1. Ciro Bigongiari, Salvatore Mangano Results of the optical properties of sea water with the OB system

2. Outline The idea MC templates Data and MC comparison Conclusions 2

3. The idea Take data with flashing optical beacon –Plot the hit arrival time distributions for all OMs Simulate many MC samples with different input values: λ a and λ s Compare hit arrival time distributions from MC samples and data Choose MC with λ a and λ s which describes best data 3

4. MC samples New CALIBOB Different MC input parameters For example: a = 35, 40, 45, 50, 55, 60, 65, 70, 75 m 9 values s = 35, 40, 45, 50, 55, 60, 65, 70, 75 m 9 values 9*9 = 81 MC samples for each data run Each data run has his: –detector geometry – charge calibration – PMT efficiency – background noise 4

5. Histogram Comparison We compare many histograms one for each OM considered To quantify the agreement between the histograms we calculate the χ 2 5 Data MC

6. Changing Absorption (MC templates) Normalized at first histogram Absorption effects the direct photons (see peak) =>More light at larger distance for larger absorption 70 m 50 m

7. Changing Scattering (MC templates) Normalized at first histogram Scattering effects the indirect photons Photons from peak region go to tail region 50 m 90 m

8. MC and Data comparison Find MC which describes data

9. Chi2 Procedure Loop over selected floors/OMs of one line Cut a fixed range of hit arrival time distribution Merge all the cut histogram ranges in one super-histogram Compare super-histogram from data with MC Repeat for all lines (except OB) χ 2 calculated with Chi2Test function of ROOT –Robust, flexible and well tested

10. Super-Histogram example MC DATA Time Entries

11. Super-Histogram MC shifted time MC DATA Time

12. OM selection Some OMs are rejected –OMs too close to the OB Floor > 13 ARS token ring effect –OMs too far away Floor < 21 Not enough statistics –OMs whose efficiency ε 1.5 –Backwards looking OMs PMT acceptance uncertainty –OMs very inclined Led emission uncertainty –OMs after visual inspection of their distributions 12

13. Binning and statistics The Chi2 values depend: 1.on histogram binning –Very small bins  large statistical errors (Small Chi2 values for all MC models) Chi2 ~ 1 Independent of the MC model  –Very large bin  small statistical errors (Large Chi2 values for many MC models) Sensitive to Attenuation length only 2.on MC statistics

14. MC and Data for Line 2 with small χ 2 Time OM1 OM2 OM3 OM1 MC Data

15. Absorption vs. scattering for Line 2 Calculate Chi2 for each MC Eta=0.3

16. All lines Run 58120 Different Lines show similar results Last Figure shows sum..over all lines

17. Data runs We use data runs taken with the 6 LEDs of the TOP group of one OB flashing at the same time 17 RunEventsOB lineOB floorIntensityDate 5812020516142High17-06-2011 5860720048142High12-07-2011 5860920028122High12-07-2011 6151450058142Low12-12-2011 6151846000042High12-12-2011 6476633592142High11-06-2012 6476946873942Low11-06-2012

18. All lines Run 58607

19. All lines Run 58609

20. All lines Run 61514

21. All lines Run 61518

22. All lines Run 64766

23. All lines Run 64769

31. Final Bologna result Take from lines the MC with smallest chi2 (four runs, eliminate too distant lines)

32. PreFinal Oudja result Take from lines the MC with smallest chi2 (four runs, eliminate too distant lines)

33. Abs per line

34. Scat per line

35. abs per run

36. scat per run

37. Conclusions Improved Data-MC comparison technique Consistent results between different lines and runs Soon new Calibob version with only λ a and λ s Results –λ a = 52 -> 49 m and rms = 6 m –λ s = 59 -> 55 m and rms = 8 m 37

38. Backup 38

39. Changing Eta (MC templates) Large eta more scattering at large angle Photons from peak region go to tail region Scattering and eta are connected => Difficult to disentangle Eta 0.4 Eta 0.15