R.Shanidze, B. Herold, Th. Seitz ECAP, University of Erlangen
(for the KM3NeT consortium)
15 October 2009 Athens, Greece
Study of data filtering algorithms for the KM3NeT neutrino telescope
Neutrino, atmospheric- and optical background rates in the ANTARES telescope. (Mediterranean Sea, depth -2475 m) • -event rate: ~ 0.1 mHz • atmospheric -rate(*): ~ Hz • Optical background (40K, bioluminescence) ~ 60 kHz/PMT PMTs (Strings x Storeys x PMT) 900 ( 12 x 25 x 3)
Introduction
ANTARES solution(**): - ‘ All data to shore concept’ flexible software filters/triggers: - External trigger (s): GCN/GRB ( SNEWS /Supernova ) - Standard triggers - T2 / T3 triggers - 40K trigger
(*) Integrated atm- rate measured in ANTARES (preliminary): (h)=0 x exp(-b(h-h0)), with 0 =1.72(+0.92-0.57) x10-3 m-2s-1, b=0.0019, ho=2000 m. (*) ANTARS DAQ: ANTARES Collaboration ( J.A. Aguilar et al., ) NIM A570(2007), 107 ( astro-ph/0610029 )
Are these filtering / triggering schemes applicable to KM3NeT telescope ?
R. Shanidze, B. Herold, T. Seitz VLVnT-09, Athens, 15 October, 2009.
Triggering “a la ANTARES”
Hit hierarchy: a) L0. PMT signal above defined threshold (0.3 p.e.) b) L1 - at least 2 x L0 in the same story within tc
- ( 20 ns in ANTARES NT) - PMT signal above threshold ( 5 p.e.)
T2/T3 Triggers (**)• 1T2 ) 2 x L1 in adjacent storeys in t2=100 nsec• 1T3 ) 2 xL1 in adjacent/next to adjacent storeys in t3=100/200 nsec• 2T3 ) 2 x 1T3 clusters in the whole detector in a time window t =2.25 sec
(*) ANTARES-Soft-2005-005(**) ANTARES-Soft-2008-009
“ standard triggers “ (*) based one selection of (n x L1 hits) [x ( “causality”)]
ANTARES storey
The software filters/triggers are based on L1 hits.
R. Shanidze, B. Herold, T. Seitz VLVnT-09, Athens, 15 October, 2009.
Rate in a single PMT large hits in ANTARES ( from M. de Jong ):
a) ANTARES PMT response ( in p.e.) from GRB trigger 61-334877 b) Corresponding rates R0~ 70 kHz, R2/3 ~ 6 kHz / 200 Hz
Contribution to the large (> 1 p.e.)amplitudes: - K40 - atm. PMT noise: ( auto-pulses,…)
L1 Rates in ANTARES
A [ p.e. ] A [ p.e. ]
Rat
e [
kH
z] a b
L1 rates = L1(coincidences) + L1 (large p.e. signals) L1 for coincidences (tc=20 ns) ≈ 600 Hz, for R0 = 70 kHz ( in addition ~20 Hz from single 40K decays )
R. Shanidze, B. Herold, T. Seitz VLVnT-09, Athens, 15 October, 2009.
ANTARES vs. KM3NeT
PMT OM
xPMT.
St.
(DU)
DU St. PMTs
(Tot.)
Dist(m)
St./DU
Vol.
(km3)
ANTARES 10” 3 x 1 25 12 300 900 14.5/70 0.01
KM3NeT 8” 3 x 2 20 127 2540 15240 40/180 2.6
3” 1x31 20 300 6000 186000 30/130 3.1
Storey types in considered for KM3NeT: ANTARES type, extended storey, MultiPMT storey
R. Shanidze, B. Herold, T. Seitz VLVnT-09, Athens, 15 October, 2009.
L1 rates in with multi-PMT OM
Study of 40K rates in multi-PMT OMs: ( talk of B. Herold at this conference )
Optical rates 40K in multi-PMT: studied with GEANT-4 simulations: 31 x 3” PMT ( QE as measured in Erlangen)
R0 in multi-PMT OM ~ 190 kHz ( ~ 6 kHz/PMT )
R. Shanidze, B. Herold, T. Seitz VLVnT-09, Athens, 15 October, 2009.
K40 coincidences Correlated rate ~ 1300 Hz : hits in the same PMT (1 PMT hit): 14 % hits in 2 or more PMTs (L1 hit): 86 % ( ~ 1120 Hz)
Accidental L1 rate (10 ns window) : • R0 = 6 kHz, R(L1)=334 Hz (calculations) • K40 simulations: ~ 300 Hz
Calculations:
with
R. Shanidze, B. Herold, T. Seitz VLVnT-09, Athens, 15 October, 2009.
L1 rates for the flexible tower storey Calculations for L1: - at lest 2 L0 hits from 6 PMTs on a storey in T=50 nsec window. (?) (time window corresponds to 6m storey )
Correlated L1 rate from 40K expected to be much smaller
R(L1) rate per stores as a function of R0.
for R0= 50/100 kHz, R2(L1)=1.9 / 7.4 kH
R. Shanidze, B. Herold, T. Seitz VLVnT-09, Athens, 15 October, 2009.
Accidental L1 rates in ANTARES
Accidental rates in ANTARES ( MC simulations/calculations) ANTARES-Phys-2006-008
Accidental rates for k(L1) hits in the ANTARES telescope for (tE = 2.2 msec, Rmax ≈ 500 kHz) Poisson statistics: Rate(k)= Rmax(mk/k! •exp(-m) ) Maximum rate
Rmax = 1/tE ( ≈ 500 kHz ) with mean number L1: m=R(L1)xNsto x tE
Significantly large number of L1 hits in the KM3NeT time window (~ 10 sec) additional strong suppression is necessary !
KM3NeT: L1 rate: > R(L1) / ANTARES storeys: x 8.5 (20) ANTARES time: ~ x 5 ANTARES Max rate: ANTARE/5 ( ≈ 100 kHz)
R. Shanidze, B. Herold, T. Seitz VLVnT-09, Athens, 15 October, 2009.
Causality relation/filter Causality relation (filter) : t ij < ( dij / cn ) + 20 nsec dij - distance between hits i and j cn - speed of light in water (n=1.35)
distance between OMs in the multi-PMT/string configuration (130m). Mean distance between OM pairs: d ≈ 1200 m corresponds to: t ≈ 5.4s many L1 pairs will pass causality filter.
d is small for the storeys in the same DU. adjacent storeys = T2 next to adjacent = T3
T2/T3 rates in ANTARES
T2(T3) is a pair or L1 with additional constraints: - time interval 100(200) ns in ANTARES - adjacent (next to adjacent) storeys Advantage:
T2 / T3 rate smaller the L1
• Number of T2 pairs ≈ N storeys
Calculations for KM3NeT are coming soon ( will be included in VLVnT-09 proceedings )
R. Shanidze, B. Herold, T. Seitz VLVnT-09, Athens, 15 October, 2009.
• Accidental L1 rate per storey is the main input in the filtering algorithms and should be studied for all KM3NeT configurations.
• Standard causality filter, due to large distances between DUs are much less effective for KM3NeT telescope. • Filtering/triggering algorithms based on T2/T3 clusters are promising approach for the high purity selection of neutruno and muon events.
• Development of new, dedicated software filtering/triggering algorithms seems to be necessary for the KM3NeT telescope.
Summary and Outlook
R. Shanidze, B. Herold, T. Seitz VLVnT-09, Athens, 15 October, 2009.