32ND INTERNATIONAL COSMIC RAY CONFERENCE, BEIJING 2011

Temporal and spatial structure of the extensive air shower front with the ARGO-YBJ experi-ment

A.K. CALABRESE MELCARNE1, G. MARSELLA2,3, D. MARTELLO2,4, L. PERRONE2,3 , S. SBANO4

(ON BEHALF OF THEARGO-YBJ COLLABORATION)1 Istituto Nazionale di Fisica Nucleare, CNAF, I-40127 Bologna, Italy2 Istituto Nazionale di Fisica Nucleare (INFN), Sezione di Lecce, I-73100, Lecce, Italy3 Dipartimento di Ingegneria dell’Innovazione, Universita del Salento, I-73100, Lecce, Italy4 Dipartimento di Fisica, Universita del Salento, I-73100, Lecce, Italygiovanni.marsella@le.infn.it

Abstract: The ARGO-YBJ experiment is an Extensive Air Shower array currently operating at the high altitude Yang-bajing Cosmic Ray Laboratory (Tibet, P.R. China 4300 m a.s.l.). The detector consists of a dense layer of Resistive PlateCounters (RPCs) covering an area of about 6700 m2. One of the major benefits of having this design is the abilitytomeasure the temporal and spatial structure of secondary particles of the shower with high time resolution and with anaccurate determination of their position at the ground. This provides a unique enviroment to explore cosmic ray showerswith respect to their front shape. We present a detailed study of the curvature and of the thickness of the shower disc usingevents with energies up to several hundreds of TeV. DetailedMonte Carlo simulations of the shower development and ofthe detector response are performed to investigate the implications of our observations. Finally, we discuss the potentialof this measurement in the evaluation of standard surface array observables with particular attention to the events withtime structure peculiarities like double shower fronts arediscussed.

Keywords: Cosmic Rays - Ground based Astronomy - Extensive Air Showers

1 Introduction

The ARGO-YBJ experiment (Astrophysical Radiationwith Ground-based Observatory at YangBajing) has beendesigned to study cosmic rays and cosmicγ-radiation at en-ergy larger than few hundreds GeV, by detecting air show-ers at high altitude with wide-aperture and high duty cycle.The apparatus is a single layer detector logically dividedinto 153 units called clusters (7.64×5.72 m2), each made of12 Resistive Plate Counters (RPCs) operating in streamermode. Each RPC is read out using 10 pads (62×56 cm2),which are further divided into 8 pick-up strips providing alarger particle counting dynamic range. The signal comingfrom all the strips of a given pad are sent to the same chan-nel of a multi-hit TDC. Pads are the time elemental unitsfor measuring the pattern of the shower front with time res-olution of ∼ 1.8 ns. The percentage of active area in thecentral array is 92%. To improve the reconstruction capa-bility, the surrounding area has been partially instrumentedwith a guard ring of RPCs extending the detector layoutup to 100×111 m2. ARGO-YBJ is operating in its com-plete layout since 2007 allowing a complete and detailedthree-dimensional reconstruction of the shower front withunprecedented spatial and time resolution.

The space-time structure of extensive air showers dependson primary mass, energy and arrival direction and on theinteraction mechanisms with air nuclei. Measurements ofshower parameters with several detection techniques wouldbe required for a detailed knowledge of the shower front.A flat array like ARGO-YBJ can measure the particles ar-rival times and their densities at ground. The digital readoutallows detecting shower secondary particles down to verylow density and the high space-time granularity is able toprovide a fine sampling of the shower front close to thecore. The time profile of the shower front can be recon-structed by the time of fired pads. Particles within severaltens of meters from the shower core are expected to forma curved front. Indeed, to reconstruct the primary parti-cle arrival direction, the space-time coordinates (positionand time of fired pads in the event) can be fitted to a conewhose axis crosses the core position at the ground. Thearrival times of the particles are fitted by minimizing thefollowing quantity:

S2=

1

W

Nhit∑

i=1

wi

(

ti − t0 −xi

cl −

yi

cm −

Ri

cα

)2

(1)

The reconstructed parameters are the direction cosinesl,mand the timet0. The sum is over the fired pads,W =∑Nhit

i=1wi beingwi the number of strips fired in the i-th

ARGO COLLABORATION AUGER, SPACIAL AND TIME STRUCTURE OF AIR SHOWERS

pad,ti is the measured time,xi, yi are the pad coordinatesandc is the light velocity. The conicity correction dependson the conicity coefficient (α) and on the pad distance (Ri)to the core in the plane perpendicular to the shower axisA Maximum Likelihood based algorithm, using a NKG-like [1] lateral distribution function [2] is used to performa reliable reconstruction of the shower core position andarrival direction up to the edge of the active carpet andslightly beyond. The time structure of the shower disc hasbeen studied as a function of the distance to the showeraxis. It is identified by the zenith angleθ and the azimuthangleφ of the primary particle arrival direction. Follow-ing the approach described in detail in[3], the curvature ofthe shower front, defined as the mean of time residuals withrespect to a planar fit, is investigated in the energy range be-tween 300 GeV and 100 TeV and its features are employedto characterize the standard surface array observables.

2 Temporal and spatial structure of airshowers

The analysis uses a data sample of about3.5 × 108 events

satisfying the trigger condition of at least 20 fired pads ina time window of 420 ns[4]. The reconstructed core po-sitions are required to be within the central carpet and aquality cut on theS2 of the fit has been applied in orderto reject mis-reconstructed events. Under these conditions,a Monte Carlo study has shown that the fraction of mis-reconstructed events with true core position external to thecentral carpet is about 10% at the lowest multiplicity anddecreases down to the few percent for increasing multiplic-ities. This analysis is restricted to events with reconstructedzenith angleθ < 15

◦.The spatial and time structure of the shower disc has beenstudied as a function of the distance to shower axis, in in-tervals of 1 m between 10 m and up to a maximum distanceof 70 m. Fig.1 (top) shows the mean of time residuals withrespect to a planar fit for different pad multiplicities. Thearrival time delay from planar fit increases with distanceup to 10 ns for particles landing further than 50 m fromthe core. No significant dependence on pad multiplicity isobserved.

Shower-by-shower fluctuations play a key role for theunderstanding of EAS morphology. The design of theARGO-YBJ detector offers a unique chance to have high-resolution pictures of shower footprints at the ground. Eachobserved event keeps track of the propagation through theatmosphere. This factor can influence the shape of timefront and the sequence of arrival times. For primaries inter-acting deeper in the atmosphere, (originating “young show-ers”), due to geometrical reasons, the arrival of particlesat a given lateral distance is expected to be more delayedcompared to primaries that have interacted higher (origi-nating “old” showers). Young showers will then exhibit asteeper time profile with respect to a planar fit. A study ofthe correlation between the conicity parameterα assignedto each reconstructed event and the atmospheric depth at

distance to shower axis (m)10 20 30 40 50 60 70

resi

dual

s (p

lana

r fit

) ns

0

2

4

6

8

10

12

14

ht8001000dEntries 6770340Mean 24.64Mean y 2.711RMS 16.91RMS y 17.3Underflow 0Overflow 27.88Integral 529.3Skewness 14.62Kurtosis 37.63

ht8001000dEntries 6770340Mean 24.64Mean y 2.711RMS 16.91RMS y 17.3Underflow 0Overflow 27.88Integral 529.3Skewness 14.62Kurtosis 37.63

ht10001200_015dEntries 5213220Mean 24.66Mean y 2.761RMS 16.88RMS y 16.67Underflow 0Overflow 28.84Integral 540.9Skewness 14.68Kurtosis 37.86

ht10001200_015dEntries 5213220Mean 24.66Mean y 2.761RMS 16.88RMS y 16.67Underflow 0Overflow 28.84Integral 540.9Skewness 14.68Kurtosis 37.86

= (200 - 400)hitN

= (400 - 800)hitN

= (800 - 1000)hitN

= (1000 - 1200)hitN

distance to shower axis (m)10 20 30 40 50 60 70

resi

dual

s (p

lana

r fit

) ns

0

5

10

15

20

ht200400_015dEntries 2.71871e+07Mean 24.91Mean y 2.789RMS 17.15RMS y 26.23Underflow 0Overflow 24.29Integral 521.6Skewness 13.17Kurtosis 32.59

ht200400_015dEntries 2.71871e+07Mean 24.91Mean y 2.789RMS 17.15RMS y 26.23Underflow 0Overflow 24.29Integral 521.6Skewness 13.17Kurtosis 32.59

= (200 - 400)hitData N

= (200 - 400)hitMC N

Figure 1: Mean of time residuals with respect to a planarfit (top), data. Comparison to simulation (proton) for padmultiplicity between 200 and 400 (bottom).

shower maximum stage, Xmax, has been performed usingCORSIKA[5] simulations. Results are shown in Fig 2 forproton primaries with zenith angle below 15◦. The conicity

hprof2

(ns/m)αConicity coefficient 0.02 0.025 0.03 0.035 0.04 0.045 0.05

)-2

Xm

ax (

g cm

350

400

450

500

550

600

650hprof2hprof2

Figure 2: The atmospheric depth at shower maximum (truevalue from simulation) as a function of the (reconstructed)conicity parameterα (statistical uncertainties only).

increases (decreases) for deeper (shallower) Xmax and thisfeature suggests to adopt this observable as an estimator ofthe shower development stage. A third degree polynomialfit to the profile shown in Fig. 2 is used to parametrize the

32ND INTERNATIONAL COSMIC RAY CONFERENCE, BEIJING 2011

correlation between the observedα and the correspondingmean Xmax. For a sample of data selected with the samecriteria used for the simulation, this method provides anestimate of the mean Xmax consistent with the true valuewithin a ten gcm−2. Despite not exaustive, this test givesan encouraging indication to be further exploited.

The potential of the knowledge of the shower front shapehas been also investigated by splitting the simulated dataset in two sub-samples, on the basis of the conicity. Eventswith reconstructedα less or larger than 0.035 ns/m are clas-sified as “old” or “young” and the reconstructed multiplic-ity of fired strips at the ground is studied as a function oftrue energy (see Fig.3).

E/TeV10 20 30 40 50 60 70 80 90 100

num

ber

of fi

red

strip

s

310

410

young

old

Figure 3: Strip numerosity as a function of true energy be-tween 10 and 100 TeV, for “old” and “young” showers withzenith angle less than 15◦.

For a given energy of primary particle, the observed multi-plicity may be very different for the two classes of events,as it is influenced by the altitude of the first interactionpoint (i.e. by the shower age). An independent measure-ment of the stage of the shower development can in prin-ciple allow an improvement of the resolution of the multi-plicity, leading to a better accuracy in the estimate of theenergy.

3 Large rms shower fronts

This section is devoted to the study of events that showparticularly wide time distribution. After a study on theshower time profile and on the lateral distribution at differ-ent time delay from the shower front, showers with largetime residual rms with respect to the shower front havebeen investigated. The longitudinal time structures in datacould help to better define selection criteria for particu-lar analysis, suche asγ/hadron separation, compositionor ”exotic” physics, and allow a better determination ofEAS disc structure and correlations between front profile,front thickness and core distance. By this analysis, sev-eral structures have been observed. The events are selectedrequiring a strip multiplicity threshold greater than 120,in order to have a proper multiplicity to reconstruct two

separated fronts, and a time residual rms from the coni-cal front greater than 15 ns, which correspond to a cut ofS2 > 225ns2. The cut in multiplicity correspond to selectthe 12% of the events, while the cut in rms reduce the sam-ple to 11% of the total events. For double shower fronts anadditional request has been defined: the time residual fromthe conical shower front is fitted with two Gaussian distri-butions. If the two Gaussians are well separated the eventis selected. In Fig 4 a double front and a wide shower areshown. In this section only double front showers analysiswill be reported.

xhit (m)-60 -40 -20 0 20 40 60yhit (m)-50-40-30-20-1001020304050

time

(ns)

1050

1100

1150

1200

1250

1300

1350

1400

xhit (m)-40 -20 0 20 40 60yhit (m) -50-40-30-20-1001020304050

time

(ns)

600

800

1000

1200

1400

1600

time:yhit/100.:xhit/100. {nev==249219}

Figure 4: Top: a wide shower. These events are character-ized by a wide and homogeous space-time distributio Bot-tom: a double front shower. Two well separated fronts areevident.

As shown in the picture, double shower fronts are two wellseparated events. In order to establish the origin of theseevents, a detailed analysis has been carried out. First of allthe two groups of hits are separated and the shower prop-erties reconstructed separately. After that reconstructedparameters such as subshower multiplicity, direction andarrival time are compared. Requiring good reconstruc-tion parameters on both subshower permits to detect purecoincidence events, compatible with the expected doubleevents coincidences in the trigger time windowsτ=2µs. Atpresent38 × 10

6 events have been processed and 27000were selected as double showers and 4000 results as dou-ble coincidences. The distribution that have been comparedbetween the subshowers are the multiplicity, the relative

ARGO COLLABORATION AUGER, SPACIAL AND TIME STRUCTURE OF AIR SHOWERS

time delay and the relative angular distribution. In case ofpure time coincidences the multiplicity distribution shouldbe similar for the two subshowers. The relative delay, de-fined as the time difference between the twoT0 from theplanar fit applied to reconstruct the subshower parameters,are compatible with the expected exponential distributione−λt, with λ = 3.5 kHz. The relative angular distributionhave been compared with a similar distribution obtainedby a toy Montecarlo. The result is fully consistent with therandom ditribution of two events distributed with acos2θ

zenith angle dependance. Finally, the observed number ofevent is in total agreement with the expected one. The ex-pected number of events have been calculated as the num-ber of coincidences expected in the trigger time windowτ taking into account the quality selection on subshowers.The expected rate is equal to:

λexp = λ1 × λ2 × τ (2)

whereλ1 andλ2 are the oberved rate of the showers sat-isfying the quality requirements for the subshowers in thesame sample of data.

The same analysis has been applied to a full sample ofMC data of several elements (Proton and Helium) at zenithangle ranging from0

o to 45o and in an energy range be-

tween 100 GeV and 1 PeV. Several Events show interestingdelayed time structures, especially in heavy elements andhigher energy range, but MC events,as expected in anal-ysis of random time coincidences, don’t reproduce dou-ble shower angular distribution, indicating that eventualin-teresting events in shower cascade development should besearched at low values of the relative angular distribution.

4 Conclusions

In this work the very high space-time granularity ofARGO-YBJ has been exploited in detail to investigatedeeply the shower front morpholgy. Various observableshave been defined to better characterize EAS properties,such as shower age and primary composition. The possi-bility to select between ”old” and ”young” shower basedon the conicity parameterα has been shown. Longitudi-nal time structures have also been detected selecting eventswith large time distribution around the front. The attentionhas been focused on double front showers and , in partcu-lar, to identify the random double coincidences expected inARGO-YBJ experiment.

References

[1] K. Greisen, Ann.Rev.Nucl.Sci.10 (1966) 63[2] G. Di Sciascio et al., Proc. of 28th International Cos-

mic Ray Conference (ICRC 03), Tsukuba, Japan, Ses-sion OG.2.5 (2003) 3015;

[3] A.K. Calabrese Melcarne, L. Perrone for the ARGO-YBJ Collaboration, Proc. of 31st International Cos-mic Ray Conference (ICRC 09), Lodz, Poland.

[4] A.Aloisio et al., The trigger system of the ARGO-YBJ experiment, IEEE Trans. Nucl. Sci., NS-51(4),(2004) 1835-1839.

[5] D. Heck et al.CORSIKA: A Monte Carlo Code toSimulate Extensive Air Showers, ForshungszentrumKarlsruhe, FZKA 6019 (1998) and references therein