arXiv:1005.0035v2 [astro-ph.HE] 12 May 2010
Observation of Ultra-high-energy Cosmic Rays
with the ANITA Balloon-borne Radio Interferometer
S. Hoover1, J. Nam2, P. W. Gorham3, E. Grashorn4, P. Allison3, S. W. Barwick5, J. J. Beatty4, K. Belov1,
D. Z. Besson6, W. R. Binns7, C. Chen8, P. Chen8,11, J. M. Clem9, A. Connolly10, P. F. Dowkontt7,
M. A. DuVernois3,12, R. C. Field11, D. Goldstein5, A. G. Vieregg1, C. Hast11, C. L. Hebert3, M. H. Israel7,
A. Javaid9. J. Kowalski3, J. G. Learned3, K. M. Liewer13, J. T. Link3,14, E. Lusczek12, S. Matsuno3,
B. C. Mercurio4, C. Miki3, P. Mioˇ cinovi´ c3, C. J. Naudet13, J. Ng11, R. J. Nichol10, K. Palladino4, K. Reil11,
A. Romero-Wolf3, M. Rosen3, L. Ruckman3, D. Saltzberg1, D. Seckel9, G. S. Varner3, D. Walz11, F. Wu51
1Dept. of Physics and Astronomy, Univ. of California, Los Angeles,
Manoa, HI 96822.
Lawrence, KS 66045.
& Leung Center for Cosmology and Particle Astrophysics, National Taiwan University,
London, United Kingdom.
Minneapolis, MN 55455.
2Dept. of Physics, Ewha Womans University, Seoul,
3Dept. of Physics and Astronomy, Univ. of Hawaii,
4Dept. of Physics, Ohio State Univ., Columbus,
5Dept. of Physics, Univ. of California, Irvine,
6Dept. of Physics and Astronomy, Univ. of Kansas,
7Dept. of Physics, Washington Univ. in St. Louis,
8Dept. of Physics, Grad. Inst. of Astrophys.,
9Dept. of Physics, Univ. of Delaware, Newark,
10Dept. of Physics and Astronomy, University College London,
11SLAC National Accelerator Laboratory, Menlo Park,
12School of Physics and Astronomy, Univ. of Minnesota,
13Jet Propulsion Laboratory, Pasadena,
14Currently at NASA Goddard Space Flight Center, Greenbelt, MD, 20771.
We report the observation of sixteen cosmic ray events of mean energy of 1.5×1019eV, via radio pulses
originating from the interaction of the cosmic ray air shower with the Antarctic geomagnetic field, a process
known as geosynchrotron emission. We present the first ultra-wideband, far-field measurements of the radio
spectral density of geosynchrotron emission in the range from 300-1000 MHz. The emission is 100% linearly
polarized in the plane perpendicular to the projected geomagnetic field. Fourteen of our observed events are
seen to have a phase-inversion due to reflection of the radio beam off the ice surface, and two additional events
are seen directly from above the horizon.
The origin of ultra-high energy cosmic rays (UHECR) re-
mains a mystery decades after their discovery [1, 2]. Key
to the solution will be increased statistics on events of high
enough energy (≥ 3×1019eV) to elucidate the endpoint of
the UHECR energy spectrum as seen at Earth. The primary
difficulty is the extreme rarity of events at these energies.
Despite steady progress with experiments such as the Pierre
Auger Observatory, there remains room for new methodolo-
gies. Cosmic rays have been detected for decades via im-
pulsive radio geosynchrotron emission [3, 5–16] but until
now not in this crucial energy range, which offers the pos-
sibility of pointing the UHECRs back to their sources. We
present data from the Antarctic Impulsive Transient Antenna
(ANITA)  which represents the first entry of radio tech-
niques into this energy range. We find 16 UHECR events,
at least 40% of which are above 1019eV, and we show com-
pelling evidence of their origin as geosynchrotron emission
from cosmic-ray showers. Our results indicate degree-scale
precision for reconstruction of the UHECR arrival direction,
lending strong credence to efforts to develop radio geosyn-
chrotron detection as a competitive method of UHECR parti-
Geosynchrotronemission arises when the electron-positron
particle cascade initiated by a primary cosmic ray encounters
the Lorentz force in the geomagnetic field. The resulting ac-
celeration deflects the electrons and positrons and they begin
to spiral in opposite directions around the field lines [17, 18].
In air, the particles’ radiation length is of order 40 g cm−2, a
kilometer or less at the altitudes of air shower maximum de-
velopment. Particle trajectories form partial arcs around the
field lines before they lose enough energy to drop out of the
shower. The meter-scale longitudinal thickness of the shower
particle ‘pancake’ is comparable to radio wavelengths below
severalhundredMHz; thus the ensemblebehaviorof all of the
cascade particles yields forward-beamed synchrotron emis-
sion which is partially or fully coherent in the radio regime.
Therefore, the resulting radio impulse power grows quadrat-
ically with primary particle energy, and at the highest ener-
gies, yields radio pulses that are detectable at large distances.
Current systems under development for detection of these ra-
dio impulses are co-locatedwith cosmic-rayparticle detectors
on the ground to aid in cross-calibration [14–16]. They de-
tect showers with primary energies in the 1017−18eV range
because of their limited acceptance. No such system has re-
ported a sample of > 1019eV UHECR events.
The ANITA long-duration balloon payload is launched
50 100 150200 250300350
FIG. 1: An example of interferometric maps of relative correlated
intensity for both Hpol (top) and Vpol (bottom) from event 3623566
which occurred in a region of Antarctica where the geomagnetic in-
clination gave an appreciable Vpol component for the shower radio
emission. The sidelobes are residuals from the relatively sparse sam-
pling of the ANITA interferometer baselines. Such maps are used to
verify the location of the emission source on the Antarctic continent,
and exclude emission that arises from known anthropogenic sources.
from Williams Field near McMurdo Station, Antarctica. It
takes advantage of the stratospheric South Polar Vortex to cir-
cle the Antarctic continent at altitudes of 35-37km while syn-
optically observing an area of ice of order 1.5M km2. During
flight, ANITA recordsall nanosecond-durationradioimpulses
over a 200-1200 MHz radio frequency band. The threshold is
a few times the received power of thermal emission from the
ice, ∼ 10 picoWatts. The direction of detected signals, deter-
mined by pulse-phase interferometric mapping (Fig. 1,),
is localized to an angular ellipse of 0.3◦×0.8◦(elevation ×
azimuth) which is projected back onto the continent to deter-
mine the origin of the pulse. ANITA’s mission is the detection
of ultra-high energy neutrinos via linearly-polarized coherent
radio Cherenkov pulses from cascades the neutrinos initiate
within the ice sheets. Virtually all impulsive signals detected
during a flight are of anthropogenic origin, but such events
can be rejected with high confidence because of their associa-
tion with known human activity, which is carefully monitored
in Antarctica. For its first flight, during the 2006-2007 Aus-
tral summer, ANITA’s trigger system was designed to max-
imize sensitivity to linearly polarized radio pulses, but pur-
posely blinded to the plane of polarization. However, the
entire polarization information – both vertical and horizon-
tal (Vpol and Hpol) – was recorded for subsequent analysis.
Grid East, km
Grid North, km
FIG. 2: Map of locations of detected UHECR events superimposed
on a Radarsat image of relative microwave radar backscatter ampli-
tude of the Antarctic continent. The red diamonds are the reflected-
event locations, the black squares are the two direct-event locations.
The dash-dot line indicates the limit of ANITA’sfield-of-view for the
flight. Note that the portion of ANITA’s field-of-view that includes
the ocean was always covered by sea-ice during the flight.
Since radio pulses of neutrino origin strongly favor vertical
polarization, due to the geometric-optics constraints on the
radio Cherenkov cone as it refracts through the ice surface,
we used the Hpol information as a sideband test for our blind
Our results were surprising: while the neutrino analysis
(Vpol) gave a null result, a statistically significant sample of
6 Hpol events was found initially , and a more sensitive
analysis now yields 16. These events are randomlydistributed
around ANITA’s flight path (Fig. 2), uncorrelated in location
to human activity or to each other, but closely correlated to
each other in their radio pulse profile and frequency spec-
trum (Fig. 3, Top). Their measured planes of polarization are
found in every case to be perpendicular to the local geomag-
netic field (Fig. 4), as expected from geosynchrotron radia-
tion. With two exceptions, the events reconstruct to locations
on the surface of the ice; the two exceptional cases have di-
rectional origins above the horizon, but below the horizontal
(from stratospheric balloon altitudes the horizon is about 6◦
below the horizontal). Earth-orbiting satellites are excluded
as a possible source since the nanosecond radio temporal co-
herence observed is impossible to retain for signals that prop-
agate through the ionospheric plasma, which is highly dis-
persive in our frequency regime. The fourteen below-horizon
eventsare invertedcomparedto the two above-horizonevents,
as expected for specular reflection (Fig. 3, Top). From these
observations we conclude that ANITA detects a signal, seen
in most cases in reflection from the ice sheet surface, which
FIG. 3: Top: overlay of the 16 UHECR event Hpol pulse shapes,
showing the inverted phase for the 14 reflected events (in blue) com-
pared to the two direct events (in red). Inset: Average pulse pro-
file for all events. Bottom: Flux density for both the averaged di-
rect and averaged reflected events. In each case the data are con-
sistent with an exponential decrease with frequency: the fitted co-
efficients of decrease with frequency are (180 ±13 MHz)−1, and
(197±15 MHz)−1, consistent with each other within fit errors. Er-
rors at low frequency (high SNR) are primarily due to systematic
uncertainty in the antenna gains, and to thermal noise statistics at
originates in the earth’s atmosphere and which involves elec-
trical current accelerating transverse to the geomagnetic field.
Such observations are in every way consistent with predic-
tions of geosynchrotron emission from cosmic-ray air show-
ers. In addition, the inherent spectral and time-domain simi-
FIG. 4: Plane of polarization of UHECR events compared to the an-
gleof themagneticfieldlocal totheevent, withtheredlineindicating
the expectation for the Lorentz force. The reflected events are cor-
rected for their surface Fresnel coefficients, and angles are measured
from the horizontal.
larity of our radio pulses, as well as their robust correlation to
geomagneticparameters, suggests that ANITA’s observations,
which are at much greater distance and higher frequency than
prior and current air-shower geosynchrotronobservations, are
less susceptible to near-field fluctuations of radio strength and
plane of polarization. Such issues have been problematic in
this field throughout most of its history.
Our data represent the first broadband measurements of
geosynchrotron emission in the UHF frequency range. The
average observed radio-frequency spectral flux density of the
above- and below-horizonevents, shown in Fig. 3 (Bottom) is
consistent with an exponential decrease with frequency. The
lack of any statistically significant difference in the spectra
for the direct and reflected events indicates that ice rough-
ness is unimportant for the average surface reflection. To es-
timate the electric field amplitude at the source of these emis-
optics treatments developed for synthetic-aperture radar anal-
ysis. Such models use self-affine fractal surface parame-
ters  and Huygens-Fresnel integration over the specular
reflection region to estimate both amplitude loss and phase
distortion from residual slopes or roughness. In our case,
we used digital-elevation models from Radarsat  to esti-
known to a few km precision. In most cases the surface pa-
rameters are found to be smooth, yielding only modest effects
on the reflection amplitude; in a minority of the events, sur-
face parameters were estimated to be rougher, but still within
the quarter-wave-rms Rayleigh criterion for coherent reflec-
tion . Fresnel reflection coefficients were determined us-
ing a mean near-surface index of refraction of n = 1.33, typi-
cal of Antarctic firn.
To estimate the primary energy for the observed events, we
used two independent approaches that determine the ampli-
tude of the radio emission and the mean angular offset of
the observed events. One approach is based on current air-
shower geosynchrotronradio emission simulations developed
for surface arrays [15, 19, 20], and the second approach is
based on data-drivenmaximum-likelihoodmodeling in which
a small set of parameters of a semi-empirical model were it-
eratively fit to the observed characteristics and total number
of the events, given the known UHECR energy spectrum .
The former method had the advantageof extensive work done
to develop full-scale air shower Monte Carlo simulations for
such radio emission; however, the simulations are not directly
relevant to the very different geometry and higher frequency
pared to most ground array observations, and which also in-
volve showers at much larger zenith angles than ground ar-
rays usually observe. The latter data-driven approach used
physically-motivated parameterizations to capture the radio
the data, including amplitude, phase, and frequency-spectral
content for all 16 events, were found to be effective in fit-
ting both the primary energy and observed angular distribu-
tion of the events. Using the ground-based geosynchrotron
models we found no self-consistent solution for the event en-
ergy and mean angular offset, and we conclude that ANITA’s
ulations. Our data-driven approach converged on a solution
which gave estimated event energies as shown in Fig. 5 along
with an overlay of the histogram of the energy distribution
of simulated events seen in reflection. The implications of
the data-driven solution are that the RF signals from these
highlyinclined,distant showers are significantlystrongerthan
predicted by current geosynchrotron models. The mean en-
ergy of the ensemble of reflected events is estimated to be
old of the Greisen-Zatsepin-Kuzmin (GZK) cutoff [26, 27],
which marks the beginning of the absorption edge of UHE-
CRs againstthe cosmicmicrowavebackgroundradiation. The
large asymmetry in the systematic uncertainty is due to the
uncertainty in the angular offset, which tends to strongly bias
toward underestimating the event energy in our models. For
the direct events, the mean energy is lower due to stronger di-
rect signals, but the acceptance – limited to a narrow angular
band around the horizon – is also much lower.
Based on our data-driven semi-empirical approach, we es-
timate a mean angle of observation relative to the true shower
axis of (1.5±0.5)◦. This angular precision is comparable to
that of ground-based cosmic-ray observatories, and adequate
to allow us to map these events back to the sky. The final
error circle is 2◦in diameter after convolving with angular
reconstruction precision and the modest tilts of each event
locale, determined from Radarsat images at 200 m resolu-
−0.3(sys) × 1019eV, approaching the thresh-
FIG. 5: Top: energies of detected UHECR events, with reflected
events in red, direct (above-horizon) events in black, and the sim-
ulated event sample (reflected events only) shown in blue. Bottom:
Map incelestial (α,δ) coordinates of theANITAevents (circles) with
2.0 degree radii, and nearby AGN (grey diamonds) from the V´ eron-
Cetty catalog. The approximate energy for each event is color coded
by the log10of the estimated event energy. ANITA’s exposure is ap-
proximately uniform across the band 5◦> δ > −30◦.
tion . The resulting map is shown in Fig. 5. Our event
positions are uncorrelated to the sky positions of the Auger
Observatory UHECR events, and the ensemble is also uncor-
related to AGN in the nearby universe. This is expected for
events in this energy range where intergalactic magnetic de-
flection is significant. While our sample of UHE events is
significantly smaller than the current totals for the Auger Ob-
servatory , according to our models the acceptance of this
method of UHE detection continues to increase at high en-
ergies, even beyond 1020eV, whereas the acceptance of all
ground-basedUHECR observatories saturate well before this.
Estimates from our simulations indicate that, after optimiza-
tion for UHECR observation, a new 30 day flight of ANITA
could detect a total of several hundred geosynchrotronevents,
with 60-80 above 1019eV, and ∼ 10 above the nominal GZK
cutoff energy. We conclude that a balloon-borne observatory
is viable at the highest cosmic-ray energies, and if the fidelity
ofmodels of the geosynchrotronprocesscontinuesto improve
at the rate it has in recent years, such an approach will be able
to further elucidate possible correlations in cosmic-ray origin
directions as well as the shape of the endpoint of the UHECR
We are grateful to NASA, the US National Science Foun-
dation, the US Dept. of Energy and the Columbia Scientific
Balloon Facility for their generous support of these efforts.
 Pierre Auger Collaboration, et al., Phys. Lett. B685 239-246,
 Nagano, M. & Watson, A. A., Rev. Mod. Phys. 72, 689-732
 Jelley, J. V. et al., Nature 205, 327-328 (1965).
 Jelley, J. V. et al., Nuovo Cimento A46, 649-667 (1966).
 Porter, N. A., Long, C. D., McBreen, B., Murnaghan, D. J. B.
& Weekes, T. C., Phys. Lett. 19, 415-417 (1965).
 Vernov, S. N., Abrosimov, A. T., Volovik, V. D., Zalyubovskii,
I. I. & Khristiansen, G. B., Pis’ma v ZhETF 5, 157-162 (1967).
[Sov. Phys. JETP Letters 5, 126-130 (1967)]
 Barker,P. R., Hazen,
Phys. Rev. Lett. 18, 51-54 (1967).
 Fegan, D. J. & Slevin, P. J., Nature 217, 440-441 (1968).
 Hazen, W. E., Hendel, A. Z., Smith, H., & Shah N. J.,
Phys. Rev. Lett. 22, 35-37 (1969).
 Hazen, W. E., Hendel, A. Z., Smith, H., & Shah N. J., 24,
 Spencer, R. E., Nature 222, 460-461 (1969).
 Fegan, D. J. & Jennings, D. M., Associated with Extensive Air
Showers.Nature 223, 722-723 (1969).
 Allan, H. R., Progress in Elementary Particles and Cosmic
Ray Physics, 10, edited by Wilson, J. G. & Wouthuysen S. G.
(North-Holland, Amsterdam, 1971), 171-304, and references
 Ardouin, D. et al., Astropart. Phys. 31, 192-200 (2009).
 Nehls, S. et al., Nucl. Instrum. Meth. A589, 350-361 (2008).
 LOPES Collaboration, W. D. Apel et al., Astropart. Phys. (in
 Falcke, H. & Gorham, P., Astropart. Phys. 19, 477-494 (2003).
 Suprun,D. A., Gorham, P. W., & Rosner, J. L., Astropart. Phys.
20 157-168, (2003).
 Huege, T. & Falcke, H., Astropart.Phys. 24, 116-136 (2005).
 Huege, T., Ulrich, R., Engel, R. 2007, Astropart. Physics 27,
 Gorham, P. W. et al., Astropart. Phys. 32, 10-41 (2009).
 Gorham, P. W. et. al., Phys. Rev. Lett. 103, 051103 (2009).
 Shepard, M. K. & Campbell, B. A., Icarus 141, 156-171,
 Liu, H., Jezek, K., Li, B., and Zhao, Z.. 2001. Boulder, CO:
National Snow and Ice Data Center. Digital media.
 Barclay, L. (editor),Propagation of Radiowaves, 2nd Ed., (Inst.
Electr. Engineers: London), 2003.
 Greisen, K., Phys. Rev. Lett. 16, 748-750 (1966).
 Zatsepin, G. T. & Kuzmin, V. A., JETP Lett. 4, 78-80
(1966)[Pisma Zh. Eksp. Teor. Fiz. 4, 114-117 (1966)].