![Cosmic-ray all-electron spectrum measured by CALET from 10 GeV to 3 TeV, where systematic errors (not including the uncertainty on the energy scale) are drawn as a gray band. The measured all-electron flux including statistical and systematic errors is tabulated in Ref. [6]. Also plotted are measurements in space [35-37] and from ground-based experiments [38,39].](https://www.researchgate.net/profile/Paolo-Maestro-3/publication/320794809/figure/fig4/AS:557602012893184@1509954087881/Cosmic-ray-all-electron-spectrum-measured-by-CALET-from-10-GeV-to-3-TeV-where-systematic_Q320.jpg)
Content uploaded by Paolo Maestro
Author content
All content in this area was uploaded by Paolo Maestro on Nov 06, 2017
Content may be subject to copyright.
Energy Spectrum of Cosmic-Ray Electron and Positron from 10 GeV to 3 TeV Observed
with the Calorimetric Electron Telescope on the International Space Station
O. Adriani,1,2 Y. Akaike,3,4 K. Asano,5Y. Asaoka,6,7,* M. G. Bagliesi,8,9 G. Bigongiari,8,9 W. R. Binns,10 S. Bonechi,8,9
M. Bongi,1,2 P. Brogi,8,9 J. H. Buckley,10 N. Cannady,11 G. Castellini,12 C. Checchia,13,14 M. L. Cherry,11 G. Collazuol,13,14
V. Di Felice,15,16 K. Ebisawa,17 H. Fuke,17 T. G. Guzik,11 T. Hams,3,18 M. Hareyama,19 N. Hasebe,6K. Hibino,20
M. Ichimura,21 K. Ioka,22 W. Ishizaki,5M. H. Israel,10 A. Javaid,11 K. Kasahara,6J. Kataoka,6R. Kataoka,23 Y. Katayose,24
C. Kato,25 N. Kawanaka,26,27 Y. Kawakubo,28 H. S. Krawczynski,10 J. F. Krizmanic,18,3 S. Kuramata,21 T. Lomtadze,29,9
P. Maestro,8,9 P. S. Marrocchesi,8,9 A. M. Messineo,29,9 J. W. Mitchell,4S. Miyake,30 K. Mizutani,31 A. A. Moiseev,32,18
K. Mori,6,17 M. Mori,33 N. Mori,2H. M. Motz,34 K. Munakata,25 H. Murakami,6S. Nakahira,35 J. Nishimura,17
G. A. de Nolfo,36 S. Okuno,20 J. F. Ormes,37 S. Ozawa,6L. Pacini,1,12,2 F. Palma,15,16 P. Papini,2A. V. Penacchioni,8,38
B. F. Rauch,10 S. B. Ricciarini,12,2 K. Sakai,18,3 T. Sakamoto,28 M. Sasaki,18,32 Y. Shimizu,20 A. Shiomi,39 R. Sparvoli,15,16
P. Spillantini,1F. Stolzi,8,9 I. Takahashi,40 M. Takayanagi,17 M. Takita,5T. Tamura,20 N. Tateyama,20 T. Terasawa,35
H. Tomida,17 S. Torii,6,7,41,†Y. Tsunesada,42 Y. Uchihori,43 S. Ueno,17 E. Vannuccini,2J. P. Wefel,11 K. Yamaoka,44
S. Yanagita,45 A. Yoshida,28 K. Yoshida,46 and T. Yuda5,‡
(CALET Collaboration)
1Department of Physics, University of Florence, Via Sansone, 1–50019 Sesto, Fiorentino, Italy
2INFN Sezione di Florence, Via Sansone, 1–50019 Sesto, Fiorentino, Italy
3of Physics, University of Maryland, Baltimore County, 1000 Hilltop Circle, Baltimore, Maryland 21250, USA
4Astroparticle Physics Laboratory, NASA/GSFC, Greenbelt, Maryland 20771, USA
5Institute for Cosmic Ray Research, The University of Tokyo, 5-1-5 Kashiwa-no-Ha, Kashiwa, Chiba 277-8582, Japan
6Research Institute for Science and Engineering, Waseda University, 3-4-1 Okubo, Shinjuku, Tokyo 169-8555, Japan
7JEM Utilization Center, Human Spaceflight Technology Directorate, Japan Aerospace Exploration Agency,
2-1-1 Sengen, Tsukuba, Ibaraki 305-8505, Japan
8Department of Physical Sciences, Earth and Environment, University of Siena, via Roma 56, 53100 Siena, Italy
9INFN Sezione di Pisa, Polo Fibonacci, Largo B. Pontecorvo, 3–56127 Pisa, Italy
10Department of Physics, Washington University, One Brookings Drive, St. Louis, Missouri 63130-4899, USA
11Department of Physics and Astronomy, Louisiana State University, 202 Nicholson Hall, Baton Rouge, Louisiana 70803, USA
12Institute of Applied Physics (IFAC), National Research Council (CNR), Via Madonna del Piano, 10, 50019 Sesto, Fiorentino, Italy
13Department of Physics and Astronomy, University of Padova, Via Marzolo, 8, 35131 Padova, Italy
14INFN Sezione di Padova, Via Marzolo, 8, 35131 Padova, Italy
15University of Rome “Tor Vergata,”Via della Ricerca Scientifica 1, 00133 Rome, Italy
16INFN Sezione di Rome “Tor Vergata,”Via della Ricerca Scientifica 1, 00133 Rome, Italy
17Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency,
3-1-1 Yoshinodai, Chuo, Sagamihara, Kanagawa 252-5210, Japan
18CRESST and Astroparticle Physics Laboratory NASA/GSFC, Greenbelt, Maryland 20771, USA
19St. Marianna University School of Medicine, 2-16-1, Sugao, Miyamae-ku, Kawasaki, Kanagawa 216-8511, Japan
20Kanagawa University, 3-27-1 Rokkakubashi, Kanagawa, Yokohama, Kanagawa 221-8686, Japan
21Faculty of Science and Technology, Graduate School of Science and Technology, Hirosaki University,
3, Bunkyo, Hirosaki, Aomori 036-8561, Japan
22Yukawa Institute for Theoretical Physics, Kyoto University, Kitashirakawa Oiwakecho, Sakyo, Kyoto 606-8502, Japan
23National Institute of Polar Research, 10-3, Midori-cho, Tachikawa, Tokyo 190-8518, Japan
24Faculty of Engineering, Division of Intelligent Systems Engineering, Yokohama National University,
79-5 Tokiwadai, Hodogaya, Yokohama 240-8501, Japan
25Faculty of Science, Shinshu University, 3-1-1 Asahi, Matsumoto, Nagano 390-8621, Japan
26Hakubi Center, Kyoto University, Yoshida Honmachi, Sakyo-ku, Kyoto 606-8501, Japan
27Department of Astronomy, Graduate School of Science, Kyoto University,
Kitashirakawa Oiwake-cho, Sakyo-ku, Kyoto 606-8502, Japan
28College of Science and Engineering, Department of Physics and Mathematics, Aoyama Gakuin University,
5-10-1 Fuchinobe, Chuo, Sagamihara, Kanagawa 252-5258, Japan
29University of Pisa, Polo Fibonacci, Largo B. Pontecorvo, 3–56127 Pisa, Italy
30Department of Electrical and Electronic Systems Engineering, National Institute of Technology, Ibaraki College,
866 Nakane, Hitachinaka, Ibaraki 312-8508, Japan
31Saitama University, Shimo-Okubo 255, Sakura, Saitama 338-8570, Japan
PRL 119, 181101 (2017) PHYSICAL REVIEW LETTERS week ending
3 NOVEMBER 2017
0031-9007=17=119(18)=181101(6) 181101-1 Published by the American Physical Society
32Department of Astronomy, University of Maryland, College Park, Maryland 20742, USA
33Department of Physical Sciences, College of Science and Engineering, Ritsumeikan University, Shiga 525-8577, Japan
34International Center for Science and Engineering Programs, Waseda University,
3-4-1 Okubo, Shinjuku, Tokyo 169-8555, Japan
35RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
36Heliospheric Physics Laboratory, NASA/GSFC, Greenbelt, Maryland 20771, USA
37Department of Physics and Astronomy, University of Denver,
Physics Building, Room 211, 2112 East Wesley Avenue, Denver, Colorado 80208-6900, USA
38ASI Science Data Center (ASDC), Via del Politecnico snc, 00133 Rome, Italy
39College of Industrial Technology, Nihon University, 1-2-1 Izumi, Narashino, Chiba 275-8575, Japan
40Kavli Institute for the Physics and Mathematics of the Universe, The University of Tokyo,
5-1-5 Kashiwanoha, Kashiwa 277-8583, Japan
41School of Advanced Science and Engineering, Waseda University, 3-4-1 Okubo, Shinjuku, Tokyo 169-8555, Japan
42Division of Mathematics and Physics, Graduate School of Science, Osaka City University,
3-3-138 Sugimoto, Sumiyoshi, Osaka 558-8585, Japan
43National Institutes for Quantum and Radiation Science and Technology, 4-9-1 Anagawa, Inage, Chiba 263-8555, Japan
44Nagoya University, Furo, Chikusa, Nagoya 464-8601, Japan
45College of Science, Ibaraki University, 2-1-1 Bunkyo, Mito, Ibaraki 310-8512, Japan
46Department of Electronic Information Systems, Shibaura Institute of Technology, 307 Fukasaku, Minuma, Saitama 337-8570, Japan
(Received 9 August 2017; revised manuscript received 11 September 2017; published 1 November 2017)
First results of a cosmic-ray electron and positron spectrum from 10 GeV to 3 TeV is presented based
upon observations with the CALET instrument on the International Space Station starting in October, 2015.
Nearly a half million electron and positron events are included in the analysis. CALET is an all-calorimetric
instrument with total vertical thickness of 30 X0and a fine imaging capability designed to achieve a
large proton rejection and excellent energy resolution well into the TeVenergy region. The observed energy
spectrum over 30 GeV can be fit with a single power law with a spectral index of −3.152 0.016
(stat þsyst). Possible structure observed above 100 GeV requires further investigation with increased
statistics and refined data analysis.
DOI: 10.1103/PhysRevLett.119.181101
Introduction.—The Calorimetric Electron Telescope
(CALET) is a Japan-led international mission funded
by the Japanese Space Agency in collaboration with the
Italian Space Agency and NASA [1]. The instrument was
launched on August 19, 2015 by a Japanese carrier H-II
transfer vehicle and robotically installed on the Japanese
Experiment Module-Exposed Facility on the International
Space Station for a two-year mission, extendable to
five years.
The primary science goal of CALET is to perform
high-precision measurements of the cosmic-ray electron
and positron spectrum from 1 GeV to 20 TeV. In the high-
energy TeV region, CALET can observe possible signa-
tures of sources of high-energy particle acceleration in our
local region of the Galaxy [2,3]. In addition, the observed
increase of the positron fraction over 10 GeV by PAMELA
[4] and AMS-02 [5] tells us that at high energy an unknown
primary component of positrons may be present in addition
to the secondary component produced during the galactic
propagation process. Candidates for such primary sources
range from astrophysical ones (e.g., pulsar) to exotic (e.g.,
dark matter). Since these primary sources naturally emit
positron-electron pairs, it is expected that the electron and
positron (hereafter, all-electron) spectrum might exhibit a
spectral structure determined by the origin of positrons.
This may become visible in the high-energy domain of the
spectrum in the case, for instance, of an acceleration limit
from pulsars or the mass of dark matter particles.
CALET instrument.—CALET is an all-calorimetric
instrument with a total vertical thickness equivalent to
30 radiation lengths (X0) and 1.3 proton interaction lengths
(λI) preceded by a charge identification system. The energy
measurement relies on two independent calorimeters: a
fine-grained preshower imaging calorimeter (IMC) fol-
lowed by a total absorption calorimeter (TASC). In order
to identify the individual chemical elements, a charge
detector (CHD) is placed at the top of the instrument.
CALET has several unique and important characteristics
[6]. They include an excellent separation among hadrons
and electrons (∼105) and fine energy resolution (∼2%)
to precisely measure the energy of electrons in the TeV
region. Particle identification and energy measurements are
performed by TASC, the 3 X0thick IMC ensuring proper
development of electromagnetic shower in its initial stage is
used for track reconstruction, and charge identification is
obtained from a CHD.
In Fig. 1, a schematic side view of the instrument is
shown with a simulated shower profile produced by a
1 TeV electron, while an example of a 1 TeV electron
shower candidate in the flight data is shown in Fig. 2.
PRL 119, 181101 (2017) PHYSICAL REVIEW LETTERS week ending
3 NOVEMBER 2017
181101-2
CALET has a field of view of ∼45° from the zenith and an
effective geometrical factor for high-energy (>10 GeV)
electrons of ∼1040 cm2sr, nearly independent of energy.
Data analysis.—We have analyzed flight data (FD)
collected with a high-energy shower trigger [20] in 627 days
from October 13, 2015 to June 30, 2017. The total
observational live time is 12 686 h, and the live time to
total observation time fraction is 84%. On-orbit data
collection has been continuous and very stable.
A Monte Carlo (MC) program was developed to simulate
physics processes and detector signals based on the
simulation package
EPICS
[21] (
EPICS
9.20 and
COSMOS
8.00); it was tuned and tested with accelerator beam test
data, and a detailed detector configuration was imple-
mented. The MC event samples are generated in order to
derive event selection and event reconstruction efficiencies,
energy correction factor, and background contamination.
These samples consist of downgoing electrons and protons
produced isotropically on the surface of a sphere with a
radius of 78 cm which totally encloses the instrument.
Energy measurement.—Energy calibration is a key issue
of CALET as a calorimeter instrument to achieve high
precision and accurate measurements. The method of
energy calibration and the associated uncertainties have
been described elsewhere [22]. Detailed calibration
achieved a fine energy resolution of 2% or better in
the energy region from 20 GeV to 20 TeV (<3% for
10–20 GeV). The validity of our simulation has been
checked with beam test data [23–25]. Regarding temporal
variations occurring during long-term observations, each
detector component is calibrated by modeling variations
of the minimum ionizing particles (MIP) peak obtained
from noninteracting particles (protons or helium) recorded
with a dedicated trigger mode. The rate of change of the
gain decreasing as a function of time is less than 0.5% per
month after one year since the beginning of operations.
Track reconstruction.—As some of the calibrations and
most of the selection parameters depend on the trajectory
of the incoming particle, track recognition is one of the
important steps in data analysis. As a track recognition
algorithm, we adopt the “electromagnetic shower tracking”
[23], which takes advantage of the electromagnetic shower
shape and the IMC design concept. Thanks to optimized
arrangement of tungsten plates between the SciFi layers,
shower cascades are smooth and stable. By using the
preshower core at the bottom of the IMC layers (at depths
of 2 and 3 X0) as initial track candidates, a very reliable and
highly efficient track recognition becomes possible.
Preselection.—In order to minimize and accurately
subtract proton contamination in the sample of electron
candidates, a preselection of well-reconstructed and well-
contained single-charged events is applied. Furthermore, by
removing events not included in MC samples, i.e., particles
with an incident angle from the zenith larger than 90° and
heavier particles, equivalent event samples between FD and
MC calculations were obtained. The preselection consists
of (1) an off-line trigger confirmation, (2) geometrical
condition, i.e., the reconstructed track must traverse the
instrument from the CHD top to the TASC bottom layer,
(3) a track quality cut to ensure reconstruction accuracy,
(4) charge selection using the CHD, and (5) longitudinal
shower development and (6) lateral shower containment
consistent with those expected for electromagnetic
cascades. The combined efficiency of preselection for
electrons is very high: >90% above 30 GeV to 3 TeV,
85% at 20 GeV at variance with only 60% at 10 GeV due to
lower trigger efficiency.
Energy reconstruction.—In order to reconstruct the
energy of primary electrons, an energy correction function
is derived using the electron MC data after preselection.
The energy deposit in the detector is obtained as the sum of
the TASC and IMC, where a simple sum is sufficient for the
TASC, while compensation for energy deposits in tungsten
plates is necessary for the IMC. The correction function is
then derived by calculating the average ratio of the true
energy to the energy-deposit sum in the detector. Because
of near total absorption of the shower, the correction factor
is very small, ∼5%, up to the TeV region.
Electron identification.—The last step of event selection
is electron identification exploiting the shower shape
FIG. 1. A schematic side view of the main calorimeter. An
example of a simulated 1 TeV electron event is superimposed to
illustrate the shower development in the calorimeter.
-50
-40
-30
-20
-10
0S
H
S L H
S L H
S L H
S L H
S L H
X-Z View
-30 -20 -10 0 10 20 -20 -10 0 10 20 30
-50
-40
-30
-20
-10
0S
H
S L H
S L H
S L H
S L H
Single
Low
High
Heavy-ion Single
Heavy-ion Low
Heavy-ion High
External Trigger
Pedernal Trigger
Y-Z View
Event ID:60308
-1
10
1
10
2
10
3
10
4
10
MIP
FIG. 2. An example of a 1 TeV electron shower candidate in
flight data.
PRL 119, 181101 (2017) PHYSICAL REVIEW LETTERS week ending
3 NOVEMBER 2017
181101-3
difference between electromagnetic and hadronic showers
[6,26]. We applied two methods—simple two parameter
cuts and multivariate analysis (MVA) based on machine
learning—to understand systematic effects and the
stability of the resultant flux. A simple two-parameter
cut is embedded into the Kestimator defined as
K¼log10ðFEÞþRE=2cm, where REis the second
moment of the lateral energy-deposit distribution in the
TASC first layer computed with respect to the shower axis,
and FEis the fractional energy deposit of the bottom TASC
layer with respect to the total energy deposit sum in the
TASC. The average REof an electromagnetic shower in
lead is roughly estimated as ∼1.6cm (one Moliere unit),
while a proton-induced shower has a wider size because
of the spread due to secondary pions in the nuclear
interactions, making it a powerful parameter for e=p
separation. On the other hand, mainly due to the difference
between radiation length and interaction length of lead
tungstate together with the large thickness of TASC, FEis a
simple but very powerful parameter for e=p separation. The
estimated performance of e=p separation in the MC
calculations is confirmed with test beam results [23,25].
For the MVA analysis, we use the boosted decision
tree (BDT) method from the toolkit
TMVA
[27]. Multiple
parameters with a significant discrimination power between
electromagnetic and hadronic showers, and for which very
good agreement between FD and MC calculations was
confirmed, are combined into a single discrimination
function, taking into account the correlations among the
parameters. Using MC information, the BDT algorithm
is trained to maximize the separation power based on the
input parameters separately for different ranges of depos-
ited energy [6]. In order to maximize the rejection power
against the abundant protons, MVA has been adopted above
500 GeV, while the K-estimator cut was used below
500 GeV. An example of BDT response distributions is
shown in Fig. 3.
Subtraction of proton background events.—In order to
extract the residual proton contamination in the final
electron sample, templates of the Kestimator and BDT
response were used, where normalization factors for MC
electrons and MC protons are included as fitting param-
eters. The value of the selection is chosen to correspond to
80% efficiency for electrons using the distribution of MC
electrons. The contaminating protons are derived as the
ratio between the expected absolute number of events from
the distribution of MC protons and the normalization factor,
independent of the spectral shape of the electrons. The
resultant contamination ratios of protons in the final
electron sample is ∼5% up to 1 TeV, 10%–15% in the
1–3 TeV region, while a constant high efficiency of 80%
for electrons is kept.
Absolute energy scale calibration.—The energy scale
calibrated with MIPs is commonly checked in space experi-
ments by analysis of the geomagnetic cutoff energy [28].
For this study, data samples obtained by the low-energy
shower trigger (E>1GeV) are selected inside an interval
of the McIlwain Lparameter [29] of 0.95–1.25. By dividing
the interval of Linto three bins—0.95–1.00, 1.00–1.14,
and 1.14–1.25—different rigidity cutoff regions are selected
corresponding to ∼15,∼13,and∼11 GV, respectively.
The cutoff energy is calculated by using the track trajectory
tracing code
AT MN C
3[30] and the International
Geomagnetic Reference Field IGRF-12 [31]. The rigidity
cutoff in the electron flux is measured by subtracting
carefully the secondary components (reentrant albedo elec-
trons) with checking the azimuthal distribution in corre-
sponding rigidity regions. It is found that the average ratio
of the expected to measured cutoff position in the electron
flux is 1.035 0.009 (stat). As a result, a correction of the
energy scale by 3.5% was implemented in the analysis.
Systematic uncertainties.—The main sources of system-
atic uncertainties include (i) energy scale, (ii) absolute
normalization, and (iii) energy-dependent uncertainties.
(i) The energy scale determined with a study of the
rigidity cutoff is 3.50.9% (stat) higher than that obtained
with MIP calibrations. As the two methods are totally
independent, the causes of this difference have to be further
investigated to clarify their contribution to the systematic
error on the energy scale. However, the uncertainty is not
included in the present analysis, and this issue will be
addressed by further studies. Since the full dynamic range
calibration [22] was carried out with a scale-free method, its
validity holds regardless of the absolute scale uncertainty.
(ii) The systematic uncertainty related to the absolute
normalization arises from geometrical acceptance (SΩ),
live time measurement, and long-term stability of the
detector [6].SΩis a pure geometrical factor for CALET
and is independent of energies to a good approximation.
The geometry of the CALET detector was accurately
measured on the ground and is introduced in the MC
model; the systematic errors due to SΩare negligibly small.
Other errors are taken into account by studying the stability
of the spectrum for each contributing factor.
(iii) The remaining uncertainties, including track
reconstruction, various event selections, and MC model
)
BDT
BDT Response (R
10.50-0.5-1
Number of Events
1
10
2
10
3
10 Flight Data
MC Electrons
MC Protons
MC Total
475.5< E/[GeV] <598.6
< 0.5
BDT
/d.o.f. = 1.14 for -0.5 < R
2
χ
FIG. 3. An example of BDT response distributions in the
476 <E<599 GeV bin. The reduced chi square in the BDT
response range from −0.5to 0.5 is obtained as 1.14.
PRL 119, 181101 (2017) PHYSICAL REVIEW LETTERS week ending
3 NOVEMBER 2017
181101-4
dependence [6], are, in general, energy dependent. In order
to estimate tracking-related systematics, for example, the
dependence on the number of track hits and the difference
between two independent tracking algorithms [32,33] were
investigated.
Electron identification is the most important source of
systematics. To address the uncertainty in the BDT analy-
sis, in particular, 100 simulated data sets with independent
training were created, and the stability of the resultant flux
was checked in each energy bin by changing the electron
efficiency from 70% to 90% in 1% steps for the test sample
corresponding to each training set. An example for stability
of the BDT analysis is shown in Fig. 4.
By combining all the energy bins, the results are
presented in Fig. 5, where the average of all training
samples with respect to the standard 80% efficiency case
(specific training result) is presented by red squares, while
error bars represent the standard deviation corresponding
to the systematic uncertainty in the flux from the BDT
analysis in each energy bin. We confirmed that our BDT
analysis exhibits good stability with respect to training
and cut efficiency. The difference between K-estimator and
BDT results is included in the systematic uncertainty of the
electron identification [6].
Based on the above investigations, the systematic
uncertainty bands, which consider all of the components
(as the relative difference between the flux under study and
the standard case flux) except for the energy scale uncer-
tainty, are shown as black lines in Fig. 5, with each
contribution added quadratically. The various sources of
systematic uncertainties have different contributions at
various energies. In the present study, we surveyed all of
the viable choices in event selection, reconstruction, and
MC models [6,21,26,34], including those that are not
optimal, and took account of all differences in the system-
atic uncertainty. Some important details of our systematic
study are described in Ref. [6]. Systematic uncertainties
will be significantly reduced as our analysis proceeds
further and statistics increase, because most of the system-
atic uncertainties come from imperfect understanding
of data.
Electron and positron spectrum.—The differential flux
ΦðEÞbetween energy Eand EþΔE(GeV) with bin width
ΔE(GeV) is given by the following formula:
ΦðEÞ¼ NðEÞ−NBGðEÞ
SΩεðEÞTðEÞΔEðEÞ;
where ΦðEÞis expressed in m−2sr−1sec−1GeV−1,NðEÞis
the number of electron candidates in the corresponding bin,
NBGðEÞis the number of background events estimated with
MC protons, SΩ(m2sr) is the geometrical acceptance,
εðEÞis the detection efficiency for electrons defined as
the product of trigger, preselection, track reconstruction,
and electron identification efficiencies, and TðEÞ(sec) is
the observational live time. While TðEÞis basically energy
independent, at lower energies it is reduced because we
only use data taken below 6 GV cutoff rigidity. Based on
the MC simulations, the total efficiency is very stable with
energy up to 3 TeV: 73%2%.
Figure 6shows the all-electron spectrum measured with
CALET in an energy range from 10 GeV to 3 TeV, where
current systematic errors are shown as a gray band. The
present analysis is limited to fully contained events, and the
acceptance is 570 cm2sr, only 55% of the full acceptance.
Our present flux is fairly consistent with AMS-02 [5],
although it is lower than the recent Fermi LAT result [36]
above a few hundred GeV. The spectrum could be fitted
to a single power of −3.152 0.016 over 30 GeV, includ-
ing the systematic uncertainties. The structures at the
highest energies are within the (stat þsyst) errors, and,
therefore, no conclusion can be drawn at the moment on
their significance. Further development of the analysis
and more statistics will allow this energy region to be
investigated in detail.
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
948.7 < E/[GeV] < 1194.3
50 100 150 200 250 300 350
mean: 0.988
stddev: 0.050
Flux Ratio -1
70 75 80 85 90
Number of Trials
BDT-Cut Efficiency [%]
FIG. 4. Stability of BDT analysis with respect to independent
training samples and BDT-cut efficiency in the 949 <E<
1194 GeV bin. Color maps show the flux ratio dependence
on efficiency, where the bin value (number of trials) increases
as color changes from violet, blue, green, yellow, to red. A
projection onto the Yaxis is shown as a rotated histogram
(in gray color).
Energy [GeV]
10 20 30 40 50 60 2
10 2
10
×
22
10
×
33
10 3
10
×
23
10
×
3
Systematic Uncertainty
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
Total Systematic Uncertainty
BDT-cut Stability
Total Systematic Uncertainty
BDT-cut Stability
FIG. 5. Energy dependence of systematic uncertainties. The red
squares represent the systematic uncertainties stemming from the
electron identification based on BDT. The bands defined by black
lines show the sum in quadrature of all the sources of systematics,
except the energy scale uncertainties.
PRL 119, 181101 (2017) PHYSICAL REVIEW LETTERS week ending
3 NOVEMBER 2017
181101-5
We gratefully acknowledge JAXA’s contributions to the
development of CALET and to the operations onboard the
ISS. We also wish to express our sincere gratitude to ASI
and NASA for their support of the CALET project. This
work was supported in part by a JSPS Grant-in-Aid for
Scientific Research (S) (Grant No. 26220708) and by the
MEXT-Supported Program for the Strategic Research
Foundation at Private Universities (2011–2015) (Grant
No. S1101021) at Waseda University.
*yoichi.asaoka@aoni.waseda.jp
†torii.shoji@waseda.jp
‡Deceased.
[1] S. Torii et al.,Proc. Sci., ICRC2015 (2015) 581.
[2] J. Nishimura, M. Fujii, T. Taira, E. Aizu, H. Hiraiwa, T.
Kobayashi, K. Niu, I. Ohta, R. L. Golden, and T. A. Koss,
Astrophys. J. 238, 394 (1980).
[3] T. Kobayashi, Y. Komori, K. Yoshida, and J. Nishimura,
Astrophys. J. 601, 340 (2004).
[4] O. Adriani et al.,Nature (London) 458, 607 (2009).
[5] L. Accardo et al.,Phys. Rev. Lett. 113, 121101 (2014).
[6] See the Supplemental Material at http://link.aps.org/
supplemental/10.1103/PhysRevLett.119.181101 for more
detailed information about the CALET instrument, electron
identification, and systematic uncertainty as well as tabu-
lated all-electron flux, which includes Refs. [7–19].
[7] K. Yamaoka et al.,inProceedings of the 7th Huntsville
Gamma-Ray Burst Symposium, GRB 2013, eConf
C1304143, 41 (2013).
[8] P. S. Marrocchesi et al.,Nucl. Instrum. Methods Phys. Res.,
Sect. A 659, 477 (2011).
[9] P. S. Marrocchesi et al.,inProceedings of the 33rd
International Cosmic Ray Conference (ICRC2013), 2013,
p. 362.
[10] P. S. Marrocchesi et al.,Proc. Sci., ICRC2017 (2017)
156.
[11] Y. Akaike et al.,Proc. Sci., ICRC2017 (2017) 181.
[12] M. Mori et al.,inProceedings of the 33rd International
Cosmic Ray Conference (ICRC2013), 2013, p. 248.
[13] Y. Akaike et al.,inProceedings of the 32nd ICRC, 2011,
Vol. 6, p. 371.
[14] Y. Akaike et al.,Proc. Sci., ICRC2015 (2015) 613.
[15] F. Palma et al.,Proc. Sci., ICRC2015 (2015) 1196.
[16] J. Femindez-Varea, R. Mayo, J. Bar, and F. Salvat, Nucl.
Instrum. Methods Phys. Res., Sect. B 73, 447 (1993).
[17] E. Longo and I. Sestili, Nucl. Instrum. Methods 128, 283
(1975).
[18] M. Aguilar et al.,Phys. Rev. Lett. 114, 171103 (2015).
[19] Y. Yoon et al.,Astrophys. J. 839, 5 (2017).
[20] Y. Asaoka et al.,Proc. Sci., ICRC2015 (2015) 603.
[21] K. Kasahara, in Proceedings of the 24th International
Cosmic Ray Conference, Rome, Italy, 1995, Vol. 1, p. 399.
[22] Y. Asaoka, Y. Akaike, Y. Komiya, R. Miyata, S. Torii et al.,
Astropart. Phys. 91, 1 (2017).
[23] Y. Akaike et al.,inProceedings of the 33rd International
Cosmic Ray Conference (ICRC2013), 2013, p. 726.
[24] T. Niita, S. Torii, Y. Akaike, Y. Asaoka, K. Kasahara,
S. Ozawa, and T. Tamura, Adv. Space Res. 55, 2500
(2015).
[25] T. Tamura et al. (to be published).
[26] L. Pacini, Y. Akaike et al.,Proc. Sci., ICRC2017 (2017)
163.
[27] A. Hocker et al.,Proc. Sci., ACAT2007 (2007) 040.
[28] M. Ackermann et al.,Astropart. Phys. 35, 346 (2012).
[29] D. F. Smart and M. A. Shea, Adv. Space Res. 36, 2012
(2005).
[30] M. Honda, T. Kajita, K. Kasahara, and S. Midorikawa, Phys.
Rev. D 70, 043008 (2004).
[31] E. Th ´ebault et al.,Earth Planets Space 67, 79 (2015).
[32] P. Brogi et al.,Proc. Sci., ICRC2015 (2015) 595.
[33] P. Maestro, N. Mori et al.,Proc. Sci., ICRC2017 (2017)
208.
[34] S. Agostinelli et al.,Nucl. Instrum. Methods Phys. Res.,
Sect. A 506, 250 (2003).
[35] M. Aguilar et al.,Phys. Rev. Lett. 113, 221102 (2014).
[36] S. Abdollahi et al.,Phys. Rev. D 95, 082007 (2017).
[37] O. Adriani et al. (to be published).
[38] F. Aharonian et al.,Phys. Rev. Lett. 101, 261104 (2008).
[39] F. Aharonian et al.,Astron. Astrophys. 508, 561 (2009).
Energy [GeV]
10 2
10 3
10
]
2.0
GeV
-1
s
-1
sr
-2
flux[m
3.0
E
0
50
100
150
200
250
CALET
Fermi-LAT 2017 (HE+LE)
AMS-02 2014
+
+e
-
PAMELA e
HESS 2008+2009
FIG. 6. Cosmic-ray all-electron spectrum measured by CALET
from 10 GeV to 3 TeV, where systematic errors (not including the
uncertainty on the energy scale) are drawn as a gray band. The
measured all-electron flux including statistical and systematic
errors is tabulated in Ref. [6]. Also plotted are measurements in
space [35–37] and from ground-based experiments [38,39].
PRL 119, 181101 (2017) PHYSICAL REVIEW LETTERS week ending
3 NOVEMBER 2017
181101-6
