Subsections


Data Acquisition and Quality Control



Introduction

Stations from the networks operated by the BSL transmit data continuously to the BSL facilities on the UC Berkeley campus for analysis and archival. In this section, we describe activities and facilities which pertain to the individual networks described in Sections 1, 3, and 4, including procedures for data acquisition and quality control, and sensor testing capabilities and procedures. Some of these activities are continuous from year to year and have been described in prior BSL annual reports. In this section, we describe changes or activities which are specific to 2008-2009.

Figure 3.24: Data flow from the BDSN, NHFN, MPBO, HRSN and BARD network into the BSL central processing facility.
\begin{figure*}\begin{center}
\epsfig{file=acq_dataflow.eps,width=10cm}\end{center}\end{figure*}

Data Acquisition Facilities

The computers and the associated telemetry equipment are now located in the campus computer facility in Warren Hall at 2195 Hearst Avenue. This building was constructed to current ``emergency grade'' seismic codes and is expected to be operational even after a $M$ 7 earthquake on the nearby Hayward Fault. The hardened campus computer facility within was designed with special attention for post-earthquake operations. The computer center contains state-of-the art seismic bracing, UPS power and air conditioning with generator backup, and extensive security and equipment monitoring.

Data Acquisition

Central-site data acquisition for data from the BDSN/HRSN/NHFN/MPBO networks is performed by two computer systems in the Warren Hall data center (Figure 3.24). These acquisition systems also collect data from the Parkfield-Hollister electromagnetic array and the BARD network. A third system is used primarily for data exchange with the USNSN and transmits data to the USNSN from HOPS, CMB, SAO, WDC, HUMO, MOD, MCCM, and YBH. Data for all channels of the HRSN are now telemetered continuously from Parkfield to the BSL over the USGS T1 from Parkfield to Menlo Park, and over the NCEMC T1 from Menlo Park to Warren Hall.

Figure 3.25: Flow of data from comserv areas through network services processing. One stream of the network services provides picks (and currently still provides codas) determined using the programs shown in the right flow path. Every 5 seconds, ground motion parameters are also determined, including PGA, PGV, PGD, and ML100 (left flow column). Parameters from the network services are available to the CISN software for event detection and characterization. Data are also logged to disk (via datalog), distributed to other computers (mserv), and spooled into a trace ring for export.
\begin{figure*}\begin{center}
\epsfig{file=BK_netserv.ps,width=10cm}\end{center}\end{figure*}

The BSL uses the programs comserv and qmaserv developed by Quanterra for central data acquisition. These programs receive data from remote Quanterra data loggers and redistribute it to one or more client programs. These clients include datalog, which writes the data to disk files for archival purposes, wdafill, which writes the data to the shared memory region for processing with the network services routines, and other programs such as the seismic alarm process, the DAC480 system, and the feed for the Memento Mori Web page.

The two computers performing data acquisition are also ``network services'' computers that reduce waveforms for processing with the CISN software (Figure 3.25). To facilitate processing, each system maintains a shared memory region containing the most recent 30 minutes of data for each channel.

Each BDSN data logger using frame relay telemetry is configured to enable data transmission simultaneously to two different computers over two different frame relay T1 circuits to UCB. Normally, only one of these circuits is actively enabled at any given time. The comserv/qmaserv client program cs2m receives data and multicasts it over a private ethernet. The program mcast, a modified version of Quanterra's comserv program, receives the multicast data from cs2m, and provides a comserv-like interface to local comserv clients. Thus, each network services computer has a comserv/qmaserv server for every station, and each of the two systems has a complete copy of all waveform data.

We have extended the multicasting approach to handle data received from other networks such as the NCSN and UNR. These data are received by Earthworm data exchange programs and are then converted to MiniSEED and multicast in the same manner as the BSL data. We use mserv on both network services computers to receive the multicast data and handle it in the same way as the BSL MiniSEED data.

In 2006, the BSL established a real-time data feed of all BSL waveforms between the BSL acquisition systems and the NCEDC computers using the open source Freeorb software. This allows the NCEDC to provide near-real-time access to all BSL waveform data through the NCEDC DART (Data Availabile in Real Time) system.

We monitor seismic stations and telemetry using the program seisnetwatch. This program extracts current informaton such as time quality, mass positions, and battery voltage and allows it to be displayed. If the parameter departs from the nominal range, the station is marked with yellow or red to indicate a possible problem.

Seismic Noise Analysis

BSL seismic data are routinely monitored for state-of-health. An automated analysis is computed regularly to characterize the seismic noise level recorded by each broadband seismometer.

PSD Noise Analysis

The estimation of the Power Spectral Density (PSD) of the ground motion recorded at a seismic station, as documented in the 2000-2001 BSL annual report (http://seismo.berkeley.edu/annual_report/), provides an objective measure of background seismic noise characteristics over a wide range of frequencies. It also provides an objective measure of seasonal and secular variation in noise characteristics and supports early diagnoses of instrumental problems. In the early 1990s, a PSD estimation algorithm was developed at the BSL for characterizing the background seismic noise and as a tool for quality control. The algorithm generates a bar graph output in which all the BDSN broadband stations can be compared by component. We also use the weekly PSD results to monitor trends in the noise level at each station. Cumulative PSD plots are generated for each station and show the noise level in 5 frequency bands for the broadband channels. The plots make it easier to spot certain problems, such as failure of a sensor. In addition to the station-based plots, a summary plot is produced for each channel is produced. The figures are presented as part of a noise analysis of the BDSN on the web at http://www.seismo.berkeley.edu/seismo/bdsn/psd/.

PDF PSD Noise Analysis

In addition to the PSD analysis developed by Bob Uhrhammer, the BSL has implemented the Ambient Noise Probability Density Function (PDF) analysis system developed by McNamara and Buland (2004). This system performs its noise analysis over all the data of a given time period (week or year), including earthquakes, calibration pulses, and cultural noise. This is in contrast to Bob Uhrhammer's PSD analysis, which looks at only the quietest portion of data within a day or week. Pete Lombard of the BSL extended the McNamara code to cover a larger frequency range and support the many different types of sensors employed by the BSL. Besides the originally supported broadband sensors, our PDF analysis now includes surface and borehole accelerometers, strain meters, and electric and magnetic field sensors. These enhancements to the PDF code, plus a number of bug fixes, were provided back to the McNamara team for incorporation in their work. The results of the PDF analysis are presented on the web at http://www.ncedc.org/ncedc/PDF/. One difficulty with using these plots for review of station quality is that it is necessary to look at data from each component separately. To provide an overview, we have developed summary figures for all components in two spectral bands, 30 - 60 s and 0.125 - 0.25 s, which These will soon be available on the web.

Sensor Testing Facility

The BSL has an Instrumentation Test Facility in the Byerly Seismographic Vault where the characteristics of up to eight sensors can be systematically determined and compared. The test equipment consists of an eight-channel Quanterra Q4120 high-resolution data logger and a custom interconnect panel. The panel provides isolated power and preamplification, when required, to facilitate the connection and routing of signals from the sensors to the data logger with shielded signal lines. The vault also has a GPS rebroadcaster, so that all data loggers in the Byerly vault operate on the same time base. Upon acquisition of data at up to 200 sps from the instruments under test, PSD analysis, coherence analysis, and other analysis algorithms are used to characterize and compare the sensor performance. Tilt tests and seismic signals with a sufficient signal level above the background seismic noise are also used to verify the absolute calibration of the sensors. A simple vertical shake table is used to assess the linearity of a seismic sensor. The sensor testing facility of the BSL is described in detail in the 2001-2002 Annual Report (http://www.seismo.berkeley.edu/).

Several projects made use of the sensor testing facility in 2008-2009. These included final testing of the new STS-1 electronics (E300) and initial testing of the STS-1 type sensors being developed jointly by Metrozet and the BSL. Data were also collected from the new pressure/temperature sensors (see below).

Meteorological Sensors

A new meterological (MET) sensor package which measures temperature, relative humidity, and pressure (THP) is being developed and tested at BSL as a replacement for the aging temperature and pressure sensors now at the BDSN stations. Temperature and pressure measurements at the stations are useful for reducing the components of the seismic background noise that are correlated with these parameters. A hygrometer has been added to the sensor package to allow measurement of the local relative humidity, a parameter which is potentially useful for estimating and correcting for GPS tropospheric propagation delays.

During the past year we have tested the BSL THP sensor package on the McCone Hall roof and at SBRN. In a cluster test at BRIB, a pair of the BSL MET sensor packages were operated with a pair of commercial MET sensors (a Paroscientific 1477-005 and a Vaisala WXT-510). Specifications for the commercial MET sensors, and specifications and calibration of the BSL MET sensors are given below.

Commercial MET Sensors A Paroscientific sensor (Model 1477-005; s/n 101728), the most accurate of the MET sensors being tested, is used as the reference standard. The pressure accuracy is $\pm{8}$Pa, the temperature accuracy is $\pm{0.5}$ $^{\circ}$C, and the relative humidity accuracy is $\pm{2}\%$ (http://paroscientific.com/pdf/MET.pdf). The Vaisala MET sensor (Model WXT-510; s/n C4760003), on loan from UNAVCO, is being tested for possible use at BARD stations. The pressure accuracy is $\pm{50}$Pa, the temperature accuracy is $\pm{0.3}$C, and the relative humidity accuracy is $\pm{3}$% (http://www.vaisala.com/instruments/products/weathermultisensor.html). In addition, this sensor measures wind speed and direction with accuracies of $\pm{3}$% and $\pm{3}$$^{\circ}$, respectively.

BSL MET Sensors The sensor elements in the BSL MET package are manufactured by Honeywell (http://sensing.honeywell.com). The pressure sensing element, Honeywell SDX15A2-A, is temperature compensated and has a typical accuracy of $\pm{0.25}$%. The specification sheet indicates that the sensor range of 0-15 psi in absolute pressure results in a 90mv ($\pm{1}$%) differential change on the outputs when the bridge is excited with 12V. The sensor is operated in a bridge circuit configuration and its sensitivity is:

$P(Pa) = 689.825 * V + 92643$

where: V is the bridge output in Volts.

The thin film platinum resistance temperature detector (RTD) is a Honeywell HEL-700 with a resistance of 1k$\Omega$ at 0$^{\circ}$C and with an accuracy of $\pm{0.3}$$^{\circ}$C. The RTD is operated in a circuit with offset and gain and its sensitivity is:

$T($° ${C}) = 2.8877 * V + 26.248$

where: V is the output in Volts.

The humidity sensing element is a Honeywell HIH-4602C which is sensitive to the relative humidity with an accuracy of $\pm{3.5}$%. The sensor is operated in a circuit which results in a overall calibrated sensitivity of:

$\%RH = ( ( V + 9.29 ) / 3.168 - Z ) / S$

where: %RH is the percent relative humidity and V is the voltage output. S and Z are given in the factory calibration sheet as Z $\sim$0.826 mV and S $\sim$31.5 mv/%RH. Thus:

$\%RH = 10.021 * V + 66.87.$

where: V is the output in Volts.

The factory specification sheet indicates that the response time is $\sim$50 seconds and the accuracy is $\pm{3.5}$%RH.

The absolute humidity (AH) is a function of temperature, and, given the temperature, AH can be derived from relative humidity (RH) via:

$AH(g/m^{3}) = ( 0.000002T^{4} + 0.0002T^{3} + 0.0095T^{2} + 0.337T + 4.9034 ) * RH $

where: T is the temperature in $^{\circ}$C.

BRIB MET sensor Cluster Test Four MET sensors (the Paroscientific, the Vaisala, and two BSL instruments) have been operating in a closely spaced cluster at BRIB since 2009.211 as seen Figure 3.26.

Figure 3.26: Cluster of MET sensors under test at BRIB. The sensors from left to right are; Paroscientific; BSL MET 2, BSL MET 1, and Vaisala. The electronics for the BSL MET sensors are remotely located approximately 40 feet away in the vault which contains the BRIB supporting hardware.
\begin{figure}\begin{center}
\epsfig{file=rau09_1_F1.eps, width=7cm, bbllx=33, bblly=90, bburx=580, bbury=702}\end{center}\end{figure}

Data Acquisition The Paroscientific and Vaisala MET sensors output digital data and they are connected to port 2 of two different NetRS (Trimble GPS reference stations) and embedded in the GPS data streams at a rate of 1 sample per minute. The two BSL MET sensor output voltages proportional to temperature, relative humidity and pressure which are digitized/recorded by a single 6-channel Quanterra Q4120 data logger at a rate of 1 sample per second. The Paroscientific and Vaisala MET data are extracted from the GPS data streams using the teqc program and the BSL MET 1 and 2 data were extracted using the qmerge program.

One problem which was encountered is that there are gaps in MET data from the Paroscientific and Vaisala sensors, the zero minute of each hour and occasionally at other times are missing and we are currently investigating the cause.

Test Results Ten days of MET data (2009.211-220) was retrieved for the four sensor packages an the data were interpolated/decimated to 1 sample per minute for analysis.

The two BSL MET sensors were calibrated via least squares using the Paroscientific as the reference. The calibration of the two BSL MET sensors channels are given in terms of the offset (a) and the sensitivity (b) with the McLauren series form: f(c) = a + b * counts. The results are presented in Table 3.13. Note that the a and b values for the second relative humidity sensor are anomalously large owing to a malfunction in the electronics. Even so the calibration is reasonable given the 5.1 percent standard error.

Comparing the Vaisala MET sensor with the reference Paroscientific MET sensor data, we find that the pressure, temperature and relative humidity standard errors are: 0.159mbar (15.9Pa), 0.292$^{\circ}$C, and 4.38%. All values are within the factory specifications for accuracy of the Vaisala WXT-510 and thus the differences are not significant.

In summary, the BSL in-house MET sensors preform very well. They accuracy and resolution are sufficient for use to reduce the components of the seismic background noise that are correlated with temperature and pressure and for estimating and correcting for geodetic GPS tropospheric propagation delays.


Table 3.13: Calibration parameters for BSL MET sensors where D is pressure in Pa, K is temperature in $^{\circ}$ C, and I is relative humidity in % .
MET Sensor a b se
1 D 93093 0.0016902 31.7
2 D 93144 0.0016531 63.0
1 K 14.83 0.0000047665 1.13
2 K 21.88 0.0000054509 1.07
1 I 5391. 0.0023869 5.12
2 I 58.70 0.000019885 3.92


STS 1 Development and testing

The BSL is participating in the NSF-funded re-development of the STS1 instruments. BSL's role has principally been to objectively test, evaluate, and compare the old STS1s with new instruments developed by Metrozet. Additionally, with input from Metrozet, BSL has developed and fabricated a new baseplate for installing these seismometers.

In December 2008, Metrozet brought newly developed STS1-H instruments to Berkeley for initial evaluation. These instruments were installed at the test facility on the new baseplate.

Several problems typical of any new installation were encountered. These include difficulties with cabling, insulation, connector orientation, and vacuum leaks. The problems were quickly resolved and the evaluation proceeded. Initial assessments indicate that both the Metrozet seismometer and the Berkeley designed baseplate are promising.

As many of the problems experienced with STS1 seismometers over time are related to their installation, the BSL has focused on developing both a new baseplate and an acceptable retrofit to the original warpless baseplate. We have developed and are testing replacement parts that would ensure the vacuum on the instruments.

Acknowledgements

Doug Neuhauser, Bob Uhrhammer, Peggy Hellweg, Pete Lombard, Rick McKenzie, and Jennifer Taggart are involved in the data acquisition and quality control of BDSN/HRSN/NHFN/MBPO data. Development of the sensor test facility and analysis system was a collaborative effort of Bob Uhrhammer, Tom McEvilly, John Friday, and Bill Karavas. IRIS and DTRA provided, in part, funding for and/or incentive to set up and operate the facility, and we thank them for their support. Vaisala WXT-510 MET sensor package was on loan from UNAVCO. Bob Uhrhammer, Peggy Hellweg, Pete Lombard, Doug Neuhauser, and Barbara Romanowicz contributed to the preparation of this section. The STS-1 project is funded by the NSF through the IRIS/GSN program (IRIS Subaward Agreement number 388). This is a collaborative project with Tom VanZandt of Metrozet, LLC (Redondo Beach, CA).

References

Ekström, G. and M. Nettles, http://www.seismology.harvard.edu/~ekstrom/Projects/WQC.html, 2005.

Gardner, W. A., A unifying view of coherence in signal processing, Signal Processing, 29, p. 113-140, 1992.

Ingate, S. et al, Workshop Report from Broadband Seismometer Workshop, Lake Tahoe, CA, http://www.iris.edu/stations/seisWorkshop04/report.htm, 2004.

McNamara, D. and R. Buland, Ambient Noise Levels in the Continental United States Bull. Seism. Soc. Am., 94, 4, 2004.

Scherbaum, Frank. Of Poles and Zeros: Fundamentals in Digital Seismology, Volume 15 of Modern Approaches in Geophysics, G. Nolet, Managing Editor, Kluwer Academic Press, Dordrecht, xi + 257 pp., 1996.

Tapley, W. C. and J. E. Tull, SAC - Seismic Analysis Code: Users Manual, Lawrence Livermore National Laboratory, Revision 4, 388 pp., March 20, 1992.

Wielandt, E. and G. Streckeisen, The leaf spring seismometer: design and performance, Bull. Seis. Soc. Am., 72, 2349-2367, 1982.

Wielandt, E. and Steim, J. M., A digital very broad band seismograph, Annales Geophysicae, 4 B(3), 227-232, 1986.

Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail: www@seismo.berkeley.edu
© 2007, The Regents of the University of California