Universidade de São Paulo

2014-04

Latest results from the KASCADE-Grande

experiment Nuclear Instruments and Methods in Physics Research A,Amsterdam : Elsevier BV,v. 742, p. 10-15,

Apr. 2014http://www.producao.usp.br/handle/BDPI/51309

Downloaded from: Biblioteca Digital da Produção Intelectual - BDPI, Universidade de São Paulo

Biblioteca Digital da Produção Intelectual - BDPI

Departamento de Física e Ciência Interdisciplinar - IFSC/FCI Artigos e Materiais de Revistas Científicas - IFSC/FCI

Latest results from the KASCADE-Grande experiment

A. Chiavassa c,n, W.D. Apel a, J.C. Arteaga-Velázquez b, K. Bekk a, M. Bertaina c, J. Blümer a,d,H. Bozdog a, I.M. Brancus e, E. Cantoni c,f,1, F. Cossavella d,2, C. Curcio c, K. Daumiller a,V. de Souza g, F. Di Pierro c, P. Doll a, R. Engel a, J. Engler a, B. Fuchs d, D. Fuhrmann h,3,H.J. Gils a, R. Glasstetter h, C. Grupen i, A. Haungs a, D. Heck a, J.R. Hörandel j, D. Huber d,T. Huege a, K.-H. Kampert h, D. Kang d, H.O. Klages a, K. Link d, P. Łuczak k, M. Ludwig d,H.J. Mathes a, H.J. Mayer a, M. Melissas d, J. Milke a, B. Mitrica e, C. Morello f, J. Oehlschläger a,S. Ostapchenko a,4, N. Palmieri d, M. Petcu e, T. Pierog a, H. Rebel a, M. Roth a, H. Schieler a,S. Schoo a, F.G. Schröder a, O. Sima l, G. Toma e, G.C. Trinchero f, H. Ulrich a, A. Weindl a,J. Wochele a, J. Zabierowski k

a Institut für Kernphysik, KIT - Karlsruher Institut für Technologie, Germanyb Universidad Michoacana, Instituto de Física y Matemáticas, Morelia, Mexicoc Dipartimento di Fisica, Università degli Studi di Torino, Italyd Institut für Experimentelle Kernphysik, KIT - Karlsruher Institut für Technologie, Germanye National Institute of Physics and Nuclear Engineering, Bucharest, Romaniaf Osservatorio Astrofisico di Torino, INAF Torino, Italyg Universidade Sâo Paulo, Instituto de Física de São Carlos, Brasilh Fachbereich Physik, Universität Wuppertal, Germanyi Department of Physics, Siegen University, Germanyj Department of Astrophysics, Radboud University Nijmegen, The Netherlandsk National Centre for Nuclear Research, Department of Cosmic Ray Physics, Lodz, Polandl Department of Physics, University of Bucharest, Bucharest, Romania

a r t i c l e i n f o

Available online 19 November 2013

Keywords:Cosmic raysExtensive air showersKneeSpectra

a b s t r a c t

The KASCADE-Grande experiment operated at KIT from January 2004 to November 2012, measuring EASgenerated by primary cosmic rays in the 1016–1018 eV energy range. The experiment detected, for eachsingle event, with a high resolution, the total number of charged particles (Nch) and of muons (Nμ).

In this contribution we present the latest results about:

(i) The measurement of the all particle energy spectrum, discussing the influence of the hadronicinteraction model used to derive the energy calibration of the experimental data.

(ii) The energy spectra derived separating the events according to the Nμ=Nch ratio. This techniqueallowed us to unveil a steepening of the spectrum of heavy primaries at E� 1016:9270:04 eV and ahardening of the spectrum of light primaries at E� 1017:0870:08 eV.

(iii) The elemental spectra (for five mass groups) obtained applying a detailed unfolding analysis technique.(iv) A search for large scale anisotropies.

& 2013 Elsevier B.V. All rights reserved.

1. Introduction

Measurements of the cosmic-rays all-particle and individual elemen-tal spectra of the primary chemical composition and of the anisotropiesin the primaries arrival directions are the tools to understand thephenomenology of cosmic rays. The KASCADE-Grande experiment wasbuilt to investigate the energy range from 1016 to 1018 eV with the maingoal of searching for a change of slope in the primary spectrum of theheavy particles and to investigate the possible transition from a galacticto an extra-galactic origin of cosmic rays in this energy range.

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/nima

Nuclear Instruments and Methods inPhysics Research A

0168-9002/$ - see front matter & 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.nima.2013.11.045

n Corresponding author at: Dipatimento di Fisica, Via Pietro Giuria 1, 10125,Torino, Italy.

E-mail addresses: achiavas@to.infn.it, andrea.chiavassa@to.infn.it (A. Chiavassa).1 Now at Istituto Nazionale di Ricerca Metrologia, INRIM, Torino.2 Now at DLR Oberpfaffenhofen, Germany.3 Now at University of Duisburg-Essen, Duisburg, Germany.4 Now at University of Trondheim, Norway.

Nuclear Instruments and Methods in Physics Research A 742 (2014) 10–15

The results obtained at lower energies by KASCADE [1] andEAS-TOP [2] as well as by other experiments suggest that the kneein the primary energy spectrum observed at 3–4�1015 eV is dueto a break in the spectrum of light elements (Zr6). Severalmodels foresee a rigidity dependence of such breaks. Therefore,a knee of the heavy component is expected at around 1017 eV.Such features can only be investigated by precise measurementsboth the all-particle spectrum (i.e. the spectrum of the entireevent sample) and the spectra of different mass groups (i.e. thespectra of event samples obtained by applying a primary massdependent selection).

The evolution with energy of the primary chemical compositionbrings also relevant information concerning the transition from agalactic origin of the primary radiation to an extra-galactic one. Mostof the astrophysical models identify in a change toward a composi-tion dominated by light (mainly protons) primaries a sign of such atransition. It is therefore of main importance to perform compositionstudies in a wide energy range and with high resolution.

In addition a search for large scale anisotropies in the arrivaldirections of cosmic rays is performed, that is an observable verysensible to the propagation of the primaries in the galacticmagnetic fields. The foreseen effect is very low (of the order of10�3–10�2) and is hidden by counting differences induced bypressure and temperature variations. To take into account sucheffects we have performed the search following the East–Westmethod [3].

In this contribution we will present the updated resultsobtained for the all-particle [4], light [5] and heavy [6] primaryenergy spectra; we will discuss the elemental spectra obtainedunfolding the Nch�Nμ spectrum [7]; and we will show the upperlimits derived from a search for large scale anisotropies [8].

2. Experimental setup

The multi-detector experiment KASCADE [9] (located at 49.11N,8.41E, 110 m a.s.l.) was extended to KASCADE-Grande in 2003 byinstalling a large array of 37 stations consisting of 10 m2 scintilla-tion detectors each, the layout is shown in Fig. 1. KASCADE-Grande[10] provides an area of 0.5 km2 and operates jointly with theexisting KASCADE detectors. The joint measurements with theKASCADE muon tracking devices are ensured by an additionalcluster (Piccolo) located close to the center of KASCADE-Grandeand deployed for fast trigger purposes. For results of the muontracking devices see [11].

The Grande detectors are sensitive to charged particles, whilethe KASCADE array detectors measure the electromagnetic com-ponent and the muonic component separately. These muondetectors enable to reconstruct the total number of muons on anevent-by-event basis also for Grande triggered events.

Basic shower observables like the core position, angle-of-incidence, and total number of charged particles (Nch) are providedby the measurements of the Grande stations. The Grande arrayaccuracy in the EAS parameters reconstruction is measuredcomparing, on an event by event basis, the values independentlydetermined by the KASCADE and by the Grande arrays. A resolu-tion of 5 m on the core position, of 0.71 on the arrival direction,and of 15% on the total number of charged particles (with asystematic difference lower than 5%) has been achieved. The totalnumber of muons is determined using the core position recon-structed by the Grande array and the muon densities measured bythe KASCADE muon array detectors. The resolution on the Nμ EASparameter is evaluated reconstructing simulated events, a � 20%accuracy has been achieved. More details on the experimentalsetup and on the event reconstruction can be found in [10].

Full efficiency for triggering and reconstruction of air-showers isreached at a primary energy of 1016 eV, slightly varying on the cutsneeded for the reconstruction of the different observables [10].

3. Results

3.1. All particle energy spectrum

The energy of the primary particle that originated the detectedEAS is determined by the KASCADE-Grande experiment by meansof the Nch and Nμ observables [4], combining these two variablesindeed we can lower the dependence from the chemical composi-tion of the primary particles. This is obtained evaluating for eachevent the so-called k parameter, that is essentially a measurementof the ratio between the muon and the charged particles numbers.

k¼ log 10ðNch=NμÞ� log 10ðNch=NμÞHlog 10ðNch=NμÞFe� log 10ðNch=NμÞH

ð1Þ

log 10ðNch=NμÞH;Fe ¼ cH;Fe log 10 NchþdH;Fe ð2ÞFrom its definition is clear that k is a number centered around

zero (one) for proton (iron) generated events, if expressed as afunction of Nch for Monte Carlo events, assuming intermediatevalues for all other primaries. The values of the k parameters aretuned by a full EAS and detector simulation, the analysis reportedin [4] is based on the QGSJetII-02 [12] hadronic interaction model.Having calculated, for each event, the k parameter the primaryenergy is estimated from the Nch value:

log 10ðE=GeVÞ ¼ ½aHþðaFe�aHÞ � k� log 10ðNchÞþbHþðbFe�bHÞ � k ð3Þ

To take into account the shower evolution in atmosphere theparameters, aH;Fe, bH;Fe, cH;Fe, dH;Fe, contained in the k and E

X coordinate (m)-600 -500 -400

-700

-600

-500

-400

-300

-200

-100

0

100KASCADE Array

Grande Stations

Y co

ordi

nate

(m)

1000-100-200-300

Fig. 1. Layout of the KASCADE-Grande experiment. The KASCADE array and thedistribution of the 37 stations of the Grande array are shown. The 192 muondetectors are placed in the outer 12 clusters of the KASCADE array (hatched area).The dashed line shows the fiducial area selected for the all-particle and heavy massgroup spectra analysis. The dotted line indicates the area used for the measurementof the light mass group spectrum.

A. Chiavassa et al. / Nuclear Instruments and Methods in Physics Research A 742 (2014) 10–15 11

expressions are derived in five different angular intervals, whoseupper limits are: 16.71, 24.01, 29.91, 35.11 and 40.01. The values ofthe parameters can be found in [4].

The all-particle energy spectrum is then measured in the fivedifferent angular bins. As shown in [4] these spectra are slightlyshifted, indicating that the EAS evolution in atmosphere is notcorrectly described by the simulations. Nevertheless these differ-ences are inside the experimental uncertainties and thus wemediate them to obtain the all particle energy spectrum measuredin the zenith angle range from 01 to 401. The residuals of the all-particle energy spectrum multiplied by a factor, in such a way thatthe middle part of the spectrum becomes flat are shown in Fig. 2.

The measured spectrum cannot be described by a single power law:a hardening around 1016 eV and a steepening at log 10ðE=eVÞ ¼16:9270:10 are observed. The statistical significance of the steepeningis 2:1s, here the change of the spectral slope is from γ ¼ �2:9570:05to γ ¼ �3:2470:08. The same spectral features are meanwhile con-firmed by the Tunka-133 [13] and Ice-Top [14] experiments.

This procedure relies on the EAS simulation and thus dependson the high-energy hadronic interaction model used. To evaluatethe systematic effects introduced in the all-particle spectrummeasurement [15] the same procedure has been repeated using

events simulated with the SIBYLL2.1 [16], EPOS1.99 [17] andQGSJetII-04 [18] hadronic interaction models.

Applying the energy calibration functions, obtained by eachmodel, to the measured data the all-particle energy spectra for thefive zenith angle bins are obtained for the four previously men-tioned models; for all of them, except QGSJetII-04, an unfoldingprocedure has been applied. Different sources of uncertainty affectthe all-particle energy spectrum. A detailed description is reportedin [4]. They take into account: (a) the angular dependence of theparameters appearing in the energy calibration functions ofthe different angular ranges. (b) The possible bias introduced inthe energy spectrum by different primary compositions. (c) The spec-tral slope of Monte Carlo used in the simulations. (d) The recon-struction quality of Nch and Nμ. The total systematic uncertainty is� 20% at the threshold (E ¼ 1016 eV) and � 30% at the highestenergies (E ¼ 1018 eV) almost independently from the interactionmodel used to interpret the data. The final all-particle spectrum ofKASCADE-Grande is obtained (see Fig. 3) by combining the spectrafor the individual angular ranges. Only those events are taken intoaccount, for which the reconstructed energy is above the energythreshold for the angular bin of interest. In general the shape ofthe energy spectrum is very similar for all models, however, a shift influx is clearly observed which amounts to � 25% increase in the caseof SIBYLL2.1 and � 10% decrease in the case of EPOS1.99. This is theconsequence of the energy shift assigned on an event-by-event basis.This result gives an estimation of the systematic uncertainty on theexperimental flux due to the hadronic interaction model used tointerpret the data, and it is essentially independent of the techniqueused to derive the flux, namely averaging the fluxes obtained indifferent angular bins. The shift in the assigned energy to the data isalso visible in the hardening around � 2� 1016 eV and in thesteepening around 1017 eV which look shifted among the modelsin general agreement with the energy shift. This result indicates thatthe features seen in the spectrum are not an artifact of the hadronicinteraction model used to interpret the data but they are in themeasured data. In the overlapping region, KASCADE-Grande data arecompatible inside the systematic uncertainties with KASCADE datainterpreted with the same model.

3.2. Energy spectra of individual mass groups

The k parameter previously defined can also be used toseparate the events in samples generated by two different primary

1016 1017 1018

I/(A

× E

-2.9

18)-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

primary energy [eV]

KASCADE-Grande

Fig. 2. The all-particle energy spectrum obtained with KASCADE-Grande. Theresidual intensity after multiplying the spectrum with a factor of E2:918 is displayedas well as the band of systematic uncertainty.

Energy (eV/particle)

1610 1710 1810

)1.

5 e

V-1

sr

-1 s

ec-2

J(E

) (m

2.5

dI/d

E x

E

1510

1610

1710

Akeno (J.Phys.G18(1992)423)AGASA (ICRC 2003)HiResI (PRL100(2008)101101)HiResII (PRL100(2008)101101)Yakutsk (NewJ.Phys11(2008)065008)AUGER (ICRC 2009)

EAS-TOP (Astrop.Phys.10(1999)1)TIBET-III (ApJ678(2008)1165)GAMMA (J.Phys.G35(2008)115201)TUNKA (Nucl.Phys.B,Proc.Sup.165(2007)74)KASCADE (QGSJET01 Astrop.Phys.24(2005)1)KASCADE (SIBYLL2.1 Astrop.Phys.24(2005)1)KASCADE (QGSJET II, M.Finger 2011)KASCADE (EPOS1.99, M.Finger 2011)

µKASCADE-Grande (QGSJET II) Nch-NµKASCADE-Grande (SIBYLL 2.1) Nch-NµKASCADE-Grande (EPOS 1.99) Nch-N

(RAW)µKASCADE-Grande (QGSJET II-04) Nch-N

Fig. 3. Comparison of the all-particle energy spectrum obtained with KASCADE-Grande data based on SIBYLL2.1 (blue), QGSJetII-02 (black), QGSJetII-04 (pink) and EPOS1.99(red) models to results of other experiments. The band denotes the systematic uncertainties in the flux estimation. (For interpretation of the references to color in this figurecaption, the reader is referred to the web version of this article.)

A. Chiavassa et al. / Nuclear Instruments and Methods in Physics Research A 742 (2014) 10–1512

mass groups. To emphasize the features of the heavy elements weselected the electron-poor events with kepðEÞZðkCðEÞþkSiðEÞÞ=2,i.e. events with a k value greater than the mean value of theexpectations for C and Si primaries (QGSJetII-02 [12] basedsimulation). The spectra of these event samples are shown inFig. 4, the band indicates changes of the spectra when the cut isvaried by one standard deviation in the kep(E) definition.

The reconstructed spectrum of the electron-poor events shows adistinct knee like feature at about 8�1016 eV. Applying a fit of twopower laws to the spectrum interconnected by a smooth knee [19]results in a statistical significance of 3:5s that the entire spectrumcannot be described with a single power-law. The change of thespectral index is Δγ ¼ �0:4870:05 from γ ¼ �2:7670:02 toγ ¼ �3:2470:05 with the break position at log 10ðE=eVÞ ¼ 16:9270:04. The spectrum of the electron-rich events (corresponding, withthis cut definition, to light and medium mass primaries) is compatiblewith a single power law with slope index γ ¼ �3:1870:01. Arecovery to a harder spectrum at energies greater than 1017 eV cannotbe excluded by this analysis.

To increase the statistics and deeply investigate this possiblehardening of the light primaries spectrum a larger fiducial area hasbeen defined, essentially accepting events at larger distances fromthe muon detector (i.e. from the KASCADE array, see Fig. 1). Themain effect of this event selection is that the 100% efficiency isreached at higher energies, that is not a problem for this analysisaimed to study a possible spectral feature at energies greater than1017 eV. In order to emphasize features of the light mass group weredefine the cut on the k parameter as kerðEÞrðkCðHeÞþkCðEÞÞ=2(again a simulation based on the QGSJetII-02 hadronic interactionmodel is used). The obtained spectrum is shown in Fig. 5; ahardening, or ankle-like feature, is clearly observed. Fitting thisspectrum with the same function used for the all-particle andheavy mass groups primary spectra we obtain a change of thespectral index from γ ¼ �3:2570:05 to γ ¼ �2:7970:08 at anenergy of log 10ðE=eVÞ ¼ 17:0870:08. The measured number ofevents above the bending is Nmeas ¼ 595. Without the bending wewould expect Nexp ¼ 467 events above this ankle-like feature.The Poisson probability to measure at least Nmeas events abovethe bending, if Nexp are expected is P � 7:23� 10�9, correspondingto a 5:8s significance.

Comparing the two previous observations it is important tonotice that the knee in the heavy component occurs at a lowerenergy compared to the bending in the spectrum of the lightprimaries. Therefore the steepening of the heavy spectrum and therecovery of the light component are not due to a bias in thereconstruction or separation procedures. It is worth pointing outthat the slope of the heavy mass spectrum above the knee-likefeature is very similar to the slope of the light mass spectrumbefore the ankle-like feature. The slope index of the light massgroup spectrum above the ankle-like feature is γ � �2:7 and thiscan be interpreted as an indication of an injection of a new(extragalactic) population of high energy cosmic-rays [5].

3.3. Energy spectra for elemental groups by unfolding analysis

The measured two-dimensional shower size spectrum of thenumber of charged particles (log 10ðNchÞ) vs. the number of muons(log 10ðNμÞ) is the basis for the unfolding analysis [1] to infer theabsolute fluxes of different mass groups. With the KASCADE-Grande resolution we can separate five different groups. Onlyevents with shower sizes for which the experiment is fullyefficient are considered, i.e. log 10ðNchÞZ6:0 and log 10ðNμÞZ5:0.In order to avoid effects due to the varying attenuation of theshower sizes for different angles of incidence, the data set used isrestricted to showers with zenith angles θr181.

The analysis objective is to compute the energy spectra of fivecosmic ray mass groups [7] (represented by protons (p), helium(He), carbon (C), silicon (Si), and iron (Fe) nuclei) from 1016 eV to1017 eV primary energies. The convolution of the sought-afterdifferential fluxes dJn=d log 10E of the primary cosmic ray nuclein into the measured number of showers Ni contributing to the celli of shower size plane, and thus to the content of this specificcharged particle and muon number bin (log 10ðNchÞ; log 10ðNμÞ) inthe previously mentioned bi-dimensional spectra, can bedescribed by an integral equation:

Ni ¼ 2πAf Tm ∑Nnucl

n ¼ 1

Z 181

01

Z þ1

�1

dJnd log 10E

pn sin θ cos θ d log 10E dθ

ð4ÞOne has to sum over all Nnucl elements contributing to the all-

particle cosmic ray spectrum, in this analysis the five representa-tive primaries. Tm is the measurement time, the factor 2π accounts

electron-rich sample

all-particle (104489 events)electron-poor sample

1020

dI/d

E x

E2.

7 (m-2

sr-1

s-1eV

1.7 )

log10(E/eV)16 16.5 1817 17.5

= -2.95±0.05

= -3.24±0.08 = -2.76±0.02

= -3.24±0.05

KASCADE-Grande

= -3.18±0.01

10 eV17 10 eV18

1019

Fig. 4. Reconstructed energy spectrum of the electron-poor and electron-richcomponents together with the all-particle spectrum for the angular range 0–401.The error bars show the statistical uncertainties; the bands assign systematicuncertainties due to selection of the subsamples.

(E/eV)10

log16.6 16.8 17 17.2 17.4 17.6 17.8 18 18.2

)1.

7 e

V-1

s-1

sr

-2 (m

2.7

dI/d

E x

E

1910

light (sep. between He-CNO)

band of systematic uncertainty2506

1487882

539322

195144

92 5543

40 188

Fig. 5. The reconstructed energy spectrum of the light mass component of cosmicrays. The number of events per energy bin is indicated as well as the range ofsystematic uncertainty. The error bars show the statistical uncertainties.

A. Chiavassa et al. / Nuclear Instruments and Methods in Physics Research A 742 (2014) 10–15 13

for the integration over the azimuth angle, and Af is the chosenfiducial area. The term pn represents the conditional probability toreconstruct a certain combination of charged particle and muonnumber respectively (i.e. to get an entry in the cell (log 10ðNchÞ;log 10ðNμÞ) if the air shower inducing particle was of the type n andhad an energy E.

In Fig. 6, the unfolded differential energy spectra of lighterprimaries (protons as well as helium and carbon nuclei, upperpanel), and the spectra of heavier ones (silicon and iron nuclei,lower panel) are depicted. In addition, all five unfolded spectraare summed up to the all-particle flux, which is also shown.The shaded band indicates the methodical uncertainties, while theerror bars represent the statistical error originating from thelimited measurement time. The uncertainties due to the interac-tion models used, i.e. QGSJET-II-02 and FLUKA2002.4 [20], cannotbe considered.

With increasing energy the heavy component gets the domi-nant contributor to the cosmic ray composition. The spectra oflighter primaries are rather featureless within the given

uncertainties, while in the iron spectrum there is a slight bendingdiscernible at around 1017 eV. The position of this knee-likestructure agrees with those observed in the all-particle and inthe heavy component spectra previously discussed.

There are no indications so far that the interaction modelsused, i.e. QGSJET-II-02 and FLUKA 2002.4, have serious deficits inthe description of the physics of hadronic interactions at theseenergies, which, however, does not mean necessarily that thesemodels must be accurate in all details. Different interaction modelsprimarily have impact on the absolute scale of energy and masses,such that model uncertainties can shift the unfolded spectra,possibly resulting in different abundances of the primaries, whilespecific structures, e.g. knee-like features of the spectra, are lessaffected by the models.

3.4. Large scale anisotropies

The search for large scale anisotropies has been performedthrough a differential method, the so-called East–West [3]method, based on the counting rate differences between East-ward and West-ward directions. This method allows us to removecounting rate variations caused by atmospheric and instrumentaleffects. The used data set contains 107 events recorded betweenDecember 2003 and October 2011. To ensure reconstructionquality, a cut on the zenith angle θ and on the number of chargedparticles at observation level (Nch) was applied: θo401 andlog ðNchÞ45:2.

Fig. 7 shows the modulation in sidereal5 time obtained usingthe East–West method. The amplitude of the first harmoniccalculated in sidereal time is ð0:2870:08Þ � 10�2 with a 0.2%Rayleigh probability of being due to background fluctuation(i.e. s¼ 3:5) at median energy 3:3� 1015 eV. The 99% C.L. upperlimit on the amplitude is 0:47� 10�2, derived according to thedistribution drawn from a population characterized by an aniso-tropy of unknown amplitude and phase as derived by Linsley [21].

To investigate a variation of the amplitude and phase of thefirst harmonic with primary energy we have performed the sameanalysis in intervals of the number of charged particles. In order tohave bins containing events of similar energy and have goodstatistics in each of them we have chosen Δlog Nch ¼ 0:4. Theresults are shown in Table 1, in none of the bins the amplitude of

(E/GeV)10

primary energy in log

)1.

50 G

eV-1

s-1

sr

-2 /

(m2.

50 E

dif.

flux

dJ/d

E ·

10

210

310sum spectrumpHeC

(E/GeV)10

primary energy in log

7 7.2 7.4 7.6 7.8 8 8.2

7 7.2 7.4 7.6 7.8 8 8.2

)1.

50 G

eV-1

s-1

sr

-2 /

(m2.

50 E

dif.

flux

dJ/d

E ·

10

210

310sum spectrumSiFe

Fig. 6. The unfolded energy spectra for elemental groups of cosmic rays, repre-sented by protons, helium, and carbon nuclei (upper panel) as well as by silicon andiron nuclei (lower panel), based on KASCADE-Grande measurements. The all-particle spectrum that is the sum of all five individual spectra is also shown.The error bars represent the statistical uncertainties, while the error bands markthe maximal range of systematic uncertainties. The response matrix used is basedon the interaction models QGSJetII-02 [12] and FLUKA 2002.4 [20].

Fig. 7. Sidereal time distribution of the number of counts (θo401 and log Nch45:2)in 20 min intervals obtained applying the East–West method. The dashed line showsthe calculated first harmonic.

5 Sidereal time: Common time scale among astronomers which is based on theEarth's rotation measured relative to the fixed stars.

A. Chiavassa et al. / Nuclear Instruments and Methods in Physics Research A 742 (2014) 10–1514

the harmonic is statistically significant and so we have calculatedthe upper limits at 99% confidence level. The Nch limits of theintervals used for the harmonic analysis are converted to primaryenergy and the median energy of the events in each bin is definedas representative energy.

4. Conclusions

The KASCADE-Grande experiment took data from January 2004to end 2012, detecting EAS generated by primary cosmic rays inthe 1016–1018 eV energy range. In this contribution we have shownthe main results obtained so far by the experiment:

1. A measurement of the all-particle energy spectrum, showing thatit cannot be described by a single slope power law. A hardeningslightly above 1016 eV and a steepening at log 10ðE=eVÞ ¼16:9270:10 are detected.

2. The measurement of the light and heavy primary mass groupenergy spectra. These spectra were obtained dividing theevents in two samples on the basis of the ratio between themuon and the charged particles numbers. A steepening atlog 10ðE=eVÞ ¼ 16:9270:04 in the spectrum of the electron poorevent sample (heavy primaries) and a hardening atlog 10ðE=eVÞ ¼ 17:0870:08 in the one of the electron rich (lightprimaries) one were observed. The slope of the heavy massgroup spectrum above the knee-like feature is similar to theone of the light mass spectrum before the ankle-like feature.

3. Applying the unfolding algorithm to the Nch vs: Nμ spectrumwe could extract the spectra of five different elemental mass

groups. These measurements rely on the hadronic interactionmodel utilized in the EAS simulation. Making use of the samehadronic interaction model to interpret the data, the measuredfluxes are in agreement with those obtained at lower energiesby the KASCADE array and with the findings above for theKASCADE-Grande energy range.

4. Upper limits on the amplitude of large scale anisotropies in fourNch bins.

Acknowledgments

The authors would like to thank the members of the engineer-ing and technical staff of the KASCADE-Grande collaboration, whocontributed to the success of the experiment. The KASCADE-Grande experiment is supported in Germany by the BMBF andby the Helmholtz Alliance for Astroparticle Physics – HAP fundedby the Initiative and Networking Fund of the Helmholtz Associa-tion, by the MIUR and INAF of Italy, the Polish Ministry of Scienceand Higher Education, and the Romanian Authority for ScientificResearch UEFISCDI (PNII-IDEI Grant nos. 271/2011 and 17/2011).

References

[1] T. Antoni, et al., Astropart. Phys. 24 (2005) 1.[2] M. Aglietta, et al., Astropart. Phys. 10 (1999) 1.[3] R. Bonino, et al., Astrophys. J. 738 (2011) 67.[4] W.D. Apel, et al., Astropart. Phys. 36 (2012) 183.[5] W.-D. Apel, et al., Phys. Rev. D 87 (2013) 081101.[6] W.D. Apel, et al., Phys. Rev. Lett. 107 (2011) 171104.[7] W.D. Apel, et al., Astropart. Phys. 47 (2013) 54.[8] C. Curcio, et al., KASCADE-Grande collaboration, in: Proceedings of the 33rd

ICRC, Rio de Janeiro, 2013.[9] T. Antoni, et al., Nucl. Instrum. Methods A 513 (2003) 490.[10] W.D. Apel, et al., Nucl. Instrum. Methods A 620 (2010) 202.[11] W.D. Apel, et al., KASCADE-Grande contributions to the 32nd ICRC, Beijing,

China astro-ph/1111.5436v1, 2011.[12] S. Ostapchenko, Nucl. Phys. B (Proc. Suppl.) 151 (2006) 143.[13] L. Kuzmichev, et al., in: Proceedings of 32nd International Cosmic Ray

Conference, Beijing, vol. 1, 2011, pp. 209.[14] M.G. Aartsen et al., Phys. Rev. D 88 (2013) 042004.[15] W.D. Apel et al., J. Adv. Space Res., http://dx.doi.org/10.1016/j.asr.2013.05.008,

in press.[16] E.-J. Ahn, et al., Phys. Rev. D 80 (2009) 094003.[17] K. Werner, Nucl. Phys. B (Proc. Suppl.) 175 (2008) 81.[18] S. Ostapchenko, Phys. Rev. D 83 (2011) 014018.[19] T. Antoni, et al., Astropart. Phys. 16 (2002) 245.[20] G. Battistoni, et al., AIP Conf. Proc. 896 (2007) 31–49.[21] J. Linsley, Phys. Rev. Lett. 34 (1975) 1530.

Table 1Results of harmonic analysis through the East–West method for five intervals ofNch.

log 10ðNchÞ E (PeV) Asid � 10�2 Hours

5.2–5.6 2.6�1015 0.2670.10 1571.45.6–6 5.5�1015 0.3970.17 16.371.66.0–6.4 1.2�1016 0.6770.41 8.472.26.4–6.8 2.5�1016 0.571.0 18.476.64 6.8 6.3�1016 4.672.2 16.171.8

log 10ðNchÞ E (PeV) U.L.�10�2 P (%)

5.2–5.6 2.6�1015 0.49 35.6–6 5.5�1015 0.77 76.0–6.4 1.2�1016 1.54 266.4–6.8 2.5�1016 2.8 854 6.8 6.3�1016 9.3 11

A. Chiavassa et al. / Nuclear Instruments and Methods in Physics Research A 742 (2014) 10–15 15