IMPACT OF WATER OPTICAL PROPERTIES ON TRACKS RECONSTRUCTION H Yepes -Ramirez IFIC (CSIC – Universitat de València) ANTARES Collaboration Meeting Moscow,

IMPACT OF WATER OPTICAL PROPERTIES ON IMPACT OF WATER OPTICAL PROPERTIES ON TRACKS RECONSTRUCTIONTRACKS RECONSTRUCTION

IMPACT OF WATER OPTICAL PROPERTIES ON IMPACT OF WATER OPTICAL PROPERTIES ON TRACKS RECONSTRUCTIONTRACKS RECONSTRUCTION

H Yepes -Ramirez H Yepes -Ramirez IFIC (CSIC – Universitat de València)IFIC (CSIC – Universitat de València)

ANTARES Collaboration Meeting

Moscow, June 06th-10th, 2011

OUTLINEOUTLINEOUTLINEOUTLINE

ANTARES Collaboration Meeting ANTARES Collaboration Meeting Moscow, Moscow, JuneJune 06 06thth-10-10thth2

Brief reminder of light propagation Brief reminder of light propagation in sea water: in sea water: ANTARES Monte Carlo

model

Simulation: Simulation: absorption and scattering length inputs, codes and data selection

Selected resultsSelected results

Conclusions and outlookConclusions and outlook

Brief reminder of light propagation Brief reminder of light propagation in sea water: in sea water: ANTARES Monte Carlo

model

Simulation: Simulation: absorption and scattering length inputs, codes and data selection

Selected resultsSelected results

Conclusions and outlookConclusions and outlook

Brief reminder of light Brief reminder of light propagation in sea waterpropagation in sea water

Brief reminder of light Brief reminder of light propagation in sea waterpropagation in sea water


Scattering phase function (Scattering phase function ())

Morel and Loisel approach

Molecular scattering (Rayleigh) Isotropic (<cos>=0)

= contribution of Rayleigh scattering

Particle scattering (Mie) Strong forward peaked (<cos>Mie=0.924)

effscatabs

effatt

111

Attenuation Length (COLIMATED BEAM)

Effective Attenuation Length (ISOTROPIC SOURCE)

Absorption lengthAbsorption length Scattering LengthScattering Length

scatabsatt 111

Mie

scatscateffscat

cos)1(1cos1

924.0cos Mie

Scattering length wavelength dependence

(Kopelevich parameterization)

][550

312.0550

34.1550

0017.0 13.07.13.4

mvvb ls

scat

b1

b = scattering coefficient.

vs, vl = scattering centers.<Cos> = Average cosine of the global distribution

Petzold values for particle scattering

)()1()()( *** MieRay

SimulationSimulationSimulationSimulation


A set of water properties inputs reliable for study:

• AIM production of the absorption and scattering spectrum for different water models for muons and neutrinos-antineutrinos, in agreement to the water models proposed for the data/MC CALIBOB comparison for optical beacon data (J Ruiz-Rivas, Collaboration Meeting in Paris 2010).

• Ten different water models (two runs each, just for a first approach) for muons and neutrinos-antineutrinos (20 files for muons, 20 files for neutrinos and 20 files for antineutrinos).



abs [m] Vs, Vl (scattering centers) [ppm] scat at 470 nm [m]

55 0.0075 53 0.17

55 0.01 41 0.17

55 0.02 22 0.17

55 0.01 41 0.11

55 0.02 22 0.02

63 0.0075 53 0.17

63 0.01 41 0.17

63 0.02 22 0.17

63 0.01 41 0.11

63 0.02 22 0.02

• Three runs with the same value and different scattering spectrum for a given absorption length.

• Three runs with different values, but is computed in such a way that the three runs will have the same effective scattering length at 470 nm, for a given absorption length.

• Three common models for comparisons (muons abs55-abs63, neutrinos + antineutrinos abs55) with previous Monte Carlo productions and data agreement Choosed the “most common” one (≈ data sampled in MC, data size to compare, muons + neutrinos production, etc.).

• OM Angular acceptance of June 2009 (Genova Meeting 2009).



GENGENWATER MODEL:

•Photon tables production (water tables) Water tables (hbook files) + Description files (ASCII files).

HITHITOM PARAMETERS:

• Hit probability computation from the water tables for a given OM parameters Hit tables (hbook files) + Description files (ASCII files).

KM3KM3SIMULATED EVENTS: GEOMETRY + KINEMATICS

• Physics events reading and OM hits production based on event geometry and hit probability tables Detector events: Signal hits (muons, not tracks from hadronic showers), physical background.

GEASIMGEASIM

MCEWMCEW

TETE

RECORECO

SIMULATIONS OF ATMOSPHERIC NEUTRINO INTERACTIONS.

• Process (and evaluation) tracks from particles coming from the hadronic showers (also muons from KM3).

TRANSLATION OF INFO ASCII FILES INTO ROOT FORMAT.

FORMAT CONVERSION TO “LOOK LIKE DATA”: electronics smearing effects (calibration, ARS response) and optical background.

RECONSTRUCTION: Reconstruction of track direction (AAfit) and ntuples information arrangement as number of hits, zenith distribution…(AntDST).

Simulation chain:



Main options and software versions in muons and neutrinos simulation:

CODE/INPUT OPTIONS/VERSIONS

GEN v3r7

HIT v3r7

KM3 v3r7

DETECTOR r12_c00_s01

GEASIM v4r10

MCEW -

TriggerEfficiency Gaussian ARS threshold file: threshold_gaus_0.33_0.08_0.1.txt

SoS file: noise_basic_harold.root (Thanks to C Bogazzi)

-n 10000000 –t 104.858 –C3 – p 0.035

-t 104.858 Frame time in ms. To determine the number of background hits to be generated in case the summary data are used.

-C3 Hit generator type: 3, Gaussian, according observed charge distribution, with time-dependent contribution of after pulses.

2010-09-23 version

Aafit v0r6 (MC, not affected by the alignment BUG), v0r8 (data)

AntDST v1r1



Methodology: runs selection, lifetime computations and weights

1. Data subsample: “Point source search with 2007 and 2008 data” (ANTARES-PHYS-2010-008).

3. Lifetime and weights for MC:

• Lifetime for muons: 2*8320/86400 = 0.19 days.

• Weights for neutrinos-antineutrinos: w3*(1.0/2*1.0e+10).

• 1044 files (12 lines detector) [08/05/2008-30/12/2008].

• Lifetime data: 76.77 days.

• ntuples from JP Gomez-Gonzalez (thanks!!!).

2. MC sampling:

• SoS file prepared by C Bogazzi (thanks!!!) from the data subsample (noise_basic_harold.root): time slices taken from all different acquisition conditions.

• Mupage for muons.

• Geasim for neutrinos-antineutrinos.

Thanks to Annarita and Carla also for the codes used as starting point and their support !!!

Thanks also to Patrick, Maarten and Aart from NIKHEF for their help about software details !!!

“Self-computed” water tables + Same SoS file 1 mupage file for abs = 63 m @ 470 nm

Remarks:

1.Water tables produced by Annarita to Juan Pablo and the mine ones, are equivalent .

2.Reproducibility of the simulation chain and equivalence among scripts are OK.

3.Run-dependent simulations (ANTARES-PHYS-2010-008) Agreement MC / data should be independent of the SoS file MC Sampling with a SoS file based on 2008 data subsample. Data/MC comparisons with the same data subsample.





Juan Pablo (Amsterdam) (abs = 55 m)My production (abs = 55 m)

Ratio Data/MC

Data > MC; R > 1 Data < MC; R < 1

tcosth > 0 && beta < 1

SoS MC sample Data sample KM3 AAFit OM acceptance

Harold noise_basic_harold.root 2 ()+2()+2() 1044 (~½ 2008) v3r7 v0r6 2009

Juan Pablo ? 391 (total) 2663 (2007-2008) v3r6 v0r6 2008

• Nice agreement in general, main differences mainly could come from statistical reasons (for ), data sample to compare and MC sampling, and probably KM3 version.



What kind of information do we have?

1. Reconstructed data:

AAFit v0r8 to avoid bug alignment in V0R6: alignment “isvalid” issue (ELOG 521/527).

2. Reconstructed MC (muons + neutrinos-antineutrinos):

AAFit v0r6 Not affected by the below bug founded (obviously).

3. Reconstructed data + MC:

• Absorption + scattering info:

• Reconstruction quality parameter.

• Number of hits used in the fit.

• Total amplitude of the hits used in the fit.

• Zenith distributions.

• Etc…(backup).

• Scattering info Time information from time residuals computed from muon tracks.

• Lack of muon events for some cases STATISTICS DEPENDENT (few runs to compare) Huge ratios (-6<<-5.5).

• Well reconstructed in general, except some extremes models which could be discarded (blue and blue sky curves: sca~22 m).

• Muons tracks quality doesn’t seem enough affected by the scattering length within ~ 58 % of confidence for ≥ 0.11.

• Neutrinos tracks quality is more sensitive to the scattering length, around 23 % of confidence, for reasonable models.

Selected resultsSelected resultsSelected resultsSelected results


Reconstruction quality parameter ():

• Lack of muon events for some cases STATISTICS DEPENDENT (few runs to compare) Huge ratios (-6<<-5.5).

• Some extremes models could be discarded (grey and red curves: sca~22 m).

• Muons tracks quality are not enough affected by the scattering and absorption length.

• Neutrinos tracks quality seems to be more sensitive to the scattering and to the absorption length spectrum.





Reconstruction quality parameter summary:

1.The quality of a reconstructed track will not be affected unless the sca trends to increase to extreme values (~22 m).

2.Muons tracks are less sensitive to the abs and sca, it is not seen for neutrinos.

3.The contribution of Rayleigh scattering could be between 0.11 < < 0.17.

4.The water models: 63/41/0.11, 55/53/0.17 and 55/41/0.11 fit better to data sca,eff ≈ 227 m.

• The most extreme model could be discarded: blue curve 63/22/0.17.

• Two models agree to peak where we expect direct photons: red line (63/53/0.17) and pink line (63/41/0.11) N < 80.

• Remaining two models agree to tail where we expect scattered photons: green line (63/41/0.17) and blue sky line (63/22/0.02).



Number of hits used in the fit (Nhit):



• The most extreme model could be discarded: grey curve 55/22/0.17.

• The most flattest distribution is seen for the green curve model (55/53/0.17), however it doesn’t match to peak level, less events are expect compared to data.

• Remaining models have a significant differences and the disagreement for peaks and tails is seen.



Number of hits summary:

1.An increased abs fit better to peaks, but a reduced abs fit better to tails.

2.The best models which could be fit to data: 63/41/0.11, 55/53/0.17 and 55/41/0.11.

3.Extreme scattering models (sca=22 m) experience the worst agreement to data, except for the very delayed photons (N > 80).

• The extreme blue and blue sky curve model could be discarded.

• Most of neutrino events are lost close to the horizon (cos≈ 0; 90º), and most of them arrive perpendicular to the detector (cos≈ 1; 0º).

• For down-going muons there is a reasonable agreement for the selected model 63/41/0.11. Not enough up-going muons probably due to the reduced MC data sample to compare, this effect is seen some values before cos < 0.



Zenith angle of the fitted track (cos):



• The extreme grey and red curves models could be discarded.

• The reduced MC data sample in the neutrino region restricts the comparison.

• A similar shape is seen for ratios for down-going muons for the founded “best models”, but once again, a mass production can be proper to explain the lack of statistics for cos > -0.1 for up-going muons and up-going neutrinos.



Zenith angle summary:

1.An abs effect is expected for tracks ( and ), just close to the horizon.

2.The bests models which could be fit to data are: 55/53/0.17, 55/41/0.11 and 63/41/0.11.

3.A mass MC production is so relevant in order to support the current analysis.



Azimuth angle of the fitted track:

• The extreme blue curve model could be discarded.

• Muons and neutrino events coming from a large range of azimuth angle, seem not to be enough affected by scattering and absorption.

• The best model which fits better to data is 63/41/0.11.




• Muons and neutrino events coming from a large range of azimuth angle, seem not to be enough affected by scattering and absorption.

• The best models which fit better to data are 55/53/0.17 and 55/41/0.11.



Azimuth angle summary:

1.The muons and neutrino tracks coming in a large range of azimuth angle, don’t seem to be enough affected due the scattering and absorption.

2.The best models which could be fit to data are: 55/53/0.17, 55/41/0.11 and 63/41/0.11.



Time residuals for the detector lines (thit - texp): tracks

Arrival time of photons is expressed relative to the expected (theoretical) arrival time (texp) which can be computed from the muon tracks parameters: resulting time residuals (r). The true arrival time (thit) could be change due to photons emitted from secondary electrons being their path influenced by the scattering.

• Data binary file (JP Gomez-Gonzalez) from:

/sps/km3net/users/jpgomez/ANTARES/Physics/v2r7/Linux-x86_64/TimeResidualsv2r5_data.exe

It requires AAFit CalReal format files: most recent ones (affected due the alignment bug):

/hpss/in2p3.fr/group/antares/user/schussl/data/AAFit/v0r6/AartsLineOffsets

(No available information on v0r8 in CalReal format yet)

• Monte Carlo binary file from:

/sps/km3net/users/jpgomez/ANTARES/Physics/v2r7/Linux-x86_64/TimeResidualsv2r5_save2.exe

My own reconstructed files with v0r6 are shielded against alignment bug.



• The extreme blue curve model could be discarded.

• Distribution of the time residuals for data is more shifted to smaller values than the MC ones Could be an effect due to the “alignment bug” (thit < texp, increased hit amplitudes) - software details? To check.

• The models 63/53/0.17 - 63/41/0.11, seem to fit better to data, specially at the tails level. However , less entries are seen for data than for MC distributions, specially at peak level.




• Distribution of the time residuals for data is more shifted to smaller values than the MC ones Could be an effect due to the “alignment bug” (thit < texp, increased hit amplitudes) or software details? To check.

• The models 55/53/0.17 - 55/41/0.11, seem to fit better to data, specially at the tails level.



Time residuals summary:

1.An

thit-texp [ns]

thit-texp [ns]

Time residuals summary:

1.A general shift ~ 2 ns is seen for MC concerning to data reconstructed with v0r6. Is it a possible effect due to the “alignment bug” or due to software details or calibration constants?

2.A reduced abs (i.e, 55/53/0.17 – 55/41/0.11) seems to fit better to data at a peak level, instead a better agreement is “slightly” seen for tails for a increased abs(i.e, 63/41/0.11).

CONCLUSIONS AND OUTLOOKCONCLUSIONS AND OUTLOOKCONCLUSIONS AND OUTLOOKCONCLUSIONS AND OUTLOOK


1. A systematic data/MC comparison for optical properties studies has been presented for a reduced MC data sample.

2. The quality of a reconstructed track could not be affected unless the sca trends to increase to extreme values (~22 m). Contribution of Rayleigh scattering could be between 0.11 < < 0.17 and sca,eff ≈ 227 m.

3. An abs effect is expected for tracks ( and ), just close to the horizon looking at zenith angle. Muons and neutrino tracks coming from a large range of azimuth angle don’t seem to be enough affected due the scattering and absorption.

4. A general shift ~ 2 ns is seen for MC concerning to data reconstructed with v0r6 in time residuals distributions, a possible effect due to the “alignment bug”, software details or calibration constants? A reduced abs seems to fit better to data at a peak level, instead a slightly better agreement is seen for tails for a increased abs try time residuals on AAFit v0r8.

5. How much the correct model could matters? The best models which fit to data are founded for 63/41/0.11, 55/53/0.17 and 55/41/01.11. but it has to be confirmed for a mass production (done up to KM3 level). Time residuals points to a reduced absorption length model.

6. A tentative “to do list”: reconstructed and no reconstructed data

7. Number of hits in the trigger, total number of hits in the event, arrival time of the hit on the PMT, before and after trigger, detector performance: expect impact on effective areas, detector angular resolution and sensitivity studies, test on BBFit? Review to the reconstruction PDF for influence of scattering and else?

MORE PLOTS IN THE BACKUP OR: http://ific.uv.es/~yepes/CM_MOSCOW_2011/talk/finalPlots

1. A systematic data/MC comparison for optical properties studies has been presented for a reduced MC data sample.

2. The quality of a reconstructed track could not be affected unless the sca trends to increase to extreme values (~22 m). Contribution of Rayleigh scattering could be between 0.11 < < 0.17 and sca,eff ≈ 227 m.

3. An abs effect is expected for tracks ( and ), just close to the horizon looking at zenith angle. Muons and neutrino tracks coming from a large range of azimuth angle don’t seem to be enough affected due the scattering and absorption.

4. A general shift ~ 2 ns is seen for MC concerning to data reconstructed with v0r6 in time residuals distributions, a possible effect due to the “alignment bug”, software details or calibration constants? A reduced abs seems to fit better to data at a peak level, instead a slightly better agreement is seen for tails for a increased abs try time residuals on AAFit v0r8.

5. How much the correct model could matters? The best models which fit to data are founded for 63/41/0.11, 55/53/0.17 and 55/41/01.11. but it has to be confirmed for a mass production (done up to KM3 level). Time residuals points to a reduced absorption length model.

6. A tentative “to do list”: reconstructed and no reconstructed data

7. Number of hits in the trigger, total number of hits in the event, arrival time of the hit on the PMT, before and after trigger, detector performance: expect impact on effective areas, detector angular resolution and sensitivity studies, test on BBFit? Review to the reconstruction PDF for influence of scattering and else?

MORE PLOTS IN THE BACKUP OR: http://ific.uv.es/~yepes/CM_MOSCOW_2011/talk/finalPlots

http://ific.uv.es/~yepes/CM_MOSCOW_2011/finalPlots

http://ific.uv.es/~yepes/CM_MOSCOW_2011/finalPlots

BACKUPBACKUPBACKUPBACKUP


5 parameters fit - 2 minimization: t0, , , x0, y0

cgctheory

k

v

kl

ctt

sin

1

tan

10(exp)

Reconstruction quality factor L:

• Linear prefit photon hit coordinates x, y, z, t

• Minimization with hit-charge weights.

• Maximum likelihood (L) fit computed from MC PDF of time residuals.

)1(log

solutionsdof

NN

L



Statistics effect on zenith distributions:

abs = 63 m @ 470 nm (official mupage production @ CERN)



Angular error ():







Number of lines used in the fit (Nlines):







Time residuals:



Documents

IMPACT OF WATER OPTICAL PROPERTIES ON TRACKS RECONSTRUCTION H Yepes -Ramirez IFIC (CSIC – Universitat de València) ANTARES Collaboration Meeting Moscow,