Subsections


Plate Boundary Deformation Project

Introduction

Figure 7.1: Map of San Francisco Bay and San Juan Bautista area with existing borehole strainmeter and continuous GPS stations. The 5 new Mini-PBO stations (blue) complement the existing configuration of BARD GPS (red triangles) and borehole strainmeter sites (red circles) and BDSN and borehole HFN seismic stations (yellow).
\begin{figure*}\begin{center}
\epsfig{file=mpbo04_1.eps, width=10.5cm}\end{center}\end{figure*}

The Integrated Instrumentation Program for Broadband Observations of Plate Boundary Deformation, commonly referred to as ``Mini-PBO'', is a joint project of the BSL, the Department of Terrestrial Magnetism at Carnegie Institution of Washington (CIW), the IGPP at UC San Diego (UCSD), and the U.S. Geological Survey (USGS) at Menlo Park, Calif. It augments existing infrastructure in central California to form an integrated pilot system of instrumentation for the study of plate boundary deformation, with special emphasis on its relation to earthquakes. This project is partially funded through the EAR NSF/IF program with matching funds from the participating institutions and the Southern California Integrated Geodetic Network (SCIGN).

Because the time scales for plate boundary deformation range over at least 8 orders of magnitude, from seconds to decades, no single technique is adequate. We have initiated an integrated approach that makes use of three complementary and mature geodetic technologies: continuous GPS, borehole tensor strainmeters, and interferometric synthetic aperture radar (InSAR), to characterize broadband surface deformation. Also, ultrasensitive borehole seismometers monitor microearthquake activity related to subsurface deformation.

The project has three components: 1) the installation of broadband deformation stations in the San Francisco Bay area; 2) the installation of GPS stations in the Parkfield region; and 3) support for skeletal operations of a 5-m X-band SAR downlink facility in San Diego to collect and archive radar data, and develop an online SAR database for WInSAR users. The BSL has participated in the first two of these components. Additional details about the Parkfield GPS stations, installed in 2001 to link the BARD network in central and northern California to the SCIGN network in southern California and currently operating in real-time streaming mode with instantaneous position analysis, are provided in the BARD chapter of this report. The remainder of this chapter describes San Francisco Bay area broadband deformation station component of this project.

The broadband deformation stations augment existing instrumentation along the Hayward and San Andreas faults in the San Francisco Bay area (Figure 7.1). During July 2001 to August 2002, five boreholes were drilled and equipped with tensor strainmeters and 3-component L22 (velocity) seismometers (Table 7.1). These were the first deployments of a new type of strainmeter developed by CIW that use 3 sensing volumes placed in an annulus with 120-degree angular separation (Figure 7.2, which allows the 3-component horizontal strain tensor to be determined. All of the stations include pore pressure sensors and 2-component tiltmeters. Four of the stations now are equipped with GPS receivers recording at 1 sample per second and Quanterra recording systems that provide 100-Hz seismic and strainmeter data. The GPS antennas at these stations are mounted at the top of the borehole casings to achieve stable compact monuments. These stations complement existing Bay Area stations of the BARD continuous network, the BDSN and HFN seismic networks, and borehole dilatometers along the southern Hayward fault and in the San Juan Bautista region.

Mini-PBO Station
Configuration


Table 7.1: Operating stations of the Mini-PBO network. Strainmeter installation date is given. Depth to tensor strainmeter and 3-component seismometers in feet. High-frequency seismic and strainmeter data, and 1-Hz GPS data are available at all sites except MHDL.
Code Latitude Longitude Installed Strainmeter Seismometer Location
        depth (ft) depth (ft)  
OHLN 38.00625 -122.27299 2001/07/16 670.5 645.5 Ohlone Park, Hercules
SBRN 37.68622 -122.41044 2001/08/06 551.5 530.0 San Bruno Mtn. SP, Brisbane
OXMT 37.49936 -122.42431 2002/02/06 662.7 637.3 Ox Mtn., Half Moon Bay
MHDL 37.84227 -122.49374 2002/08/06 520.6 489.2 Golden Gate NRA, Sausalito
SVIN 38.03318 -122.52632 2002/08/29 527.0 500.0 St. Vincent CYO School, San Rafael


The general configuration of borehole instrument installation at each Mini-PBO station is shown in Figure 7.3. A 6.625" steel casing was cemented into a 10.75" hole to 500-650' depth to prevent the upper, most unconsolidated materials from collapsing into the hole. Below this depth a 6" uncased hole was drilled to the target region for the strainmeter and seismometer packages. Coring, in order to identify the region with the most competent rock for the strainmeter, was attempted with only moderate success at a few of the holes and was not attempted at St. Vincents. We found that video logs provided a reasonable substitute. The target region of each hole was filled with a non-shrink grout into which the strainmeter was lowered, allowing the grout to completely fill the inner cavity of the strainmeter within the annulus formed by the sensing volumes to ensure good coupling to the surrounding rock.

The 3-component seismometer package was then lowered to just above the strainmeter, on a 2" PVC pipe, and neat cement was used to fill the hole and PVC pipe to entirely enclose the package. The pipe above this depth was left open for later installation of the pore pressure sensor. To allow water to circulate into the pipe from the surrounding rock for the pore pressure measurements, the steel casing was perforated, a sand/gravel pack was emplaced, and a PVC screen was used at this depth. At each hole, the casing was then cemented inside to about 200', and outside to about 20' depth. A 12" PVC conductor casing was cemented on the outside from the surface to 20' to stabilize the hole for drilling and to provide an environmental health seal for shallow groundwater flow. The annulus between the 12" conductor casing and the 6.625" steel casing was cemented to about 10' depth and above was left decoupled from the upper surface to help minimize monument instability for the GPS antenna mounted on top of the steel casing.

The BSL developed a GPS mount for the top of the borehole casings to create a stable, compact monument. This design will be used at the more than 140 PBO strainmeter stations to be installed over the next 5 years. The antennas, using standard SCIGN adapters and domes for protection, are attached to the top of the 6-inch metal casing, which will be mechanically isolated from the upper few meters of the ground. The casing below this level is cemented fully to the surrounding rock. The GPS mount design consists of two 11-inch diameter stainless steel flanges. The lower slip- and- weld type flange is welded onto the top of the 6 5/ 8"- inch borehole casing providing a level surface for the second flange. The upper blind-type flange, to which the 1 1/ 4" stainless steel pipe used to connect to the SCIGN DC3 adaptor is attached, is bolted to the lower flange using four 3/ 4" by 3" stainless steel bolts. Two half- inch stainless steel dowels are press fit with high location precision (radius 7.500" +/- 0.001" ) into the lower flange. Two matching holes are machined into the upper flange with a high location precision (radius 7.500" +/- 0.001" ) and hole diameter precision (0.005"). One of the dowels is offset to insure unique directional alignment. During the period July 2003-June 2004, the BSL completed the installation of GPS systems using the borehole mount at St. Vincents School for Boys (SVIN) near San Rafael and at the Ox Mt site (OXMT) near Half Moon Bay.

Two-component tiltmeters were installed at all the stations by the USGS in 2003. Data from these sensors are recorded at 10-minute intervals and telemetered using the GOES system. Pore pressure sensors are also installed at all the stations and data are recorded at 1 Hz on the Quanterra dataloggers, except at Marin Headlands, where 10-minute interval data are also recorded on the Zeno datalogger.

The 1-Hz GPS, and 100-Hz strainmeter and seismometer data is acquired on Quanterra data loggers and continuously telemetered by frame relay to the BSL. Low frequency (600 second) data (including strainmeters, for redundancy) is telemetered using the GOES system to the USGS. All data is available to the community through the Northern California Earthquake Data Center (NCEDC) in SEED format, using procedures developed by the BSL and USGS to archive similar data from 139 sites of the USGS ultra-low-frequency (UL) geophysical network, including data from strainmeters, tiltmeters, creep meters, magnetometers, and water well levels.

The BSL is supervising GPS, power, frame relay telemetry, and Quanterra 4120 datalogger installation and maintenance at all the broadband deformation stations. Power, telemetry, and dataloggers are currently installed at all stations except MHDL, where we are waiting for PG&E to install the power drop. Telemetry at SVIN was established in June 2003 using Wi-LAN radios, a new type of radio that the BSL is currently beginning to adopt. These radios act as Ethernet bridges, providing superior access to console control on the Quanterras. The radios can also provide a spanning tree network structure for a regional wireless network, which allows greater flexibility for future network installations. We will use Wi-LAN radios at MHDL as well to provide a link between the borehole site and the frame-relay network interface circuit, which is located about 0.5 miles from the site.

Figure 7.2: Tensor strainmeter diagram. These instruments are a modification of hydraulic sensing dilatometer design to achieve a volume strain sensitivity of 10**(- 12) with constant frequency response from 0 to more than 10 Hz and a dynamic range of about 130 dB. The design incorporates a second bellows- DT- valve sub- system which provides extended dynamic range, complete preservation of baseline during required instrumental resets, and redundant sensing electronics. Figure courtesy A. Linde (USGS).
\begin{figure}\begin{center}
\epsfig{file=mpbo04_2.eps, width=8cm}\end{center}\end{figure}

Figure 7.3: The Mini-PBO borehole configuration at St. Vincents, showing the emplacement of the strainmeter and seismometer instruments downhole. The GPS receiver is mounted on the top. Figure courtesy B. Mueller (USGS).
\begin{figure}\begin{center}
\epsfig{file=mpbo04_3.eps, width=8cm}\end{center}\end{figure}

The BSL and USGS are addressing minor problems at the stations. Highly correlated low-amplitude noise is contaminating the seismic and strain channels at several of the stations, mostly due to electrical ground loop issues. The USGS and CIW are also investigating anomalies in the strainmeter channels, including unusual steps in several of the instruments related to resetting of the secondary valves that serve as an overflow buffer in the event of a large strain signal. These investigations have included tests of several variations of the strainmeter electronics boxes, which may have components that are poorly isolated. grounded. These tests, still ongoing, also introduce grounding loop problems that contaminate the high-frequency strain channels. In January 2004, the vertical component of the geophone at OHLN began to develop problems, which were most evident during seismic events. Testing indicated that the spring may have not functioned properly and that the sensor had lodged against a stop. The geophone returned to normal operation after it was pulsed with a high amplitude (.5V 4Hz) sine wave, which presumably dislodged the sensor.

Broadband Deformation Data

We are in the initial stages of assessing the data quality of the broadband deformation instrumentation. The borehole seismic packages provide good signal to noise characteristics compared to the NHFN stations due to their relatively deep installation. The systems have the best signal to noise near their 2-Hz characteristic frequency, but typical microseismic noise around 0.1 Hz is not evident (Figure 7.4). It is possible that the microseismic noise could be resolved if the systems included a pre-amplifier. We are planning to test this at OXMT and MHDL when the power and telemetry issues at those sites are resolved. These stations currently sample at 100-Hz, so they miss some of the seismic energy at high frequencies that are observed on the 500-Hz Parkfield borehole stations.

Analysis of GPS observations show that the short-term daily repeatabilities in the horizontal components are about 0.5-1 mm, and annual signals with about 1 mm amplitude. These values are similar to those obtained with more typical monuments, such as concrete piers or braced monuments, but it is too early to assess the long-term stability of the borehole casing monument, which might also be affected by annual thermal expansion effects on the casing.

Figure 7.4: Background noise measured by the borehole seismic packages at OHLN, SBRN, and SVIN. Component 1 is vertical. The systems have the best signal to noise ratio near their 2-Hz characteristic frequency. Typical microseismic noise around 0.1 Hz is not evident.
\begin{figure}\begin{center}
\epsfig{file=mpbo04_4.eps, width=8cm}\end{center}\end{figure}

The newly designed tensor strainmeters appear to faithfully record strain signals over a broad frequency range, except at the longest periods where the strains show a long-term exponential signal. This large signal is most likely due to cement hardening effects and re-equilibration of stresses in the surrounding rock in response to the sudden appearance of the borehole. These effects can last for many years and are the principal reason that borehole strainmeters cannot reliably measure strain at periods greater than a few months.

At periods around 1 day, tidally induced strains are the dominant strain signal. Since the response of the strainmeter volumes is difficult to estimate independently, theoretically predicted Earth tides are typically used to calibrate the strainmeters. Figure 7.5 shows the approximate microstrain of the OHLN strainmeter over a several month period interval, and some of the steps required to clean the data, including removing the tides and atmospheric pressure effects. The remaining signal is highly correlated with rainfall, indicating the extent that hydrologic events can affect strain. The proximity of the strainmeters to the San Francisco Bay complicates the determinations of theoretical tides due to ocean and bay loading effects. Figure 7.6 gives results from a tidal analysis provided by D. Agnew (UCSD) that shows good agreement in amplitude and phase of the two principal tidal components used for strainmeter calibrations (M2 and O1) at most of the stations, but a negative M2 tide at Ohlone, which is physically implausible.

At higher frequencies, strains due to seismic events are also evident. Figure 7.7 shows borehole strain measurements with clear seismic phases at OHLN for a 2003 Carlsberg Ridge earthquake. The dilatational component is bandpass filtered to show 100-300 second period Raleigh waves. Also shown are synthetic straingrams computed from a summation of the fundamental normal mode branch assuming a 1D Earth model (PREM), showing reasonably good agreement in amplitude. We are developing methods to use seismic surface waves to estimate the amplitude calibrations of the strainmeters to help resolve some of the problems observed with the tidal comparisons. The synthetic surface-wave and tidal-period amplitude calibrations currently agree to within 40%, which we expect will be significantly improved by using better Earth models and more sophisticated analysis techniques, such as phase-matched filtering.

We are also examining the strain data for other types of transient behavior, such as episodic creep or slow earthquake displacements. These multiparameter stations are providing a prototype for more than 140 planned borehole strainmeter, seismometer, and GPS stations in Cascadia and along the San Andreas fault as part of the Plate Boundary Observatory.

Acknowledgements

This project is sponsored by the National Science Foundation EAR/IF program with matching funds from the participating institutions and the Southern California Earthquake Center (SCEC) (PI Romanowicz).

Under Mark Murray's supervision, André Basset, Bill Karavas, John Friday, Dave Rapkin, Doug Neuhauser, Tom McEvilly, Wade Johnson, and Rich Clymer have contributed to the development of the BSL component of the Mini-PBO project. Several USGS colleagues, especially Malcolm Johnston, Bob Mueller, and Doug Myren, played critical roles in the drilling and instrument installation phases.

Figure 7.5: Four-month subset of OHLN data, detrended to flatten the first 50 days (middle trace), separated using BAYTAP-G (Tamura et al., 1991) into tidal, atmospheric pressure, and "cleaned" data components (with arbitrary vertical offsets). The atmospheric pressure time series measured at the site was also used for this decomposition. Approximate microstrain values are based on peak-to-peak tidal amplitude. The remaining large strain signals in the cleaned data are highly correlated with rainfall measured at an instrument located about 5 km from the site (40 cm total cumulative rainfall during this interval), and therefore are probably not geophysically interesting.
\begin{figure}\begin{center}
\epsfig{file=mpbo04_5.eps, width=8cm, clip=}\end{center}\end{figure}

Figure 7.6: Tidal analysis performed by D. Agnew on areal strain from the Mini-PBO stations, showing agreement within about 20% in both amplitude and phase between the O1 and M2 tides, except at Ohlone (oh), whose negative M2 tide is physically implausible.
\begin{figure}\begin{center}
\epsfig{file=mpbo04_6.eps, width=9cm}\end{center}\end{figure}

Figure 7.7: Comparison between data (blue) and synthetic (red) straingrams at the OHLN strainmeter for the 2003 Carlsberg Ridge earthquake. The dilatational component is bandpass filtered to show 100-300 second period Raleigh waves. The synthetics are computed from a summation of spheroidal modes assuming a 1D Earth model (PREM). Surface-wave and tidal-period amplitude calibrations agree to within 40%, which we expect will be significantly improved by using better Earth models and more sophisticated analysis techniques, such as phase-matched filtering.
\begin{figure}\begin{center}
\epsfig{file=mpbo04_7.eps, width=9cm}\end{center}\end{figure}

Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94720-4760
Questions or comments? Send e-mail: www@seismo.berkeley.edu
© 2004, The Regents of the University of California