ringwald, Phys. Rev. Lett. 88
 T. J. Weiler, Astropart. Phys. 11 (1999) 303.
 T. Weiler, Phys. Rev. Lett. 49 (1982) 234.
 V. Berezinsky, Proc.of 11thInt.Workshop NeutrinoTelescopes
(ed. Milla Baldo Ceolin, 2005) p. 339; astro-ph/0509675.
 S. Barwick et al., J. Glaciol. 51 (2005) 231.
 N. D. Hargreaves, J. Glaciol. 21 (1978), 301; C. S. M. Doake,
Geophys. J. Roy. Astr. S. 64 (1981) 539.
 T. Furukawa, K. Kamiyama and H. Maeno, Proc. NIPR Sym-
posium on Polar Meteorology and Glaciol. 10 (1996) 13.
 R. Gandhi, Nucl. Phys. Proc. Suppl. 91 (2000) 453.
 E. Zas, F. Halzen, and T. Stanev, Phys. Rev. D45 (1992) 362.
 O. E. Kalashev, et al.,Phys. Rev. D66, 063004 (2002).
 R. J. Protheroe & P. A. Johnson, Astropart. Phys. 4 (1996) 253.
 C. Aramo et al., Astropart. Phys. 23 (2005) 65.
 This relatively low livetime was caused by a software bug lead-
ing to a slow readout of the GPS time for each event.
 G. C. Hill and K. Rawlins, Astropart. Phys. 19 (2003) 393.
 J. Alvarez–Mu˜ niz and E. Zas, Phys. Lett. B411 (1997) 218.
 P. Gorham et al., Phys. Rev. D72 (2005) 023002.
 M. Ackermann et al., Astropart. Phys. 22 (2004) 127.
 L. A. Anchordoqui et al., Phys. Rev. D66 (2002) 103002.
 I. Kravchenko et al., Astropart. Phys. 20 (2003) 195.
 P. W. Gorham et al., Phys. Rev. Lett. 93 (2004) 041101.
 N. Lehtinen, P. W. Gorham, A. R. Jacobsen, and R. A. Roussel-
Dupr´ e, Phys. Rev. D69 (2004) 013008.
 The band from 230-350 MHz had higher Tsysand is excluded
here due to the low precision of the measurement.