Transcript EAS - indico in2p3

The Tianshan Radio Experiment
for Neutrino Detection
Genesis & status of the TREND project
Autonomous radio-detection of air showers
North
The 21cm array
Radiointerferometer for the study
of the Epoch of Reionization (Wu
XiangPing, NAOC) completed in
2007.
4 km
East
West
South
DAQ
3 km
The TREND site
Simulated
galactic bckgd
Short waves
Urumqi
Ulastai
Measured
bckgd noise
Beijing
TREND
@ Ulastai
• Ulastai, Tianshan mountains, XinJiang autonomous province
(2650m asl)
TREND genesis
(2007-2008)
• Q: is the 21CMA set-up usable for CRs detection?
– Meeting @ Nantes (April 2008)
– Site survey (July 2008)
• A: yes!... And we get free access to it +
technical support from 21CMA.
The TREND « sales strategy »
• Autonomous EAS radio detection & identification
as a key issue in the persepective of a giant radio
array for EAS.
• Radio in R&D phase: need to explore different
technological options.
• TREND as an opportunity:
– low elm bckgrd @ Ulastai
– 21CMA setup to be used for ~free (for France)!
– Large radio-setup instrumental to improve our
understanding of EAS radio info
• Long term plans: neutrino telescope
The TREND contributors
• China: CAS
– NAOC: Wu XiangPing, Thomas Saugrin (2009-2012), Zhao Meng (computing),
Deng JianRong*, Zhang JianLi**, Gu Junhua**
– IHEP: Hu HongBo, Gou QuanBu, Feng Zhaoyang**, Zhang Yi**
• France: CNRS-IN2P3
– LPNHE: OMH, Patrick Nayman***, Jacques David***, David Martin***
– SUBATECH: Pascal Lautridou (2008-2013), Daniel Ardouin (2008-2010), Didier
Charrier (radio antennas)
– LPC: Valentin Niess
*: after 2010
– CC: Fabio Hernandez (computing, 2010-2013)
**: after 2012
***: after 2014
Nearly everybody at a small fraction of time on TREND!!!
EAS autonomous radiodetection at the
Tianshan Radio Experiment for Neutrino detection
(January 2009 - June 2014)
TREND DAQ
optical fiber
DAQ room:
8b 200MS/s ADC
+CPU (soft trigger)
+disk
50-200MHz filter
84dB
21CMA acquisition
50-100MHz filter
64dB
Driving concepts:
- use existing elements
- allow for high trig rate (200Hz/antenna)
optical fiber
TREND acquisition
Total chain gain: G=1000-5000
pod
TREND acquisition chain
Coax cable to
optical
transmitter
(<300m)
fiber to the
DAQ room
(<4km)
Amplification
& filtering
The TREND-50 setup
• 50 monopolar «Butterfly antennas» deployed in summer-automn
2010 over a total surface ~1.5km². Average antenna step = 150m.
• Stable operation between January 2011 & June 2014.
• EW orientation in 2011-2012, then NS.
TREND-15
Ardouin et al.,
Astropart. Phys 34,
2011
<arXiv:1007.4359>
TREND-50
~1.5 km²
TREND DAQ
•
Analog radio signal transfered through
optical fiber to DAQ room.
•
On the fly parrallel digitization at
computer level (200MS/s, 8bits).
•
T0 soft trigger if
antenna amplitude > Nxσnoise (N in 6-10)
• Online search for time coincidences
between antennas: T1 if at least 4 antennas in
causal time frame.
• 1024 samples (≈5 μs) written to
disk for all antennas in T1.
Nxσnoise
snoise
(~10µV @ antenna level)
• For coincident triggers: offline signal
direction reconstruction by triangulation
• Plane wave treatment: direction (Θ, φ)
• Point source treatment: position (x, y, z)
TREND DAQ driving concept: DAQ designed to accept large trigger rate (up to
200Hz/antenna). Candidate selection performed through offline treatment.
TREND antenna sensitivity
Major radio source: thermal emission from
the Galactic plane.
Signal noise level (A. U. )
Galactic plane @ 408MHz
Visible in Ulastai sky between 15h & 23h LST.
TREND antennas clearly exhibit a
~3dB increase in noise level when
the Galactic plane is in the sky.
~3dB
Local sideral time
Experimental evolution in good
agreement with simulated response
(following Lamblin et al, ICRC2007).
TREND trigger performances
• T0 rate <100Hz for 90% of the time on all antennas.
• DAQ efficiency ~ 70%.
• Large trigger rate variations at all time scales on all
antennas: «noise bursts»
• Noise is correlated between antennas: common
(physical) origin.
• Time delay between consecutive events & point
reconstruction points dominantly towards HV sources.
R3577
Antenna 120 (2000m away)
Antenna 112 (1400m away)
Antenna 106 (700m away)
Antenna 101
2011-2012 data
10 ms
R3577
TREND antenna
Reconstructed
source position
2011-2012 data:
317 DAQ days analyzed
3.7 109 triggers recorded
2.4 108 coincidences
~10Hz average coinc
rate over whole array
(~20 EAS/day expected)
RADIO PERFORMANCES:
DIRECTION RECONSTRUCTION
• Plane
track reconstruction :
- 3037 events in 4 minutes
- Θ > 60°
- Max multiplicity: 40
Point source recons
mult ≥ 22 antennas
σ = 0.7°
Total angular resolution <1.5° on the track
(and improves with smaller zenithal angle)
Estimated antenna trigger timing error: ±10ns
Baseline relative calibration: 𝐴𝐶𝑎𝑙 =
𝐴𝐴𝐷𝐶
𝜎𝑏𝑙𝑖𝑛𝑒
OK if senv >> selec :
then sbline ~K senv => K a sbline at time t
Cross-check with plane track signals
Average amplitude <Ai>
RADIO PERFORMANCES:
CALIBRATION
Distant radio source (>4km)
Signal intensity almost identical on all antennas
Antenna 150
s = 0.15
Good check for amplitude calibration
~15% amplitude resolution
(ACal-<ACal>)/<ACal>
TREND issues
• «You get what you pay for»: system reliability questionnable
50-100MHz noise level (dB)
– Sudden drops in gain [not solved]
– Aging (antennas, amplifiers, optical system, computers…)
- Significant maintenance effort required
- Reduced detection efficiency
- Monitoring of efficiency &
absolute calibration (very) challenging [in progress]
TREND-50 EAS search
EAS identification: principle
EAS
(0.2mHz)
Background events
(10Hz)
Discriminating parameters
Selected
EAS
candidates
Residual
background
EAS identification: principle
Simulated
EAS
EAS
(0.2mHz)
Background events
(10Hz)
Discriminating parameters
(optimized with simulated EAS
& bckd events)
Selected
simulated
EAS
Selected
EAS
candidates
Residual
background
EAS simulation
p@ E in [3 1016 – 3 1017] eV with
isotropic sky distrib & random core position
Shower dvlpmt (CONEX)
elm emission (EVA)
Slow!
Antenna response (NEC2)
(if distance<800m)
400 showers/E x 20 core positions x 15 antennas
120’000 voltage computations  240’000h CPU
Using DIRAC+VO France-Asia (IHEP, KEK, CC-IN2P3 & LPNHE)
EAS simulation
• Simulated antenna signal (
digitized @ 200MS/s (o)
)
• Vsimu x G + noise ( ) using
experimental (G, noise)
• Applying TREND trigger
condition with th = 8snoise (
)
• Shower considered detected if
5+ antennas triggered.
• Standard datat treatment &
reconstruction.
Discriminating parameters
• Spherical wave recons: point source
reconstruction of backgrd sources
close to array, EAS more distant.
R>3000m
• Signal shape: prompt signal for EAS
Data:
45% killed
Data:
66% killed
R>3000m
Simulated EAS:
92% pass
Simu:
100% pass
Discriminating parameters
• Array trigger pattern should be continuous for EAS
(E-field linear polarization at 1st order, random for bckgd)
Data
85% killed
Untrigged antennas:
hole in trigger pattern
Simulated EAS
E = 5 1017eV
85% pass
Continuous trig zone
Trigged antenna
Limited array size + monopolar antennas
(+ system unreliability) reduce cut efficiency.
Environment cuts
• Bckgd events strongly correlated in time & space
• Consecutive coincs: reject EAS candidate if 1+ coinc with 4+
antennas in common within 30s.
• Same direction events: reject EAS candidate if 1+ coinc with 2+
antennas in common and |Dj|<10° within 10 minutes.
Cut efficiency:
from 2.4 108 to 465 events
Cut
% survival
Ncoincs final
Simu % survival
« 50Hz » cut
24%
5.9 107
To be determined
Pulse duration
56%
3.3 107
100%
Multiplicity > 4
57%
1.9 107
-
Valid direction
reconstruction
79%
1.5 107
100%
Radius > 3000m
33%
5 106
92%
Q < 80°
14%
7 105
/
Trigger pattern/
Extension
15%
10 5
85%
Neighbourgs
(direction)
3%
2600
To be determined
Neighbourgs
18%
465
To be determined
No cut is related to wave (absolute) arrival direction.
TREND EAS
candidates
Normalized EAS candidates
zenithal distrib
2011-2012 data
(EW polar, 317 DAQ days)
465 candidates
Zenith angle [deg]
90°
60°
30°
Normalized EAS candidates
azimuthal distrib
Excess to
North
Deficit to
East & West
Azimuth angle [deg]
Simulated skymap
• For given direction (q, j): 20 random xcore with min dist to
array < 800m.
• For given shower geometry (q, j, xcore):
– check if antennas signals are above threshold (8xsnoise)
– If OK for 5+ antennas, tag this geometry as ‘trigged’.
• For each direction (q, j), compute ratio Ntriggered/Nsimulated
(Nsimulated = 20 in principle)
Simu voltage x calib + noise
Simulated sky maps
(Zhang Jianli & Gu Junhua)
90°
90°
60°
60°
30°
60°
30°
30°
8 1016 eV
5 1016 eV
90°
1 1017 eV
20/20
shower detection
90°
90°
60°
60°
30°
30°
in progress
0/20
shower detection
2 1017 eV
5 1017 eV
Data-Simu comparison
Azimuthal distribution
Zenithal distribution
Data
Simu
dN/dj (Normalized)
dN/dq (Normalized)
• Combining 8.1016 &
1017eV simulated
data sets.
• Comparable zenithal,
azim and multiplicity
distributions (except
for very inclined
showers: reflexion
issues or cuts?)
• Expected nb of
events for threshold
= 1017eV: ~6000 in
317 days before
analysis cuts. 465
observed… Detection
efficiency <10% ?!
Event multiplicity
Simu
Nb antennas/event
Data
Nb antennas/event
ToDo: full MC simulation
• Simulate EAS events with proper distributions in flux,
direction, core positions & energies.
• Generate expected antenna response to these EAS
events at fixed random times.
• If 5+ triggers, insert these simulated events in
experimental data (after experimental EAS candidates
have been removed).
• Process these data through standard analysis chain.
• Produce simulated maps & compare to data
-> Background rejection performances
-> Detection threshold
-> Detection efficiency
TREND EAS Candidates
2013-2014 data
(NS polar, 125 DAQ days)
14 candidates
- Very low stat BUT distribution differs
significantly from EW polar, as expected for EAS.
Simulated skymap p@1 1017eV
90°
60°
30°
TREND-50 2013-2014
• Possible causes for much fewer candidates:
– Array maintenance degraded (>30% antennas off)
– Bckgd noise significantly higher, affects DAQ duty cycle
& acceptance (environment cuts)
Aug 2012
March 2011
2011: 2012:
EW polar
June 2014
Dec 2012
2013-2014:
NS polar
<Coinc rate> ~100Hz
<Coinc rate> ~15Hz
TREND-50 summary
• Initial goal reached: autonomous radio detection and
identification of EAS with limited bckgd contamination
(<~ 20%) thanks to low DAQ dead time.
• Limitations:
–
–
–
–
Low detection efficiency (set-up layout & stability)
Environment cuts kill detection efficiency when bckgd rises.
Event-by-event discrimination not possible.
Physics output with these data questionnable.
• Larger array with more stable detection chain would
surely perform better…
To Do
• Perform full acceptance study: insert simulated EAS in
real data and run standard analysis
- ‘True’ simulated EAS skymap
- Detector efficiency estimation
• EAS sample analysis (LDF,
spectrum, …)…
• Requires absolute calibration
of the amplitude.
TREND-50 antenna
Absolute calibration tests
Summer 2013
TREND early days (2009-10)
•
2009: 6 log periodic antennas : reconstruction algorithm
development + autonomous trigger proof of principle.
2010: 15 log-periodic antennas + 3 scintillators: independant trigger
& analysis of scint data (EAS) & radio data (EAS radio candidates).
Selection of radio EAS candidates with
dedicated algorithm
Reconstruction of 3-fold
scintillator coincidences  EAS
Radio data
(subset)
Scintillator
data
Some radio EAS candidates are coincident with scintillator
coincidences + direction recons match!
Nants
θradio
θscints
ϕradio
ϕscints
4
61±3
67±5
359±2
3±4
4
52±1
49±3
195±2
191±4
5
42±1
36±3
55±4
56±5
4
45±1
49±3
12±1
10±5
7
56±2
53±4
323±2
331±5
First EAS
identification with
autonomous radio
array
Ardouin et al., Astropart. Phys
34, 2011 <arXiv:1007.4359>
800 m
•
400 m