Comparison of BB and SM at PRSN joint stations

Md5.3 event off Anegada at 11:35UTC on 2 March 2006 produced measurable ground motionswith good signal to noise across the entire PR / VI region. No broadband clipped, though all short period stations (except LSP, with a suspect performance) did. All continuous SM, both 24-bit and 18-bit ETNA's also have measurable motions (unless broken). I am still awaiting any confirmation on which dial-up or stand alone sensors triggered.

These motions provide an opportunity to observe the relative calibration of the sensors from the joint stations, as strong motion and broadband records should be identical.

Unfortunately, of the available joint stations, only AOPR, MPR, CBYP and HUMP had both components functioning. AGPR currently houses a 3 channels datalogger only; CULB went down about 5days prior to the event, and is still not back up; IMO has been down for over 1 year.

For the active stations, data was retrieved from Antelope in SAC format, then:
- all data high pass filtered at 10s
- all data decimated to 20sps
- acceleromater data integrated to velocity
- the ratio of the peak motions in each component was compared.
- plots of each component is presented, along with ratios of peak motions.

If we assume the Antelope metadata for the broadband is correct (which is very liekly as the individual station Ml magnitudes for many events are very similar to that observed at PRSN, the ratio is an observed constraint on the actual gain of the strong motion sensors. I will use this to calibrate the strong motion metadata, and hopefully correct the magnitude errors Antelope currently has for the SM joint stations.

An analysis of the behaviour follows:


Summary

HUMP : good correlation between all components. E components SM is opposite polarity to BB. Difference in gain is about 1600; with a CMG3T, this suggests 10V 2g EpiSensor, as we expect.

AOPR : at PRSN, we clearly have E and Z component switched. Once this is mapped, difference in gain is about 2100; with a CMG ESP, this again suggests 10V 2g EpiSensor, as we expect.

CBYP : horizontal components match very well - appears to be some long period noise in the SM vertical components, but still similar to BB. Diff in gain about 545; with a CMG ESP, this suggests 2.5V 2g EpiSensor, WHICH IS NOT AS EXPECTED, and must be changed. Juan Lugo, Jaffet is this possible?

MPR : SM is clearly malfunctioning.It should be replaced and checked. No analysis possible


Individual Station Performance



HUMP


Figure 1: HUMP : comparison of velocity timeseries from all 3 channels - 50s at start of event. Note E components have opposite polarity pdf version


Figure 2: HUMP : comparison of velocity timeseries from all 3 channels - closeup of 10s data pdf version

Ratio is about 1600
BB sensor is a CMG 3T : 1500V/cm/s
So accel gain is approx : 1.067V/cm/s/s
Expect 1.019V/cm/s/s - error about 5%

AOPR


Figure 3: AOPR : comparison of velocity timeseries from all 3 channels - 50s at start of event pdf version


Figure 4: AOPR : comparison of velocity timeseries from all 3 channels - closeup of 10s data pdf version


Figure 5: AOPR : comparison of velocity timeseries from all 3 channels - with E and Z component of SM swapped pdf version

Ratio is about 2100
BB sensor is a CMG ESP : 2000V/cm/s
So accel gain is approx : 1.05V/cm/s/s
Expect 1.019V/cm/s/s - error about 3%

CBYP


Figure 6: CBYP : comparison of velocity timeseries from all 3 channels - 50s at start of event pdf version


Figure 7: CBYP : comparison of velocity timeseries from all 3 channels - closeup of 10s data pdf version

Ratio is about 545
BB sensor is a CMG ESP : 2000V/cm/s
So accel gain is approx : 0.273V/cm/s/s
Expect 1.019V/cm/s/s

This is over a factor of 4 out! Could the SM sensor not really be 10V/g? Is the BB sensor really a CMG ESP? If the BB is correct, it suggests the SM sensor is acutally a 2.5V/g sensor, as designed for use with ETNA, not a 24-bit logger.

MPR


Figure 8: MPR : comparison of velocity timeseries from all 3 channels - 50s at start of event pdf version


Figure 9: MPR : comparison of velocity timeseries from all 3 channels - closeup of 10s data pdf version

Clearly the SM sensor is not operating correctly.
John Clinton