Station Maintenance

Ongoing network maintenance involves regular inspection of the collected seismic waveform data and spectra for nearby seismic events, and also from noise samples. Other common problems include changes to background noise levels due to ground loops and failing preamps, as well as power and telemetry issues. Troubleshooting and remediation of problems often require a coordinated effort, with a technician at the BSL to examine seismic waveforms and spectra while the field technicians are still on site. BSL technicians and researchers regularly review data and assist in troubleshooting.

The NHFN station hardware has proven to be relatively reliable. Nonetheless, numerous maintenance and performance enhancement measures are still carried out. In particular, when a new station is added to the backbone, extensive testing and correction for sources of instrumental noise (e.g., grounding related issues) and telemetry through-put are carried out to optimize the sensitivity of the station. Examples of maintenance and enhancement measures that are typically performed include: 1) tests on radio links to ascertain reasons for unusually large numbers of dropped packets, 2) trouble shooting sporadic problems with numerous frame relay telemetry dropouts, 3) manual power recycle and testing of hung Quanterra data loggers, 4) replacement of blown fuses or other problems relating to dead channels identified through remote monitoring at the BSL, 5) repair of frame relay and power supply problems when they arise, and 6) correctingtenna problems that arise due to various causes,such as weather or cultural activity.

Quality Control

• PSD and Real Event Displays: Our commonly used quality checks on the performance of the borehole installed network include assessments of the power spectral density (PSD) distributions for background noise and quick checks following large teleseismic and moderate local earthquakes of the data records across the NHFN stations. Shown in Figure 3.11 is such a display for NHFN geophone channels from a recent Bay Area earthquake (20 July 2007, M4.2 Piedmont, CA). It is immediately apparent from this simple display that station MHDL was dead and needed immediate attention.

  It is also apparent from the buzz underlying the earthquake signal in this display that the grounding schemes for stations SBRN and HERB may be in need of modification. One of the most pervasive problems at NHFN stations is power line noise (60 Hz and its harmonics at 120, 180, and 240 Hz). This noise reduces the sensitivity of the MHH detectors, and at any given station this type of noise source changes over periods of weeks to months, requiring continued vigilance and adaptability of the grounding scheme in order to maintain the desired high sensitivity to low amplitude seismic signals.

• Geophone Calibration Tests: Comparisons of the inferred ground accelerations generated by local earthquakes from co-sited NHFN geophone and accelerometer pairs show that the waveforms generally are quite coherent in frequency and phase response, but that their inferred ground accelerations differ significantly. At times, the amplitudes differ by up to a factor of $$2 while the times of the peak amplitudes are identical. This implies that the free period and damping of the geophones are well characterized. However, it also indicates that the generator constant is not accurate (assuming that the corresponding ground accelerations inferred from the accelerometers are accurate).

  Generally speaking, the accelerometers, being an active device, are more accurate and also more stable than the geophones, so it is reasonable to assume that the most likely reason for the difference is that the assumed generator constants for the geophones are inaccurate. Rodgers et al. (1995) describe a way to absolutely calibrate the geophones in situ and to determine their generator constant, free period and fraction of critical damping. The only external parameter that is required is the value of the geophone's inertial mass.

  We have built a calibration test box which allows us to routinely perform the testing described by Rodgers et al. whenever site visits are made. The box drives the signal coil with a known current step and rapidly switches the signal coil between the current source and the datalogger input. From this information, expected and actual sensor response characteristics can be compared and corrections applied. Also, changes in the sensor response over time can be evaluated so that adjustments can be made, and pathologies arising in the sensors due to age can be identified. Once a geophone is absolutely calibrated, we also check the response of the corresponding accelerometer.

Figure 3.11: Plot of unfiltered P-wave seismograms, recorded on the geophones of the 14 operational NHFN borehole stations for a recent Bay Area earthquake (20 July 2007, M4.2 Piedmont, CA). The stations have been ordered by increasing distance from the event (top to bottom). It is immediately apparent from this simple display that station MHDL was dead and needed immediate attention.

Table 3.7: Typical data streams acquired at NHFN sites, with channel name, sampling rate, sampling mode and FIR filter type. C indicates continuous, T triggered, Ca causal, and Ac acausal. Typically the DP1 continuous channel is archived and the remaining high sample rate data (i.e., CL and DP channels) are archived as triggered snippets. Prior to Sept. 2004, however, only triggered data was archived for all high sample rate channels.

Sensor	Channel	Rate (sps)	Mode	FIR
Accelerometer	CL?	500.0	T	Ca
Accelerometer	HL?	200.0	C	Ca
Accelerometer	BL?	20.0	C	Ac
Accelerometer	LL?	1.0	C	Ac
Geophone	DP?	500.0	T,C	Ca
Geophone	EP?	200.0	C	Ca
Geophone	EP?	100.0	C	Ca
Geophone	BP?	20.0	C	Ac
Geophone	LP?	1.0	C	Ac