The current version of dewarMon was run over the s4 data (2/21/05-3/29/05) and part of the s5 data (12/12/06-2/23/07) at LHO to compare dewar glitch rates from the two science runs, to gain a better understanding of the dewar glitches, and to find the best way to reduce the number of false alarms. DewarMon is based on the glitchMon glitch finding algorithm developed by Masahiro. It uses filtered timeseries from seismic channels at each station and identifies dewar glitches by searching for coincident glitches from two different channels at a given station. This method still results in a high percentage of false alarms however. To help distinguish actual dewar glitches from other events, accelerometers were placed on the two Mid-X dewars on Dec 12 2006.
The table below compares the results from the two runs. During this portion of s5 the accelerometer data made it easy to identify false alarms at the MX station. However, for all other stations (and MX during s4) false alarms were thrown out by hand. Another note: Many of the LVEAX and LVEAY glitches are the same because I don't have a good way of determining which dewar they originated from. A future version of dewarMon will probably combine them.
The next few plots concern the MX dewar glitches only. There is a clear correlation between the frequency of the dewar glitch and the level to which the dewar is filled. During s4, the frequencies increase as the dewar levels decrease. During s5, the frequencies increase until the CP5 dewar is filled, at which point they drop suddenly and start increasing again. Based on these plots and the accelerometer plots in the table above, it seems that most or all of the dewar glitches during this portion of s5 are coming from the uninsulated CP5 dewar.
 |
 |
 |
 |
Because neither MX dewar was filled during s4 and there are no dewar accelerometer channels available, it isn't clear which dewar the s4 MX glitches were coming from. If I assume that both dewars have the same frequency vs. dewar level behavior, then overlay the plots from s4 and s5, then it appears as if some or most of the s4 glitches came from the CP6 dewar. The CP6 dewar was insulated between s4 and s5, which would explain why the number of glitches at MX is so drastically reduced.
 |
 |
One hypothesis for the dewar glitches is that they tend to occur during extreme outdoor temperatures or temperature gradients. To test this, I plotted average temperatures vs. time of day along with the s5 MX dewar glitch times and temperatures. The second plot is of average temperature gradients vs. time of day. No correlation is seen between glitches and temperature gradients, but the glitches do seem to occur during colder times (overnight). The histograms below also show only the tendency for glitches to occur when it is colder.
 |
 |
 |
 |
Still to come: