Log 5

Justin Vandenbroucke

justinv@hep.stanford.edu

Stanford University

11/1/02 – 11/7/02

 

This file contains log entries summarizing the results of various small subprojects of the AUTEC study.  Each entry begins with a date, a title, and the names of any relevant programs (Labview .vi files or Matlab .m files – if an extension is not given, they are assumed to be .m files).

 

11/2/02

Offline adjustment of triggering values

calcFilterMaxes.m

plotFilterMaxes.m

Online, a triggering value is written for each event.  It is the filtered amplitude of the first sample to go above threshold.  It is also scaled by estimated phone gain.   I have a new philosophy regarding phone gains which is to tolerate the uncertainty of a factor of 2.  We can choose the best 10 values from each minute without caring if the values are correct relative to other phones or in absolute terms (Pascals).  We then tolerate a factor of 2 uncertainty in energy threshold, but order of magnitude is sufficient anyway.  The loudest events will be above both the threshold and the threshold/2.  So offline we re-filter time series and determine new triggering values.  The original, online ones are called trval, the new ones are called fmax.  fmax differs from trval in two ways: it uses unrescaled filtered voltages, and it is the maximum over the first 179 samples of the time series (not the first value above threshold).  The plots below summarize distributions of trval and fmax for all full days of data in the December range (see decHourRange.m).  The first plot has ~7 separate baselines presumably because there were ~7 different scale factors that did not correspond exactly to the correct gains.  We see that considering only fmax above 0.05 would cut the event rate significantly.

11/4/02

Slanted detector

plotDetector.m

detectorPlane.m

The detector is not perfectly horizontal but follows the slope of the sea floor.  From the plots below, the slope can be estimated to be ~ 1 % downward as we go to the north.  This makes sense as it is in the direction of the open ocean.  The phones span 28 m vertically.  The vertical heights of the phones relative to the center phone are as follows:

The slant is important for defining our fiducial volume.  Because the radiation disk is so thin (~ 10 m), for multi-phone coincidence, we will not detect a disk that is parallel to the sea surface, only one that is parallel to the actual plane of the detector.  The plane of the detector was calculated (by least squares with the phone positions).  The normal to the plane is pointed 16 east of north, and is inclined at 0.1 from the vertical, corresponding to a sea-floor grade of 0.4 %.  These are all consistent with estimations from the plots.  The phones span 17 m perpendicular to the plane.  Phone heights relative to the plane are as follows:

11/4/02

Data Management

badDay.m

See comments in badDay.m for useful information on data management and avoiding reading bad/corrupt/differently formatted data.

 

11/5/02

Effective area

detectionArea.m

Consider an idealized detector with exactly equal spacing (spacing of 1531 m, the mean length of the 6 spokes of the actual detector) and all phones exactly coplanar.  Imagine a neutrino incident from exactly vertical, creating radiation of radius R in the plane of the detector.  Depending on where the neutrino interacts in the detector plane, it will trigger N phones.  We can color each point in the detector plane depending on N.  The first 5 plots give this for various values of R.  For relationship between R and neutrino energy E0, see the next log entry.  Roughly, E0 ~ R/(0.5 km) * 1020 eV.  So the five plots correspond to 1, 2, 3, 4, 5 * 1020 eV.

11/7/02

Dimensions of detectable radiation pattern

plotDetectionVolumeRect.m

calcDiskVertices.m

For a given neutrino energy, there is a surface of the acoustic radation pattern along which the filter value will be 0.05.  This is our detection surface – we will detect the neutrino at a given phone if it is within this surface relative to the neutrino.  The first two plots give these surfaces for a range of energies (the plots only differ in number of contours given) ~ 1-6 * 1020 eV.  The 3 subsequent plots give the dimensions of the contours.  These are all calculated directly from the contours, which are calculated from Nikolai’s simulation code.

11/7/02

Isotropy considerations

plotAngularFluxDependence.m

According to Gandhi et al, PRD 58, 093009, the total interaction length for ultra-high energy neutrinos on nucleons is Lint = 100 km for 1020 eV and 200 km for 1021 eV.  The first plot below gives, for a detector 1 km below the sea surface, the distance toward the surface as a function of zenith angle, L.  Then we assume the flux of neutrinos reaching the detector goes as exp[-L/Lint].  The resulting flux dependence on angle is given in the second plot.  With a zenith angle as high as 80 degrees, 95% of the neutrinos still reach the detector.  So we can assume a nearly isotropic detector (equal horizontal and vertical spacing) is as good as one with very long horizontal spacing and very short vertical.

11/7/02

3D detector picture

detectorPicture.m

This plot shows the hydrophones and interpolated sea floor.  It indicates the general slope and slight deviations from the plane.  Data on the sea floor roughness on the scale of a few meters (blocking the area directly between phones) would be useful.

11/7/02

Detection volume

calcVolume.m

plotVolume.m

The plots below give detection volume requiring n phones be hit.  The neutrino is assume to be incident from zenith.  The 3 plots have different mesh spacings in x,y,z.  It can be seen that they are all quite similar, so the coarsest mesh is assumed to be sufficient.  The numbering of n is incorrect for the first 3 plots (to correct it, in the legend let n -> n+1) but is corrected in the 4th.  Computation time is about 15 min.  We must next loop over theta and phi to determine acceptance in km-3 sr-1.

Run plotVolume for more pretty flower pictures, this time representing volume for each (E0, n) pair, rather than n for each radius as we plotted for area.

11/7/02

Ultrasonic torpedo?!

plotHourlyIntersections.m

The Navy uses pingers on ships, subs, and torpedoes.  The pingers emit short pulses that are essentially sine waves – short pulses with narrow bandwidth.  The pingers are at several different frequencies.  We have seen what appear to be pinger signals in our data.  They are typically of a single frequency, high amplitude, and long duration.  It would be useful to write a function that used these qualities to determine whether a single-phone event is a ping.

                Pinger signals should be a fruitful subject of study.  They could be used to practice position and energy reconstruction, and should give us some insight into the effect of refraction on signal reconstruction.

                The plots below give the picture of a torpedo being shot from a ship and traveling from sea surface to sea floor.  But if I calculated correctly, it is traveling at 7500 m/s – 5 times the speed of sound!  We should figure out what’s going on here.