Seismic Veto at LHO

introduction

We were looking for a way to veto glitches at 70-110 Hz from a seismic stimulus. We settled on using low frequency narrow band asq.

• 839998920 - 841273020 (Aug 19 2006 05:01:46 UTC - Sep 02 2006 22:56:46 UTC)
• H1 science data only, minus last minute of segment
• H2 is ignored in this set (see to do list)
• Sam's filters from September 15th probably change things
• minute trends generated with PSLmon

Figure 1: AS_Q 0.9-1.1 Hz

H1 AS_Q 0.9 - 1.1 Hz, SO pend. mode, LO second harmonic pend.
Threshold for veto at 100
minutes vetoed: 138/17593 (0.75%)

Figure 2: AS_Q 1.1-1.3 Hz

H1 AS_Q 1.1 - 1.3 Hz, BSC stack mode
Threshold for veto at 2000 (I think it should be lower)
minutes vetoed: 225/17593 (1.3%)

Figure 3: AS_Q 5-7 Hz

H1 AS_Q 5 - 7 Hz, BSC stack mode harmonic (truck band)
Threshold for veto at 15,000
minutes vetoed: 224/17593 (1.3%)

Figure 4: AS_Q 70-100 Hz

the target band- we want to remove the outliers from here.

Figure 5: vetoed data

all data points present (green is unchanged)

red: asq 0.9-1.1Hz pink: asq 1.1-1.3Hz blue:asq 5-7Hz green: asq 70-110 Hz 3% of data points removed by combination of veto channels Note far right of histogram: 0.9-1.1 (red) gets 5/87 (6%) glitches greather than 0.0015 1.1-1.3 (pink) also get 5 5-7 (blue) gets 29 of 87 (33%) glitches larger than 0.0015 Total is 39 of 87, or 45%

Figure 6: vetoed data removed

after removing all the vetoed data from figure 5, we now have the green histogram.

Figure 7: efficiency vs max in minute

Next Steps

• remove H2 out of lock data
• better time resolution
• compare various thresholds of veto
• more veto bands
• run for all of S5
• use darm_err instead
• suggestions?