AN UPDATE ON ABSORPTION LENGTH MEASUREMENT WITH THE OB SYSTEM ANTARES Collaboration Meeting Paris (France), September 20th-24th H Yepes, J Zuñiga IFIC (CSIC – Universitat de València)

A BRIEF REMINDER OF THE EXPERIMENTAL PROCEDURE DATA TAKING STATUS DATA ANALYSIS STATUS: Multi-wavelength analysis and OB systematic effects studies OUTLINEOUTLINE ANTARES Collaboration Meeting Paris, September 20th-24th 2

THE EXPERIMENTAL PROCEDURE ANTARES Collaboration Meeting Paris, September 20th-24th 3 Experimental method: 1. One single top LED of the lowest OB in the line flashes upwards. 2. Signal hits are plotted and fitted (between R min, R max ) by means of an exponential function. F2 3. Quality cuts applied: To avoid the electronics dead time (related to R min ): region where the probability to get more than one photoelectron is negligible (i.e < 1 %). To avoid noise fluctuations at large distances (related to R max ): region where the signal will be greater than the noise. Low efficiency OMs cleaning: from the noise hits projections, only those between (  +3 ,  - 3  ) are considered. Low and flat level noise along the line is required (<100 kHz). 1.A BRIEF REMINDER OF THE EXPERIMENTAL PROCEDURE: Remarks: 1. The efficiencies for the OMs are computed from the normalization of the signal hits to their own noise hits. 2. The total error assigned is computed as the quadratic sum of the statistical and dispersion errors. ANTPLOT-CALI

DATA TAKING STATUS ANTARES Collaboration Meeting Paris, September 20th-24th 4 The experience from the analysis has let the optimization of data taking: Golden runs taken by request, once conditions are met: LOW AND FLAT LEVEL SHAPE along the line required. Different lines/OBs/LEDs/LEDs intensities (L4F2, L4F9, L8F2, L8F9, L2F2 all faces) to study MAINLY systematic effects and influence of depth on absorption length (L2F9, L8F9). Runs at different wavelengths have been also performed. Updated until 16/08/2010 Number of Golden runs (maximum LED intensity) Collaboration Meeting Clermont-Ferrand 42 Collaboration Meeting Paris +30 TOTAL72 GOLDEN RUN

DATA TAKING STATUS ANTARES Collaboration Meeting Paris, September 20th-24th 5 NEW MEASUREMENTS PERFORMED AT = 532 nm by means of the laser beacon: Reference fit criterion (R min ): Take distances where the probability to get more than one phe is negligible: x = number of signal reaching the OM  = number of signal hits / number of flashes Laser beacon runs selection: Standard laser beacon runs at maximum polarizer voltage value. Low and flat level shape along the line, N flashes >= 100k. High intensity at 532 nm (green) LED intensityR min [m]R max [m]P(phe>1) H (blue, 470 nm) % H (UV, 400 nm) % H (green, 532 nm) % [nm] Golden runs 470 (blue) (UV) (green)30 TOTAL101

DATA ANALYSIS STATUS I ANTARES Collaboration Meeting Paris, September 20th-24th 6 Then the variable follows Student’s t distribution: with ν = n–1 degrees of freedom Therefore the error is taken to be: σ = t ∙ s/√n In order to have 68.27% errors, the one-side tail of the cumulative Student function must be 84.13% and thus t = 1.32 (for n=3) or t = 1.83 (for n=2). If only one OM in the storey do not use that storey in the fit. Each storey provides 3 intensity measurements (3 OMs). Assume these measurements are independent and gaussian distributed (check this hypothesis later with results). Assign one signal intensity per storey computed as the average of the 3 OMs – OM signal extraction as before: subtract background, correct by efficiency, discard bad OMs Compute error by means of Student’s t (this is the usual treatment for the estimation of the mean and standard deviation of a gaussian distribution when none of the two parameters are known): TREATMENT OF ERRORS:

DATA ANALYSIS STATUS I ANTARES Collaboration Meeting Paris, September 20th-24th 7 CHANGING TO 40 K EFFICIENCIES: Noise based efficiencies can give correlated errors if the background is not constant in time. 40 K is not affected by variations of the bioluminescence background in time. 40 K concentration (salinity) stable along the full detector deep range within 0.1 %. The volume density of decays slightly increases with increasing depth du to overpressure (<0.5 %). LIGHT OUTPUT OF 40K PER UNIT VOLUME IS CONSTANT OVER DEPTH  Independent of water transparency. γ 40 K 40 Ca e - (  decay) The relative efficiencies or “sensitivities” of a triplet s i of OMs, can be defined as Rate ij = R 0 * s i * s j for i, j = 1,2,3 and R 0 = 16.2 ±1.0 Hz (equivalent to 0.3 p.e threshold) (absolute normalization). OMs delivering such rate are declared “nominal sensitivity” OMs. Solving the system of these 3 equations, can be produced OM sensitivity tables based on 40 K runs. The method requires all 3 OMs of the storey working: If 1 OM is dead  Assume that the other 2 OMs have the same efficiency  1 equation, 1 unknown. If 2 OMs are dead  All 3 OMs efficiencies set to 0.

DATA ANALYSIS STATUS I ANTARES Collaboration Meeting Paris, September 20th-24th 8 Bad fits? Increasing rates in time? Time distributions? Blue UV Green Blue UV Green TRANSMISSION LENGTH RESULTS: One UV run. Six L2 runs-batch. Some under-over flows not showed.

DATA ANALYSIS STATUS I ANTARES Collaboration Meeting Paris, September 20th-24th 9 ANOMALOUS CASES: UV Guilty !!! Over-underflows have been affected by that very small error assignment  Are there not enough statistical fluctuation by the 3 independent OMs measurements? Starting point to perform the special “multi-faces” runs. There is not a similar effect on the six L2 runs-batch There is not a hint of background problems, along the line. Run performed at low and flat background conditions. Blue 50370

DATA ANALYSIS STATUS I ANTARES Collaboration Meeting Paris, September 20th-24th 10 Neither in time…corrected by 40 K efficiencies Time and charge distributions look as the expected golden runs taken before. UNKNOWN PROBLEM !!!  Analysis ongoing Some low counting OMs …

DATA ANALYSIS STATUS I ANTARES Collaboration Meeting Paris, September 20th-24th 11 If the unexplainable runs-batch are removed: Blue UV Green Blue UV Green

DATA ANALYSIS STATUS I ANTARES Collaboration Meeting Paris, September 20th-24th 12 Blue UV Green Blue UV Green

DATA ANALYSIS STATUS I ANTARES Collaboration Meeting Paris, September 20th-24th 13 The mean value of the distribution of the L errors from the fits shows an agreement with the RMS of the transmission length distribution: UV  1.0 m Vs 1.5 m Green  1.1 m Vs 1.0 m Blue  3.0 m Vs 2.9 m. The time stability and the RMS distribution confirms the showed results in the latest Collaboration Meetings, if the runs-batch are avoided (a deepest analysis is being performed). Blue UVGreen

DATA ANALYSIS STATUS I ANTARES Collaboration Meeting Paris, September 20th-24th 14 Pulls distributions  Evidence of BIAS and verification of error coverage Blue UVGreen Blue UV Green BIAS studies ongoing 

DATA ANALYSIS STATUS I ANTARES Collaboration Meeting Paris, September 20th-24th 15 [nm] Entrie s L ± RMS[m] Average σ fit (RMS) [m] Mean Prob (  2 )RMS Prob (  2 )Entries with Prob (  2 ) < 1% 470 (Blue) ± (1.0) (UV)2036.5± (1.0) (Green) ± (0.4) SUMMARY: BLUE: Reasonable fit probabilities Variability of L ~5% (RMS/L): RMS of L in agreement with average σ fit : 3.0 m vs. 2.9 m Change of L with time not much larger than statistical Somewhat high probabilities: a few entries close to 1 and: Mean Prob (  2 ) = 0.68 RMS Prob (  2 ) = 0.31 UV: Good fit probabilities Variability of L around 3% (RMS/L): RMS of L distribution in agreement with average σ fit :1.0 m vs. 1.5 m Mean Prob (  2 ) = 0.74 RMS Prob (  2 ) = 0.29 Green: Good fit probabilities Variability of L around 6% (RMS/L): RMS of L distribution in agreement with average σ fit :1.2 m vs. 1.0 m Mean Prob (  2 ) = 0.53 RMS Prob (  2 ) = 0.30 Mean Prob (  2 )  should be 0.5 RMS Prob (  2 )  should be 1/√12 = 0.29 STABILITY IN TIME IS CONFIRMED FOR DIFFERENT WAVELENGTHS !!! * L error is not divided by √n since it is not an statistical error, it is a systematic one.

DATA ANALYSIS STATUS: SYSTEMATICS ANTARES Collaboration Meeting Paris, September 20th-24th 16 OPTICAL BEACON FACES: LED SYSTEMATICS There are 3 independent light intensity measurements by storey. There are 6 LEDs placed over the 6 LED Beacon faces. Optical Beacon choosed for analysis  L2F2. Take a look on the LED influence for different periods in time  6 runs are performed by day for different periods in time, equivalent to the 6 LED Beacon faces. IDEA  Study the dependence LED flashing - OM light collected, per storey in time. Run batch, just for this study !!!

DATA ANALYSIS STATUS: SYSTEMATICS ANTARES Collaboration Meeting Paris, September 20th-24th 17 Amount of light collected by the OMs at different periods of time using all LOB faces: high light intensity region ( F3, not used in fit): For high light intensity region in the line, a dependence to the LED seems not to be found. The amount of light percentage collected by one particular OM is higher /lower than the other ones:  OM dependent.  Angle between photon – OM ? Angular acceptance not used.

DATA ANALYSIS STATUS: SYSTEMATICS ANTARES Collaboration Meeting Paris, September 20th-24th 18 Amount of light collected by the OMs at different periods of time using all LOB faces: medium light intensity region ( F12, used in fit): For medium light intensity region (the one used in the fit), the systematics are not evident. At high distances, a correction by angular acceptance could carry out.  Next step: correction by alignment.  The obtained value for the transmission length doesn’t has large changes, without to take into account the angular aceptance, then, should we to perform such analysis? A SECOND ORDER CORRECTION.

CONCLUSIONS AND MILESTONES ANTARES Collaboration Meeting Paris, September 20th-24th 19

BACKUPBACKUP 20 NOISE SUBTRACTION: RATE OF CORRELATED COINCIDENCES: Defined as the integral under the coincidence peak (excluding pedestal) normalized to the effective duration of observation period, and properly corrected for dead time of the electronics and data acquisition. Gaussian fit to compute the rate. Average value ~ 14 Hz (R 0 ). R 0 may include the loss of glass transparency due to biofouling (if any) and similar effects, so it may be less than for "ideal" Monte Carlo OM. OM angular acceptance can be constrained by the 40 K measurements. NOISE LEVEL Fit a constant in the [-1000, -50] ns range (B level ) and substract the noise contribution (Q noise, N noise ): N signal = N hits(tot) – N noise = N tot – B level (T min - T max ) = N tot – N bins (T min - T max )