a Xi :as o-ph/0108270 1 16 Aug 2001
Ve y high-ene gy γ- ay obse a ions o he C ab nebula and
o he po en ial sou ces wi h he GRAAL expe imen
F.A que os1, J.Balles in2, M.Be enguel3, D.M.Bo que1, E.F.Camacho4,
M.Diaz5, H.-J.Gebaue 5, R.En iquez1, R.Plaga5
1Facul ad de Ciencias Fisicas, Uni e sidad Complu ense, E-28040 Mad id, Spain
2CIEMAT-Depa amen o Ene gias Reno ables,
Pla a o ma Sola de Alme ia, E-04080 Alme ia, Spain
3Depa amen o de Lenguajes y Compu aci´on,
Uni e sidad de Alme ia, 04120 Alme ia, Spain
4Escuela Supe io de Ingenie os, Uni e sidad de Se illa, E-41012 Se illa, Spain
5Max-Planck-Ins i u ¨u Physik, 80805 M¨unchen, Ge many
Oc obe 23, 2018
Abs ac
The “Gamma Ray As onomy a ALme ia” (GRAAL) expe imen uses 63 helios a -
mi o s wi h a o al mi o a ea o ≈2500 m2 om he CESA-1 ield a he “Pla a o ma
Sola de Alme ia” (PSA) o collec Che enko ligh om ai showe s. The de ec o is
loca ed in a cen al sola owe and de ec s pho on-induced showe s wi h an ene gy
h eshold o 250 ±110 GeV and an asymp o ic e ec i e de ec ion a ea o abou 15000
m2. A compa ison be ween he esul s o de ailed Mon e-Ca lo simula ions and da a is
p esen ed.
Da a se s aken in he pe iod Sep embe 1999 - Sep embe 2000 in he di ec ion o he
C ab pulsa , he ac i e galaxy 3C 454.3, he uniden i ied γ- ay sou ce 3EG 1835+35
and a “pseudo sou ce” we e analyzed o high ene gy γ- ay emission. E idence o a
γ- ay lux om he C ab pulsa wi h an in eg al lux o 2.2 ±0.4 (s a ) +1.7
−1.3(sys ) ×
10−9cm−2sec−1abo e h eshold and a signi icance o 4.5 σin a o al measu ing ime
o 7 hou s and 10 minu es on sou ce was ound. No e idence o emission om he
o he sou ces was ound.
Some di icul ies wi h he use o helios a ields o γ- ay as onomy a e poin ed ou . In
pa icula he e ec o ield-o - iew es ic ed o he cen al pa o a de ec ed ai showe
on he la e al dis ibu ion and iming p ope ies o Che enko ligh a e discussed. Upon
es ic ion he sp ead o he iming on o p o on induced showe s sha ply dec eases
and he econs uc ed di ec ion becomes biased owa ds he poin ing di ec ion. This is
shown o make e icien γ-had on sepa a ion di icul .
1
1 In oduc ion - aims and plan o he pape
Measu ing a mosphe ic Che enko adia ion is p esen ly he mos e ec i e way o de ec
cosmic γ- ays wi h p ima y ene gies be ween abou 100 GeV and 1 TeV [1]. In o de
o each low ene gy h esholds wi h echniques based on Che enko ligh , la ge mi o
collec ion a eas a e needed. GRAAL is an expe imen ha employs he la ge mi o a ea
o an exis ing owe sola -powe plan o his pu pose.
This pape b ie ly desc ibes he GRAAL de ec o and epo s esul s abou he de ec ion o
γ- ays om cosmic sou ces. In addi ion some gene al lessons we lea n abou he helios a -
ield app oach o γ- ay as onomy a e epo ed. In sec ion 2 he GRAAL de ec o is
desc ibed and compa ed o o he helios a - ield de ec o s o Che enko ligh . Sec ion 3
desc ibes he e en econs uc ion based mainly on he a i al ime o signals a he cen al
de ec o . Sec ion 4 ea s he Mon e Ca lo simula ion o he expe imen . Sec ion 5 explains
how he da a se used o he analysis o his pape was chosen om he o al se o all
aken da a. The da a educ ion p ocedu es—and he undamen al p oblems bese ing i —
a e explained in sec ion 6 and he esul s a e p esen ed in sec ion 7. Finally some concluding
ema ks a e o e ed in sec ion 8. A mo e de ailed epo abou hese esul s will be a ailable
in wo heses[2, 3].
2 The GRAAL de ec o
2.1 The CESA-1 helios a ield a he PSA
CESA-1 is a helios a ield comp ising o 300 s ee able mi o s o he no h o a cen al owe
loca ed wi hin he “Pla a o ma Sola de Alme ia”(PSA) a sola he mal-ene gy esea ch
cen e ope a ed by he Spanish CIEMAT. The PSA is loca ed in he dese o Tabe nas
(37◦.095 N, 2◦.360 W) abou 30 km om he ci y o Alme ia and he sea, a he oo hills o
he Sie a-Ne ada moun ains (heigh a.s.l. o 505 m). The 63 helios a s used o GRAAL
ha e a mi o a ea o 39.7 m2each and consis o 12 ec angula “ ace s” (sub mi o s)
wi h a sphe ical cu a u e ha a e “can ed” (adjus ed ela i e o he o e all ame) o a
oughly sphe ical o e all helios a shape. The beam sp ead unc ion o he helios a s has
a RMS o abou 0.25◦. Each helios a is indi idually s ee able wi h s epping mo o s ia a
cen al PC. Fo he pu pose o GRAAL a con ol p og am was de eloped ha allowed o
pe o m he special acking needed o he use o he ield o Che enko as onomy.
The helios a ocus he Che enko ligh o ai showe s om he di ec ion o po en ial
gamma- ay sou ces o so wa e adjus able “aiming poin s” in he cen al owe (see ig.1).
The so-called “con e gen iew”[4]— he poin ing o he helios a s owa ds a poin in he
a mosphe e co esponding o an a mosphe ic dep h o 230 g/cm2in he gene al di ec ion
o he po en ial sou ce o gamma ays—was always applied.
The ela i ely hin glass used o he helios a mi o s (4 mm hickness) —leading o a low
2
o e all hea capaci y—and he p oximi y o he ocean lead o equen dew o ma ion on
he mi o s in he win e . To p e en mic o d op o ma ion, all mi o s we e sp ayed e e y
second day wi h a ensid solu ion in he e ening using a specially cons uc ed sp ay ca .
This p ocedu e was ound o wo k well a e all mi o s had been cleand wi h sul onic acid
om aces o silicon gel—a common con aminan in glass p oduc ion.
2.2 De ec o se up
2.2.1 Seconda y op ics
Che enko ligh om ou g oups o helios a s (wi h 13,14,18,18 membe s, espec i ely) is
di ec ed on o ou single non-imaging “cone concen a o s” each con aining a single la ge-
a ea pho omul iplie ube (PMT). The ligh collec o s ha e he o m o unca ed Wins on
cones wi h an opening angle o 10◦. Each cone has a on diame e 1.08 m and a leng h
o 2.0 m. The cones a e housed in a special enclosu e ha is as ened o he ou side o he
cen al owe a he 70 m le el (see ig.2). Each cone is di ec ed on o a poin on he g ound
in he helios a ield and collec s he ligh om all helios a s which a e loca ed wi hin he
ellipse p ojec ed by he cone opening angle on he g ound (see ig. 3).
A he end o each cone, a six-s age 8 inch hemisphe ical PMT, op imized o ope a ion
unde high backg ound ligh le els (EMI 9352KB) is si ua ed. The ubes we e ypically
ope a ed wi h abou 1300 - 1600 V a a gain o abou 8000. The signals we e ansmi ed
ia AC coupling o one as ampli ie di ec ly adjacen o he PMT and a second one nea
he da a acquisi ion elec onics wi hin he owe . These ampli ie s ha e a bandwid h o
abou 350 MHz and a gain o abou 15 each. The inal FWHM wid h o Che enko pulses
is abou 3.6 ns, and mainly de e mined by pa h leng h di e ences wi hin he PMT.
The incoming ligh om an ai showe consis s o a ain o pulses om he di e en
helios a s, usually ully sepa a ed by pa hleng h di e ences. The a i al ime and ampli ude
o each helios a can hus be de e mined wi h a lash-ADC in a sequen ial mode ( ig.4).
2.2.2 T igge logic
Two comple ely independen igge s a e used. Fo he “sequence igge ” a e a disc im-
ina ed signal abo e 30 mV a ga e o 40 ns leng h is opened a e a delay o 20 ns. I a
u he signal is de ec ed du ing ga e du a ion, ano he ga e o 40 ns is opened wi h a delay
o 20 ns. I a hi d signal is de ec ed in his second ga e an e en - igge ga e o 200 ns is
opened. I he i s and second cone ha e a coinciden e en - igge he inal e en igge
is o med.
Fo he “cha ge(q) igge ” a iming-ampli ie in eg a es he signal wi h an exponen ial ime
scale o 100 (200) ns o Cone 1+2 (3+4). The in eg a ed signal is ed in o a disc imina o
in all ou cones and opens a coincidence ga e o 200 ns du a ion i a p ese h eshold is
su passed (see also ig.16, whe e he Mon e Ca lo simula ion o his igge is discussed).
3
Figu e 1: Scheme o he expe imen seen om he side, no h is o he igh . The Che enko
ligh o a schema ic ai showe (no o scale wi h espec o he ield) is concen a ed by he
helios a s o he CESA-1 ield o a ocus a he cen al owe . A dedica ed pla o m moun ed
a he ou side o he owe a he 70 m le el houses ou Wins on cones which ecei e ligh
om 13 - 18 helios a s in he ield. The da a-acquisi ion elec onics is loca ed inside he
owe .
The singles a e o his in eg a ed signal is he “q- a e” ( able 2 - 5). A majo i y coinci-
dence equi emen o “3 ou o 4 cones” is equi ed o he inal e en igge .
Bo h igge s a e always in a logical OR mode in da a aking. The e en a e o he
“sequence igge ” depends sensi i ely on he incoming di ec ion o he showe bu is el-
a i ely insensi i e on he le el o nigh -sky backg ound induced backg ound ligh (NSB).
The “cha ge igge ” is mo e s ongly in luenced by he NSB, bu igge s on coinciden
signals independen o de ailed hypo heses on he a i al- ime s uc u e.
2.2.3 Da a eadou
GRAAL achie es a good ime esolu ion because he e exis only ou sho cables ha un
exclusi ely wi hin he pla o m enclosu e om he pho omul iplie s o he da a acquisi ion
elec onics. We egis e all ou pulse ains in only one Digi al Oscilloscope (Le C oy LC
564A) wi h a bandwid h o 1 GHz and a ime bin o 500 ps. This ensu es ha he FWHM
o indi idual pulses o abou 3.6 ns is negligibly inc eased by elec onics e ec s. The digi al
scope is ead ou in sequence mode o e a GPIB in e ace in o a PC, eaching a speed o
abou 260 “wa e o ms”/sec (i.e. 1000 ime bins o 0.5 ns wid h wi h 1 by e each), which
is su icien o a dead ime below 10 % o ou mas e igge a e which always emains
below 5 Hz and is ypically abou 2-3 Hz (each igge con aining ou wa e o ms).
4
1 m
Figu e 2: Uppe plo : Side iew o he de ec o pla o m a he 70 m le el o he cen al
owe . Two o he ou Wins on cones poin ing owa ds hei espec i e helios a sub ields
wi h he la ge-a ea PMTs a he ends a e ske ched. The wall o he cen al owe is a he
le wi h a manhole o en e he pla o m. Lowe plo : View om abo e, all ou Cones a e
shown, he hal ci cle is he wall o he cen al owe .
2.2.4 Calib a ion
The ime and ampli ude calib a ion o ou se up is pe o med using blue LEDs (Nichia
NSPB 500, maximal ou pu a 470 nm) wi h a calib a o module ha is as ened a he
window o he Wins on cones. The amoun o ligh emi ed by he indi idual LEDs is
de e mined wi h a Quan acon RCA C31000 (a pho omuliplie yielding a well sepa a ed
single pho oelec on (p.e.) peak) ha was p e iously calib a ed by de e mining i s single
p.e. peak and luc ua ion beha iou . The LED ope a ing ol age is adjus ed so ha one
LED pulse co esponds o abou 100 p.e. These LED pulses a e egula ly used in each
un o e i y ime and ampli ude calib a ion. In addi ion a LED module wi h highe o al
ligh ou pu shines on o he helios a ield. When he helios a s a e b ough in o a “back
e lec ion” posi ion, he e lec ed LED pulses a e used o e i y he geome y and check he
5

64 m
Cone 2
Cone 1
Cone 3
Cone 4
Figu e 3: Scheme o he de ec ion geome y seen om abo e, no h is o he op o he page.
The small ci cle is he owe , he iled double squa e symbolize he helios a s o CESA-1 in
he 2nd o 7 h ow om he owe . The ligh om one o he ou g oups o helios a s used
in GRAAL - indica ed by he ellipses - is concen a ed in o one o he ou cones. The cone
numbe ing indica ed is used h oughou he ex .
mi o quali y.
The iming and gain p ope ies o he elec onics chain we e calib a ed on-line wi h a PC-
con olled Phillips Scien i ic PS7120 cha ge injec ion module. Cha ge pulses wi h p ope ies
simila o PMT pulses and di e en ampli udes we e injec ed di ec ly a e he PMT in
each un.
2.2.5 Remo e ope a ion
All ope a ions (like opening o he doo , high- ol age con ol e c.) a he cen al ecei e
and he acking o he helios a ield a e unde emo e con ol ia he in e ne . Va ious
en i onmen al pa ame e s like humidi y, ambien ligh , wind speed, a es e c. a e checked
by he da a-acquisi ion compu e . Unde condi ions ha indica e some mal unc ion, a
physicis on shi is phoned by he PC and can check all pa ame e s and images o web
came as, emo ely. Fo he ope a ion o he helios a ield and eme gencies only he egula
nigh -ope a o o he PSA is on-si e in all obse a ion nigh s.
2.3 Di e ences o he basic “CELESTE” e sus “GRAAL” cen al- ecei e
app oach
A e ea ly es s o he use o helios a ields o γ- ay as onomy [5] he basic idea o
T¨ume [6, 7] o image he Che enko ligh o one helios a o a single pho omul iplie has
been wo ked ou in echnical de ail in he p oposal o he CELESTE expe imen [4]. I was
6
0 1000800600400200 Time (ns)
0 1000800600400200 Time (ns)
Signal (mV) Signal (mV)Signal (mV)Signal (mV)
0 1000800600400200 Time (ns)
0 1000800600400200 Time (ns)
400
1000
500
40
20
0
0
200
400
0
0
200
Cone1
Cone2
Cone3
Cone4
Figu e 4: The signal heigh in mV a e ampli ica ion eco ded in all ou cones om one
ypical ai showe is displayed as a unc ion o ime. The igge occu s a 500 ns. The y-gain
depends on ampli ude, a 100 mV one mV co esponds o ypically 0.25 pho o elec ons.
Each peak co esponds o he Che enko -ligh lash o he showe e lec ed by a di e en
helios a . The dis ibu ion o ligh in ensi y on he g ound wi hin he ield o iew o he
cones is e y une en, no e he di e en y-scales.
hen p o en echnically a he Themis helios a a ay [8]. Two o he helios a - ield expe i-
men s “STACEE”[9, 10] and “Sola 2”[11] ollow his basic design. Recen ly CELESTE[12]
and STACEE[13] epo ed he de ec ion o VHE γ ays. The majo di e ences be ween his
well documen ed me hod o de ec ai showe s wi h helios a a ays and he “non-imaging”
p inciple o GRAAL—which collec s he ligh om 13-18 helios a s in a Wins on cone on o
asingle la ge-a ea pho omul iplie —a e desc ibed in he ollowing.
•The mos impo an d awback o he non-imaging app oach o GRAAL is ha he nigh -
sky backg ound is highe oughly by he numbe o helios a s iewed by one cone. This
esul s in a ypical expec ed backg ound o 8-10 p.e./ns in GRAAL, compa ed o 0.7 p.e./ns
in CELESTE. The ha dwa e ene gy h eshold o he de ec ion o γ- ays in p inciple achie -
able wi h he same mi o a ea used is abou 4 imes highe in GRAAL. Fo pulses a abo e
h eshold he pe o mance o he wo app oaches is no expec ed o be e y di e en be-
cause a simila amoun o Che enko ligh is ga he ed by GRAAL and CELESTE.
•The ad an age o he non-imaging app oach is i s g ea e simplici y leading o sa ings by
abou a ac o 5-10 in ha dwa e cos s. The p esence o only ou da a-acquisi ion channels
makes au oma iza ion and emo e con ol mo e easible, leading o compa able sa ings in
ope a ion cos s. In i s p esen con igu a ion GRAAL no mally uns unde emo e con ol
wi h only a PSA ope a o (who is p esen o main enance o he acili ies independen ly
o GRAAL) on-si e. The small numbe o channels allows o use lash-ADCs wi h a ime
7
esolu ion o 0.5 ns/bin, highe han any o he Che enko expe imen .
•In CELESTE he angula ield-o - iew in he sky o each PMT is designed o be cons an
a 10 m ad ( ull angle). In GRAAL his is impossible because he con ibu ing helios a s’
dis ance om he collec ing cone a ies. This ield-o - iew he e o e a ies be ween 6.5 and
12.1 m ad. I is no easy o de e mine he “op imum” alue o he ield o iew since i
depends on se e al di e se ac o s. The o al accep ance has o be de i ed om de ailed
Mon e Ca lo simula ions e en in case o a ixed accep ance. The e o e his di e ence seems
o li le impo ance.
•Because he non-imaging app oach o GRAAL equi es ha g oups o di ec ly adjacen
helios a s in he ields a e chosen, i s con igu a ion is mo e compac . In GRAAL 63 he-
lios a s ha co e an a ea o abou 160 ×80 m2a e used, whe eas CELESTE p esen ly
uses 40 helios a s ha co e an a ea o 240 ×200 m2, i.e. he sampling densi y is abou
a ac o 5 lowe . F om he Mon e-Ca lo simula ions i seems ha wi h a es ic ed ield
o iew he i egula s uc u e o he ligh pool in had onic showe s ends o be mo e p o-
nounced a la ge dis ance scales, so a mo e ex ended a ay ends o be ad an ageous o a
possible γ-had on sepa a ion.
•In he non-imaging app oach i is impossible o a oid a empo al o e lap o he signal
om ce ain helios a s depending on he poin ing di ec ion. This educes he numbe o
imes/ampli udes usable in he econs uc ion by abou 20%. When he inciden di ec ion
lies no hwa d ( his is he case o he sou ce 3EG 1835+35 a he loca ion o GRAAL), he
o e lap becomes s onge leading o a subs an ial dec ease in he quali y o econs uc ion.
On he posi i e side, calib a ion is easie when signals om se e al helios a s a e measu ed
in he same PMT.
3 E en econs uc ion
3.1 So wa e- igge h eshold
The nigh -sky backg ound (NSB) RMS luc ua ion was es ima ed om a po ion o he
lash-ADC eco ded aces ha do no con ain Che enko signals, o each e en and cone
indi idually. The a i al ime o all de ec ed signals wi h an ampli ude exceeding n ×
σNSB was de e mined om he eco ded ull pulse shape in he ela ed lash-ADC. The
pa ame e n was chosen o be ypically be ween 5 and 7. These a i al imes we e co -
ec ed o pa h leng h di e ences in cables and wi hin PMTs wi h he online-calib a ion
(sec ion 2.2.4) and we e hen used o econs uc he iming showe on o he indi idual
e en s. A i al imes close o each o he han 6 ns we e excluded o a oid any bias om
o e lapping pulses. Signals ha sa u a ed any channel we e also excluded om u he
analysis. Only he NREMAIN emaining signals we e used in he u he analysis.
Be o e u he analysis a so wa e- igge h eshold was applied. In o de o allow a mean-
ing ul econs uc ion o showe pa ame e s NREMAIN ≥5 was equi ed. n was chosen a
8
Zeni h (deg)
Numbe o e en s
Azimu h (deg)
Zeni h (deg)
Numbe o e en s
Azimu h (deg)
Figu e 5: P ojec ions o he numbe o showe s as a unc ion o showe di ec ions as econ-
s uc ed om he iming da a. Shown is de ia ion o he econs uc ed di ec ion om he
poin ing di ec ion on he ele a ion-axis (le wo panels a. and c.) and azimu h-axis ( igh
wo panels b. and d.). The o igin hen co esponds o he poin ing di ec ion as de e mined
by he o ien a ion o he helios a s. Two componen s a e appa en : a peak nea he o igin,
and a “ la backg ound” co esponding o e en s mis econs uc ed in di ec ion (see ex ).
The da a sample comp ises o 32 hou s o ON-sou ce ime on he C ab pulsa (uppe panels
a. and b.) and an equal amoun o OFF-sou ce ime (lowe panels c. and d.) aken unde
a iable wea he condi ions in he season 1999/2000. The “Gaussian plus linea unc ion”
i is pe o med o each subsample. I is seen ha he Gaussian - co espondig o success ully
di ec ion econs uc ed e en s - is always cen ed wi hin <0.05◦.
a alue as low as possible, be o e a la ge numbe o NSB induced “ ake” signals we e ound
o en e he sample. The lowes possible alue o n was ound o depend on he sou ce
posi ion somewha , due o he a ying empo al o e lap o signals in he ace. The inal
choice was: n =5(7,9,7) o he C ab (3C454, 3EG+1835, pseudo sou ce) sample. This
so wa e h eshold also equalized he e ec o he NSB on he econs uc ion. A highe
le el o NSB σNSB leads o a co espondingly highe so wa e h eshold. This is expec ed
o co ec o he e ec o a lowe ha dwa e igge h eshold and dec eased econs uc ion
e iciency wi h highe NSB o i s o de . The choice o n in he analysis o MC da a was
di e en and is explained in sec ion 4.3.
9
0
50
100
150
200
250
300
0 10 20 30 40 50 60 70
Numbe o peaks
Numbe o showe s
Figu e 8: Numbe o showe s wi h a gi en numbe o peaks iden i ied in all ou eco ded
iming aces. The ull (dashed) line is o MC γs (p o ons), and he do ed line o ex-
pe imen al da a aken unde simila inciden angles. The o al numbe o MC showe s was
no malized o he expe imen al da a o compa ison.
hese ails ha inc ease he mean o he expe imen al lsq2
dis ibu ion.
4.4.3 To al-cha ge spec um
Fig.12 displays he “ o al cha ge” spec um bo h in da a and Mon e Ca lo. The “ o al
cha ge” is de e mined by in eg a ing he a ea unde all peaks de ec ed in he lash-ADC
aces adding all ou cones in one e en . Fa abo e h eshold he expe imen al spec um
ollows a powe law wi h a di e en ial index o abou -1.6—which is much la ge han ha
o he p ima y spec um o -2.7. The eason o his is a e y la ge sca e in he co ela ion
o o al cha ge and ene gy. The Mon e Ca lo simula ed spec um looks quali a i ely simila
o he expe imen al da a bu ollows a sligh ly s eepe index o abou -1.9. One eason o
his is ha a abo e he h eshold he cu o in simula ed p o on ene gy a 10 TeV is
al eady expec ed o ha e a s eepening e ec on he MC spec um.
4.5 E ec s o he small ield-o - iew on econs uc ed showe p ope ies
To gain he ad an age o using many la ge mi o s wi h only one cen al de ec o helios a s
need o ha e a ocal leng h abou a ac o 20 - 30 la ge han hose o he elescopes used o
he imaging o VHE γ- ay showe s. Fo space easons in he cen al owe he ligh de ec o
a he ocus canno be scaled up by such eno mous ac o s. Mo eo e he cons uc ion o an
imaging came a o each helios a would be p ohibi i ely expensi e. These wo ac o s o ce
a c ucial comp omise in Che enko de ec o s using helios a ields: he ield-o - iew has o
be chosen abou one o wo o de s o magni ude smalle in solid-angle han in adi ional
16

0
50
100
150
200
250
0 10 20 30 40 50 60 70
Numbe o iden i ied peaks
Numbe o showe s
Figu e 9: Numbe o showe s wi h a gi en numbe o peaks ha we e a ached o indi idual
helios a s and we e used in he inal de e mina ion o he showe di ec ion. The ull (dashed)
line is o MC γs (p o ons), and he do ed line o expe imen al da a aken unde simila
inciden angles. The o al numbe o MC showe s was no malized o he expe imen al da a
o compa ison.
Che enko elescopes.
Ou Mon e Ca lo simula ions show ha abou 60% o he Che enko ligh o showe s
induced by gamma- ays wi h small ene gies (100 GeV) is collec ed in he GRAAL se up,
a numbe ha is accep able when aking in o accoun he la ge mi o a ea. Ne e heless,
we ind ha he angula es ic ion has se e al disad an ageous e ec s.
4.5.1 S uc u e o iming on o p o on induced showe s
I is well known ha he a i al imes in p o on-induced showe s ha e a much wide sca e
a ound he mean a i al ime han in gamma-induced showe s due o hei mo e i egula
de elopmen in he a mosphe e[19]. The expe imen al de e mina ion o his sca e has
been p oposed o be an e icien me hod o gamma/had on sepa a ion[20]. Fig.13b. shows
he s uc u e o he showe on o a ypical gamma, ig.14b. o a ypical p o on showe
om he Mon e Ca lo simula ion wi hou simula ion o he de ec o . The la ge sca e o
he p o on showe is e iden . In panel c. o hese igu es he showe on is shown wi h
a es ic ion on he inciden angle o he pho on. Only pho ons wi h an inciden angle
di e en by less han 0.3◦ om he di ec ion poin ing owa ds he showe maximum om
a posi ion on g ound we e e ained. This es ic ion has a e y simila e ec o he small
ield o iew dicussed abo e. Upon angula es ic ion he showe on na ows bo h o
p o ons and gammas, bu he e ec is s onge o he p o ons. Panels a. o hese igu es
demons a e he cause o his beha iou . The es ic ed ield o iew mainly p e en s he
17
0
50
100
150
200
250
10 -2 10 -1 1 10
lsq 2
Numbe o showe s
Figu e 10: The dis ibu ion o lsq2
. The ull (dashed) line is o MC γs (p o ons), and he
do ed line o expe imen al da a aken unde simila inciden angles. The o al numbe o
MC showe s was no malized o he expe imen al da a o compa ison.
de ec ion o Che enko pho ons emi ed a below he maximum a abou 11 km heigh .
De ia ions om he ideal sphe ical iming- on a e mainly due o he deeply pene a ing
pa o he showe . P o ons a e mo e pene a ing and a e he e o e mo e a ec ed by he
angula es ic ion.
The o al e ec is ha wi h a small ield o iew, p o ons and gammas ha e i ually
iden ical, nea ly sphe ical showe on s, wi h e y li le sca e , displayed and compa ed
wi h expe imen al da a in ig.11. This makes e icien gamma/had on sepa a ion wi h
iming me hods in helios a ields all bu impossible (see igs.11,15).
4.5.2 Recons uc ed di ec ion o p o on induced showe s
Ano he impo an me hod o disc imina e gamma- and p o on induced showe s is o ex-
clude all showe s ha do no a i e om he sou ce di ec ion wi hin he angula esolu ion
as de e mined wi h a i o he iming on . Wi h a es ic ed ield o iew he e is a bias
o he showe di ec ion econs uc ed om iming in o ma ion owa ds he sou ce di ec ion
(see ig.15). The ield-o - iew “selec s” a pa o he showe which lies owa ds he showe
maximum o a showe a i ing om he sou ce di ec ion. The iming- i hen inds he
di ec ion o his subpa o he showe , which is biased owa ds he sou ce di ec ion. In
Mon e-Ca lo simula ions o p o on induced showe s we ound ha he mean di e ence be-
ween ue showe di ec ion and econs uc ed showe di ec ion is 0.71◦±0.002◦(s a is ical
e o ), whe eas he mean di e ence be ween sou ce di ec ion and econs uc ed showe di-
ec ion is only 0.44◦±0.002◦(s a is ical e o ). This bias dec eases he ac ion o p o on
showe s which can be excluded due o hei angula dis ance o he sou ce di ec ion.
18
0
500
1000
1500
2000
2500
3000
x 10 3
-20 -15 -10 -5 0 5 10 15 20
Time de ia ion (ns)
Numbe o peaks
Figu e 11: The de ia ion o measu ed a i al imes om he inal i ed sphe ical showe
on o MC γs ( ull), p o ons (dashed) and expe imen al da a (do ed). The isible sha p
educ ion o e en s wi h a ime de ia ion somewha smalle han 5 ns is due o he ac
ha he econs uc ion p og am allows he exclusion o 3-5 peaks wi h a de ia ion om he
showe on la ge han 5 ns (see ex ) om he inal i .
4.5.3 Ene gy esolu ion
The es ic ion in he ield o iew dec eases he ene gy esolu ion p og essi ely o la ge
showe s, because he ac ion o he showe image seen canno be in e ed. We ind ha nea
ou ene gy h eshold o gamma ays he esolu ion de i ed om choosing he o al cha ge
eco ded in all 4 cones as simple p ima y-ene gy es ima o is abou 110% and wo sens
apidly o highe ene gies.
5 Da a selec ion
5.1 De ec o condi ion
Only nigh s in which all ou de ec o channels and he helios a s in he ield we e unc-
ioning no mally acco ding o he eco ded moni o iles we e chosen o u he analysis.
5.2 Me eo ological selec ion
I was ound ha he econs uc ion quali y depends on he a mosphe ic ansmission. Fo
example in nigh s which we e isibly hazy wi h a high ela i e humidi y abo e 80% (a
ela i ely equen nigh ly wea he condi ion a he PSA), he o al igge a e was low,
he a io o well econs uc ed e en s o e en s wi h a mis econs uc ed angula di ec ion
(called “PT” below) was educed by up o a ac o 2 and he lsq2
o he i o he iming
19
To al cha ge (mV)
10 -3
10 -2
10 -1
104105
Figu e 12: The numbe o showe s as a unc ion o “ o al in eg a ed cha ge” in all 4 Cones in
one showe . The dashed line a e expe imen al da a, he ull line is om he MC simula ion.
The cu es we e no malized o he same numbe o showe s. The x-axis is in uni s o
summed lash-ADC ampli udes in mV.
on signi ican ly inc eased. This is p obably he esul o selec i e abso p ion, by which
Che enko ligh om he deeply pene a ing pa o he ai showe , wi h inc eased empo al
luc ua ions, domina es he eco ded signal. As γ-induced showe s de elop mainly in he
uppe a mosphe e a selec ion o da a wi hou selec i e abso p ion is impo an . Besides
a ela i e humidi y below 70% and gene ally clea skies we equi ed he ollowing c i e ia
om he econs uc ed da a o a gi en nigh . The pa ame e limi s o each indi idual
poin ing di ec ion we e chosen such ha a se o “good” nigh s —de ined as showing ai ly
cons an pa ame e alues—was e ained. The pa ame e limi s hus sligh ly a ied o
di e en poin ing di ec ions.
Fi s a cu o exclude uns able wea he condi ions was applied. Fo his he ac ion o
e en s wi h a econs uc ed angle a om he sou ce di ec ion was chosen.
Condi ion 1: 0.95 <RO <1.05
RO = (Numbe o e en s wi h econs uc ed di ec ion >3◦ om poin ing di ec ion ON
sou ce) / (Numbe o e en s wi h econs uc ed di ec ion >3◦ om poin ing di ec ion OFF
sou ce)
The o he wo un-cu c i e ia a e mean o exclude nigh s wi h low a mosphe ic ansmis-
sion.
Condi ion 2: Ra e a e all so wa e cu s in OFF sou ce di ec ion >50/min
Condi ion 3: PT >0.8
PT = (Numbe o e en s wi h econs uc ed di ec ions <1◦ om poin ing di ec ion OFF
sou ce)/ (Numbe o e en s wi h econs uc ed di ec ions >3◦ om poin ing di ec ion OFF
20
0
250
500
750
1000
1250
1500
1750
2000
2250
0 10000 20000
Heigh (m)
Numbe o C-pho ons
0
5
10
15
20
25
30
0 100 200
102103104
Co e dis ance (m)
A i al ime (ns)
0
5
10
15
20
25
30
0 100 200
102103104
Co e dis ance (m)
A i al ime (ns)
Figu e 13: Time s uc u e o a ypical gamma- ay ini ia ed showe . b. The a i al ime
as a unc ion dis ance om he co e in me e s o a ypical gamma showe . The shading
is p opo ional o he Che enko -pho on densi y. c. Same as b. bu only hose pho ons
wi h an a i al di ec ion wi hin 0.3◦ om he di ec ion owa ds he showe maximum om
a posi ion on he g ound a e displayed o he same showe s. a. Numbe o Che enko -
pho on emi ing elec ons in he showe as a unc ion o heigh a.s.l.
sou ce)
These “me eo ological cu s” a e se e e unde he wea he condi ions a he PSA. In he
da a sample on C ab in Feb ua y/Ma ch 2000 only 22% o all da a aken on he C ab
pulsa passed all cu s.
6 Da a educ ion— he p oblem o di e en condi ions in he
sou ce and OFF-sou ce egion
A undamen al p oblem o all Che enko expe imen s—specially o hose a emp ing o
de ec an excess due o gamma- ays in he o al a e—is he ac ha he nigh -sky back-
g ound be ween ON- and OFF-sou ce di e s in gene al. This can in luence he coun ing
a e and analysis e iciency in a ious ways. This p oblem is mos c i ical o he helios a -
a ay based expe imen s because hey aim o wo k wi h a igge h eshold nea o andom
luc ua ions o he nigh -sky backg ound in a single channel. We discuss he obse ed di -
e ences in he ON- and OFF-sou ce egion in sec ion 6.1. Sec ions 6.2 and 6.3 discuss he
e ec o a di e ence in he ON- and OFF-sou ce in ensi y o he NSB on he o al a e
and he econs uc ion, espec i ely. Sec ion 6.4 desc ibes he me hod we inally chose o
calcula e an excess o e en s in he ON-sou ce di ec ion.
6.1 De ailed compa ison o condi ions ON and OFF sou ce
Fo he coun ing condi ions chosen in he 1999/2000 season, he “q- igge ” (see sec ion
2.2.2) leads o andom e en igge s due o nigh -sky noise. Because his andom a e is
21

0
200
400
600
800
1000
1200
1400
0 10000 20000
Heigh (m)
Numbe o C-pho ons
0
5
10
15
20
25
30
0 100 200
10 102103
Co e dis ance (m)
A i al ime (ns)
0
5
10
15
20
25
30
0 100 200
1 10 102
Co e dis ance (m)
A i al ime (ns)
Figu e 14: Time s uc u e o a ypical p o on ini ia ed showe . The panels show he same
quan i ies as in he p e ious ig.13. No e ha he p o on emi s a much smalle ac ion o
ligh wi hin he es ic ed ield-o - iew because o i s la ge angula ex ension.
e y sensi i e o he nigh -sky backg ound, sligh ly highe NSB le els in OFF (as obse ed
o all po en ial sou ces see ables (2 - 5)) p oduce a highe e en - igge a e in OFF (see
en y “ aw e en s” in ables 2-4) The andom a e was calcula ed om he eco ded single
a es. A discussion o he o al a e a e a co ec ion o he andom igge and o he small
e ec s is gi en below in sec ion 6.2. F om es da a i was shown ha he econs uc ion
o he iming showe on always ails o andom- igge e en s, so ha he e en numbe
“a e econs uc ion” (“ ec. e en s” in ables 2 - 4) is expec ed o be ee om nigh -sky
backg ound induced andom igge s.
All 4 sou ces discussed in his pape show a sligh ly highe NSB in he OFF-sou ce egion.
This e ec is mos p onounced o he sou ce 3C 454, whe e— om he da a epo ed in
able 3— he cu en (q- a e) is 13 (25)% highe in OFF han in ON. Howe e , he measu ed
RMS luc ua ion is only 0.4% highe in OFF han in ON and his di e ence is smalle o
he o he sou ces (0.04% o he C ab pulsa ). By measu ing he andom noise in comple e
da kness, we de e mined a cons an nigh -sky backg ound independen noise le el wi h a
RMS o 0.8658. Sub ac ing his cons an noise quad a ically om he o al noise we ge
he con ibu ion om he NSB alone (numbe in b acke s in hi d column o able 2 - 5).
Fo he sou ce wi h he la ges di e ence in noise le el, he NSB-induced componen di e s
in ON- and OFF-sou ce posi ion by abou 2.5%, so ha he di e ence in b igh ness a he
wo posi ions can be es ima ed o be abou 5%.
An e ec ha is e y di icul o emo e is a sligh expec ed educ ion in he igge h esh-
old due o a highe NSB. Due o luc ua ions, smalle e en s can c oss he igge h eshold.
The opposi e e ec — ha la ge e en s a e dec eased in ampli ude due o luc ua ions and
ail o c oss he h eshold—happens less o en due o a CR spec um ha s eeply alls wi h
ampli ude.
This e ec was s udied by calcula ing he o al mean cha ge o all ON e sus all OFF
sou ce e en s ( see ables 2 - 5 en y “mean q”). I small e en s a e p e e ed, he o al
22
Showe di ec ion o mo ion
Res ic ed ield o iew
Res ic ed ield o iew
sub
Gamma induced showe
P o on induced showe
Figu e 15: Ske ch o illus a e he e ec o a small- ield o iew - necessi a ed by he
helios a - ield app oach - on he de e mina ion o he iming s uc u e. A gamma- ay in-
duced showe is symbolized in he le pa o he igu e and a p o on induced one wi h
a sligh ly di e en incidence di ec ion on he igh . The p o on showe is spa ially mo e
ex ended and symbolized as a collec ion o small sub showe s. The es ic ed ield o iew
“p ojec s” ou sub showe s in he cen al pa o he showe ou o he mo e ex ended p o-
on showe . O he mo e pene a ing and la e ally ex ended sub showe s - ha inc ease he
luc ua ion in he iming on - do no con ibu e o he ligh de ec ed wi hin he es ic ed
ield o iew. One sub showe wi h an incidence di ec ion biased owa ds he poin ing di ec-
ion (symbolized by he label “sub”) is p e e en ially de ec ed and hus biases econs uc ed
di ec ions owa ds he poin ing di ec ion.
mean cha ge should be smalle by a ce ain ac o bwi h inc eased NSB. The o al a e
should be inc eased by a ac o o oughly 1.4
b o ou se up.
I can be seen om he esul s in he ables (2 - 4) ha wi hin he s a is ical e o o
ypically somewha less han a pe cen he o al mean cha ge is he same o all sou ces.
Howe e , wi hin his e o a signi ican educ ion o he h eshold - one which would p oduce
a educ ion in ON-OFF a es o he same o de o magni ude as an expec ed signal om
he c ab nebula - canno be excluded in his way.
6.2 E ec o NSB di e ences on o al igge a e - Mon e-Ca lo simu-
la ion
The e ec o he NSB di e ences on he o al igge a e was simula ed by aising he
amoun o andom noise by 5% o e i s usual alue. The de ec o Mon e Ca lo models he
elec onic pulse shaping and he esponse o he disc imina o in de ail (see ig.16), and so
he e ec i e change in h eshold, due o he inc eased noise le el could be deduced o be
23
Table 2: Cu en (mean o 4 Cones), q- a e: single igge a e o cha ge in eg a ing
channel (mean o Cone 1+2), σNSB: RMS luc ua ion o he measu ed NSB (in lash-
ADC uni s) in he i s 100 channels (be o e signal), numbe in b acke s is NSB induced
backg ound alone, log(mean q): base-10 loga i hm o mean ne -cha ge (in lash-ADC
uni s) o all e en s in sample, aw e en s: all ha dwa e- igge ed e en s which aces we e
eco ded, ec. e en s: numbe o e en s a e angula econs uc ion and so wa e igge ,
cen . e en s: no malized numbe o e en s in cen al angula egion (wi hin 0.7 deg ees
o poin ing di ec ion), calcula ed as explained in sec ion 6.4. Rows a e o he samples
wi h poin ing owa ds he C ab pulsa (“ON”) and on a sky posi ion (“OFF”) wi h a igh
ascension 2.625 deg ees la ge han in he ON di ec ion. The o al da a- aking ime ON
was 430 minu es wi h an equal amoun o OFF ime.
cu en [µA] q- a e[kHz] σNSB [ADC uni s] log(mean q)
ON 19.0 ±0.4 1.35 0.9493 (0.3893) 2.940 ±0.004
OFF 19.3 ±0.3 1.49 0.9497 (0.3902) 2.937 ±0.004
EXCESS -0.3 -0.14 -0.0004 (-0.0009) 0.003 ±0.006
aw e en s ec. e en s cen . e en s
ON 68702 33384 9415
OFF 75198 33056 8678
EXCESS -6496 ±379 328 ±258 737 ±165
abou 6±2%. Ex apola ing, we deduce an expec ed spu ious excess a he OFF sou ce
posi ion o he sou ce wi h he la ges di e ence in ON and OFF noise (3C454, see able
3) o abou 1%, his co esponds o abou 1.4 σs a in his case. As he di e ence in he
noise le els be ween ON and OFF is smalle in he case o he o he 3 sou ces discussed in
his pape , his e ec does no ye con ibu e signi ican ly. Howe e , i is clea ha a e y
ca e ul co ec ion o i becomes necessa y when he a ailable s a is ics g ows.
6.3 E ec o NSB di e ences on econs uc ion - So wa e padding
Finally a di e ence in NSB leads o a sligh ly di e en noise le els in ON and OFF da a.
Fo example o he C ab da a he RMS noise in he ON-sou ce da a was ound o be abou
0.5% smalle han in he OFF-sou ce da a. The e ec o his di e ence on he econs uc ion
p ocedu e was s udied by a i icially adding noise a he so wa e le el (“so wa e padding”).
Fig.17 demons a es ha he ac ion o e en s nea he sou ce di ec ion (“PT” o sec ion
5.2) dec eases wi h inc easing NSB, bu ha he e ec is impo an only a ela i ely la ge
inc eases on he o de o a ew pe cen . F om he esul s shown in igu e 17 i was de i ed
ha an inc ease o RMS noise by 1% dec eases he o e all econs uc ion e iciency by
abou 0.4% and he peak o ail a io PT (sec ion 5.2) by 0.8%. This e ec emains small
24
Table 3: En ies as in able 2 o he samples wi h poin ing owa ds he adio sou ce 3C
454.3 (“ON”) and on a sky posi ion (“OFF”) wi h a igh ascension 2.625 deg ees la ge
han in he ON di ec ion. The o al da a- aking ime ON was 550 minu es wi h an equal
amoun o OFF ime.
cu en [µA] q- a e[kHz] σNSB [ADC uni s] log(mean q)
ON 17.7 ±0.4 3.1 0.9505 (0.3922) 3.119 ±0.003
OFF 20.3 ±0.3 4.1 0.9540 (0.4006) 3.113 ±0.003
EXCESS -2.6 -1.0 -0.0035 (-0.0084) 0.006 ±0.004
aw e en s ec. e en s cen . e en s
ON 42516 30570 7525
OFF 44949 30889 7625
EXCESS -2433 ±296 -319 ±248 54 ±141
o he obse ed ac ional di e ences o he RMS NSB-noise in ON and OFF (on he o de
a ew en hs o a pe cen a maximum (see able 2- 5)), and was neglec ed in he p esen
analysis. I mus be no ed ha so wa e padding is no a pe ec simula ion o he eal
condi ions, because he in luence o he NSB on he ha dwa e- igge condi ion—which can
in p inciple in luence showe p ope ies and econs uc ion—is no simula ed.
6.4 Calcula ion o he excess
To a oid he p oblem men ioned in sec ion 6.2 we chose a me hod ha no malizes any
excess o he a io o ON- and OFF-sou ce e en s o he inal esul s epo ed in he nex
sec ion.
The no malized excess EXCESSnwas calcula ed acco ding o he ollowing equa ion:
EXCESSn= ONin −OFFin ON
OFFou
(3)
He e (ON,OFF)in s ands o he numbe o e en s wi hin 0.7◦ om he sou ce and OFF-
sou ce di ec ion, espec i ely whe eas (ON,OFF)ou s ands o he numbe o e en s wi h di-
ec ions de ia ing mo e han 2◦ om he sou ce di ec ion. The s a is ical e o o EXCESSn,
ERRnwas calcula ed acco ding o:
ERRn=
ONin + OFFin ×ON
OFF2
ou
+

(1 + ON
OFF ou )ONou ×OFF2
in
OFF2
ou



0.5
(4)
7 Resul s
25
Angula dis ance (deg)
Numbe o showe s
Angula dis ance (deg)
Numbe o showe s (on-o )
Figu e 18: The uppe plo (a.) shows he numbe o e en s as a unc ion o angula dis ance
o econs uc ed di ec ion om sou ce di ec ion o ON-sou ce e en s ( ull line) and OFF-
sou ce e en s (dashed line). No no maliza ion o any kind was applied o his plo . The
lowe plo (b.) shows he di e ence ON - OFF, no malized o he numbe o e en s in
he ou e angula egion, acco ding o eq.3 Da a om he C ab pulsa aken unde good
me eo ological condi ions acco ding o he cu s discussed in sec ion 5.2 we e used. The
s a is ical e o s o he indi idual bins a e shown.
Re e ences
[1] F.K enn ich, TeV Gamma-Ray As onomy in he new Millennium, P oc. 7 h Taipei
As ophysics Wo kshop, ASP Con e ence Se ies, Vol. XX, 2001, de. C.M. Ko,as o-
ph/0101120.
[2] D.Bo que, PhD hesis, o be published, Uni e sidad Complu ense, Mad id, (2001).
[3] M.Diaz, PhD hesis, o be published, Uni e si ¨a Heidelbe g, (2002).
[4] D. Dumo a e al. (CELESTE coll.), Che enko
Low Ene gy Sampling & Timing Expe imen - CELESTE expe imen al p oposal,
h p://wwwcenbg.in2p3. /As opa icule (1996).
[5] S.Danahe , D.J.Fegan, N.A.Po e , T.C. Weekes, T.Cole, Possible applica ions o
la ge sola a ays in as onomy and as ophysics, Sola Physics 28 (1982) 335-343.
[6] O.T. T¨ume ,T.J. O´Neill, A.D.Zych, R.S.Whi e, A la ge a ea VHE de ec o wi h
low-ene gy h eshold, P oc. 21s ICRC, Adelaide, 4 (1990) 238-241.
32

Angula dis ance (deg)
Numbe o showe s
Angula dis ance (deg)
Numbe o showe s (on-o )
Figu e 19: Good-wea he da a o he po en ial sou ce 3C 454.3 plo ed as in ig. 18
[7] O.T. T¨ume , A.D. Ke ick, T.J. O´Neill, R.S.Whi e, A.D.Zych, Sola One Gamma
Ray obse a o y o in e media e high ene gies o 10 - 500 GeV, P oc. 22nd ICRC,
Dublin 2 (1991) 634-637.
[8] B.Giebels e al. (CELESTE coll.), P o o ype Tes s o he CELESTE Sola A ay
Gamma-Ray Telescope, as o-ph/9803198, Nucl.Ins .& Me h. A412 (1998) 329.
[9] The STACEE coll., The Sola Towe A mosphe ic Che enko E ec Expe imen
(STACEE) Design Repo , EFI p ep in 97-17 (1997).
[10] M.C. Chan ell e al.(STACEE coll.) P o o ype Tes Resul s o he Sola Towe A mo-
sphe ic Che enko E ec Expe imen (STACEE), as o-ph/9704037, Nucl.Ins .Me h.
A408 (1998) 468.
[11] J.A.Zwee ink e al., The Sola Two Gamma-Ray Obse a o y: As onomy be ween
20 and 300 GeV, P oc. 26 h ICRC 5 (1999) 223.
[12] M. de Nau ois e al. (CELESTE coll.), S a us and cu en sensi i i y o he CELESTE
Expe imen , as o-ph/0010265 (2000); J.Holde , Obse a ion o Mkn 421 wi h he
CELESTE Expe imen , as o-ph/0010264 (2000).
[13] S.Ose e al. (STACEE coll.), High Ene gy Gamma-Ray Obse a ions o he C ab
Nebula and Pulsa wi h he Sola Towe A mosphe ic Che enko E ec Expe imen ,
as o-ph/0006304 (2000), Ap.J. submi ed.
[14] D. Heck, J. Knapp, J.N. Capde ielle, G. Scha z, T. Thouw, CORSIKA: A
Mon e Ca lo Code o simula e Ex ensi e Ai Showe s, Fo schungszen um Ka l-
33
Zeni h (deg)
Excess e en s
Azimu h (deg)
Excess e en s
Figu e 20: The di e ence o he numbe o e en s in ON sou ce di ec ion and OFF sou ce
di ec ion o he C ab da a sample shown in ig. 18 as a unc ion o de ia ion o he zeni h
(uppe plo a.) and azimu h angle (lowe plo b.) om he sou ce di ec ion.
s uhe Repo FZKA 6019 (1998); ind his epo and u he in o ma ion a :
h p://ik1au1. zk.de/heck/co sika/
[15] R.Plaga, J.Fe nandez, J.Gebaue , M.Haege , A.Ka le, On he possible use o he sola
powe plan CESA-1 ield a Tabe nas as Che enko de ec o , 24 h ICRC (Rome)
Vol.1 (1995),1005.
[16] F.A que os e al., The Mini-GRAAL p ojec , P oc. “Towa ds a Majo A mosphe ic
Che enko De ec o IV”,O.C.de Jage (ed.) (1997) 240-246.
[17] B.Wiebel, Chemical composi ion in high ene gy cosmic ays, Repo WUB 94-08
(1994), a ailable on h p:// wpos6.physik.uni-wuppe al.de:8080
/Public/pape s-public.h ml
[18] A.M.Hillas e al.,Spec um o TeV Gamma Rays om he C ab Nebula, As ophys.J.
503 (1998) 744.
[19] J.R.Pa e son, A.M.Hillas, Op imizing he Design o Ve y High Ene gy Gamma-Ray
Telescopes, Nucl.Ins .& Me h. A278 (1989) 553.
[20] V. R. Chi nis, P. N. Bha , Possible Disc imina ion be ween Gamma Rays and
Had ons using Ce enko Pho on Timing, As opa .Phys. 15 (2001) 29-47.
34
Figu e 21: The ene gy lux as a unc ion o ene gy as de e mined he e (diamond) compa ed
o de e mina ions in o he expe imen s (adap ed om [12]).
35