At the two sites, we deployed two different borehole strainmeters, both modifications of the original Sacks-Evertson (Sacks et al., 1971) design. Instruments of the original design have provided critical deformation data for studies of slow earthquakes (e.g., Sacks et al., 1981; Linde et al., 1988; Linde et al., 1996) and revealed new detail in volcanic activity (Linde et al., 1993; Linde et al., 1994). We installed a three-component strainmeter at Site 1150 and a single-component (dilatometer) version at Site 1151. This section will describe these designs as well as the electronic control system and signal conditioning that provides the analog signals for the A/D converters.
The dilatometer, which measures a single component (dilatation) of strain, resembles the Sacks-Evertson strainmeter. Figure F4 illustrates its basic principle of operation. The sensing volume is filled with a liquid (silicone oil). As the cylinder is deformed, oil is forced in or out of the attached bellows; the top surface of the bellows (B) correspondingly moves up or down. That motion is monitored by a differential transformer (DT). For a given strain, the bellows moves a distance proportional to the sensing volume divided by the bellows cross-sectional area. That ratio is very large (~40,000) and so the instrument provides high hydraulic (noise free) gain. This results in the instrument's having great sensitivity. To keep the strainmeter within its operating range over indefinite time intervals, a valve is opened for a few seconds as needed to allow oil to flow to or from a reservoir (R) that is decoupled from the strain field.
For installation into seafloor boreholes, significant modifications were required (Fig. F5). As before, strain changes are averaged over the 10-ft-long sensing section, providing considerable advantage over other designs that typically monitor a length of ~10-15 cm. The instrument was designed to operate at the bottom of a 1-km hole drilled in the ocean bottom under 3 km of water. For this high-pressure environment, we require walls in (19 mm) thick for an instrument diameter of 8 in (20 cm). Pressure testing verified that the dilatometer is capable of operating at pressures of up to 9000 psi (60 MPa). The actual pressure at installation time was ~6000 psi (40 MPa). The installation procedure (see "Installation Techniques") requires that cement be pumped through the strainmeter. An off-center cylinder (2-in diameter, -in wall thickness) allows this while retaining space for the DT-B subsystems.
The primary signal for strain changes is from a DT (DT1) monitoring the position of a -in-diameter bellows (B1). The B1-DT1 subunit frequency response to strain is constant from 0 to >10 Hz. The sensitivity is ~10-12 in strain with a maximum range of ~10-5. In operation, we expect strain changes greater than that maximum. The design allows for larger changes by incorporating a valve (V1) controlled by the electronic package. When V1 is opened, oil flows from or into the sensing volume allowing B1 to return to the equilibrium position. The combination of V1-B1-DT1 was the complete subunit for monitoring the strain changes in the original Sacks-Evertson design. We expect that the instrument will experience a long-term strain rate so that B1 will continue to move slowly away from the equilibrium position. Both to preserve dynamic range for short-term excursions and to obtain lower noise performance, we maintain the bellows close to equilibrium. If the B1 signal level is >15% of its range continuously for more than 15 min, the electronic control unit (Fig. F5; see "Electronic Control and Signal Conditioning") will open V1 and then close it again 10 s later. Because the instrument response remains constant over the seismic frequency range (order of hertz), the bellows experiences the strain changes caused by seismic waves. For large or nearby earthquakes, as will occur in the vicinity of Sites 1150 and 1151, such strain changes may exceed the maximum range of B1. The electronic control unit (Fig. F6; see "Electronic Control and Signal Conditioning") continuously checks that B1 is in a safe operating range. If the strain level becomes >60% maximum, V1 is immediately opened to prevent damage. Since this may happen when an earthquake occurs that is important to our program, we could lose valuable data while the valve is open (for ~10 min in this case). We avoid such data loss by incorporating a second monitoring subunit. The exit port from V1 does not connect directly to the oil reservoir but, rather, to a second, larger-diameter bellows (B2). The position of that 2-in-diameter bellows is monitored by a second DT (DT2); a second valve (V2) allows oil to flow between B2 and the unstressed reservoir (R). Whenever V1 is opened and closed, V2 will subsequently be opened after a delay of either 2 hr (following a low-threshold V1 trigger) or 8 hr (following the high-level trigger). Additionally, the output from DT2 is continuously checked and it is kept close to its equilibrium position using the same logic as for DT1. During normal operation the control system does not allow both valves to be open at the same time. Thus the sensing volume is always closed from the reservoir so that we are constantly recording the strain changes. Of course, when V1 is open the sensitivity is reduced by a factor of ~7. Under the common open condition (exceeding the low threshold for 15 min), there is normally no substantial change in the strain condition; only during the emergency high-level condition are we likely to have significant strain changes. But having a temporary gain decrease is of no real concern since the primary sensitivity is so high. When both valves are closed (the usual state during operation), B2-DT2 is monitoring the volume of a fixed mass of oil which is decoupled from the Earth's strain field. DT2 output then gives a measure of temperature change at the installation depth. The system parameters are such that we have a temperature change sensitivity of ~10-5°C.
The three-component instrument installed at Site 1150 operates using the same hydraulic principle employed in the single-component instrument but incorporates three separate sensing volumes. A schematic illustration (Fig. F7) shows a cross section of the sensing portion that consists of an inner ( in thick) and an outer ( in thick) wall with the annulus between them being divided into three equal segments. The objective is to measure, in effect, changes in three diameters (at 120°) of the cylinder, which allows us to record shear changes as well as areal or volume changes. Because the instrument is not circularly symmetric, we have made finite element calculations to verify that nonsymmetry in the design does not introduce significant error into the measurements. As with the high-pressure dilatometer design, this instrument was pressure tested up to 9000 psi to ensure that it could operate under installation conditions. The sensing length is 10 ft and the diameter is 8 in, the same as for the dilatometer. Apart from having three sensing volumes rather than one, the operation of this strainmeter is the same as for the dilatometer. Connected to each oil-filled sensing volume is a monitoring subunit identical to the B1-DT1-V1 combination in the dilatometer. The exit port connects to a second monitoring subunit, again identical to that in the dilatometer. Opening and closing of the valves occurs under the same conditions as in the single-component design, but now there are three sets of signals to monitor. If any of the three primary DTs (DT1a, DT1b, and DT1c) exceeds a threshold condition, all three primary valves (V1a, V1b, and V1c) are opened and then closed again. The secondary valves (V2a, V2b, and V2c) also operate in parallel. This strainmeter also provides high sensitivity and very broad-band frequency response. Again, both sets of valves are normally closed during operation, and the second set of DTs gives a sensitive measure of downhole temperature change.
The strainmeters must be bonded well to the host medium to provide meaningful information about changes in the strain state. Prior to the deployment of the original Sacks-Evertson strainmeter, the typical technique employed was to make the connection by means of a mechanical clamping system. Such systems had the advantage that the instrument could be unclamped and retrieved for repair in the event of instrument malfunction. But this came at the expense of instability in the recorded data; such mechanical systems are unable to maintain contact with the surroundings at the level of stability required for good quality data. The technique adopted for the Sacks-Evertson design sacrificed the retrieval capability to preserve the data quality. In all previous cases for installation on land we have used an expansive grout to make the connection between the instrument and the surrounding rock. A quantity of grout was lowered into the hole with a container that opened on reaching the bottom of the hole. After withdrawing that container, the strainmeter was lowered into the hole and sunk through the grout to be completely immersed. As the cement cured, it expanded and locked the instrument to the rock. Thus, the strainmeter subsequently was able to follow strain variations faithfully in the surrounding rock.
For this ocean-bottom installation that procedure must be modified. Sending the cement into the hole first is not feasible. The instrument system is designed to allow cement to be pumped through the instrument assembly and out below it and then fill to above the sensors. Because the strainmeter is already prestressed by the high pressure, it responds to variations in strain in the surroundings without the use of an expansive grout. Tests of the standard cement mixture used for grouting casing in the ODP have verified that its properties are suitable for connecting the strainmeter to the rock.
The strainmeters are operated and controlled by an electronic unit powered by the seawater battery, which provides a 42.6-V supply. A regulated voltage supply of 6.8 V powers the DTs in the single-component instrument. For the three-component version, all six DTs are powered in parallel (because of the limit to the number of conductors in the cable). We use a regulated supply voltage of 12 V, which is again regulated to 6.8 V in the instrument to eliminate cross talk generated by voltage drop in the 1100-m-long cable (resistance = 16 /km). The outputs of the DTs (the strain signals) are sampled at 20 samples per second (sps) by a multiplexed 16-bit A/D converter. Those digitized data are monitored by a microprocessor that is programmed to send commands to open and close the valves when the threshold conditions are met (see "Dilatometer Design"). A potential divider provides to the A/D converter an input of one-tenth the supply voltage. If the supply voltage falls below 24 V, indicating failure of the power supply unit, the valves are opened to protect the strainmeter. A capacitor across the input to the electronics has sufficient capacity to allow this protection even if the voltage supply is disconnected.
The strain signals are also sent to amplifiers that provide inputs to the 24-bit A/D converter modules (see "DM24"). The DT1 and DT2 signals have unity gain amplifiers so that the signal is always within the input voltage range of the digitizers. But a single 24-bit converter (which has ~2 bits of noise) does not cover the complete range of the strainmeter, particularly for the higher frequencies where earth noise is low. We also provide an amplified (20 × gain) output of the DT1 signals. This signal is high-pass filtered with a 3-dB corner frequency of 0.83 mHz (20-min period), which gives the high gain for the higher frequencies and also minimizes saturating the A/D converters as a result of long-period changes.
The ocean borehole seismometer (OBHS) is a package consisting of a three-component seismometer and a 24-bit digitizer (DM24) assembled in a stainless-steel pressure cylinder 1 m long with a 12.7-cm outside diameter. The seismometer is model CMG-1T made by Guralp Systems, Ltd., which consists of three orthogonal sensors, stacked vertically in the canister with a vertical sensor above the two horizontal sensors (Fig. F8).
The OBHS vertical sensor is a modified Lacoste type, and the horizontal sensor is an inverted pendulum. The inertial mass is a boom (a solid machined beam) supporting a transducer coil. The vertical sensor mass is supported by a prestressed triangular spring to support its weight and has a natural period of ~0.5 s. The horizontal sensor mass is centered by an unstressed flat triangular spring and has a natural period of ~1 s. The effective mass of each sensor is ~180 g. The springs are connected to the frames with a temperature compensating wire that minimizes the effect of temperature variation. A compact design is achieved chiefly by the short stiff springs and short boom.
The adjustments required for operation consist of leveling the boom of the vertical sensor and tilting the bases of the horizontal sensors to center the mass movements in their equilibrium positions. These are made by small (1 cm in diameter and 3 cm long) direct current (DC) motors operating gear mechanisms to tilt the horizontal sensor bases and to apply a small extra force to the vertical sensor's boom.
Before and during the installation in the borehole, the instrument may be subjected to severe motion that can damage the mass support hinges. The masses must be locked securely in their frames so that the hinges can be released. This operation is performed by a small motor-driven clamp, which is controlled by a command to the DM24.
The sensors employ feedback to expand their bandwidth and dynamic range. The sensor's response is determined by the characteristics of the feedback loop. The mass position is sensed by the capacitative position sensor. The voltage from the sensor, which is proportional to the displacement of the mass from its equilibrium position, is amplified and fed to a coil on the mass. The current in the coil forces the mass to its equilibrium position. With a high loop gain, motion of the mass is essentially prevented; the feedback voltage is then a measure of the force and, thus, the acceleration applied to the mass. The block diagram of the feedback system is shown in Figure F9. The system velocity response is identical to that of a conventional long-period sensor with a velocity transducer whose natural resonant period is 360 s with a damping factor of 0.707. The velocity output (flat to 100 Hz) is low-pass filtered (<50 Hz) before digitization. The mass position output can be used for periods longer than 360 s.
The output signals are digitized by the DM24. The detailed description of the DM24 module is given in "Seafloor Instruments". The velocity outputs are digitized at 100 sps, and the mass position outputs at 4 sps. The DM24 also digitizes the signal from the temperature sensor in the OBHS cylinder. All digital data are sent to the MEG through a 1.1-km-long cable. The communication link to the MEG is a four-wire RS422 serial link at 9600 bits per second (bps). As well as sending the signal data, the DM24 receives the time reference signals from the MEG and synchronizes the OBHS clock to the reference; this clock stamps the time in the digitized records.
The microprocessor in DM24 performs various controls of the sensors, such as unlocking/locking of the masses and bases of the horizontal sensors, and centering the masses. These controls are initiated by a command sent from the MEG or autonomously by the OBHS system. The masses are recentered whenever they deviate from the center position by more than half the range of travel.
The PMD seismometer is a three-component seismometer having a fairly broad frequency band response. This unit is a PMD model 2123 repackaged for borehole use. The amplitude response is essentially flat from 25 to 0.03 Hz with useful output down to 0.01 Hz.
The PMD seismometer sensors are robust and require no leveling or clamping. This makes them particularly suitable for borehole installations. The inertial component is a liquid—water with potassium iodide in solution—in a container which includes permeable grids that serve as cathode and anode. Between the grids a small bias voltage (<0.7 V) is applied, resulting in an ion current. Vibration of the instrument produces changes in the motion of the ionized liquid through the electrodes, resulting in a modulation of the ion current. These modulations are the output signal of the seismometer.
A prototype tested at the Nokogiriyama test borehole southeast of Tokyo had a power spectral density noise level [(m/s2)2/Hz] of about -150 dB at frequencies between 30 and 10 mHz. This is ~13 dB higher than for a CMG-1T unit (see "Ocean Borehole Seismometer") in the same borehole. Note that the Nokogiriyama site is not particularly quiet because it is only ~1 km from Tokyo Bay. The noise threshold of the PMD seismometer is probably reliable, but the site noise is 20 dB or more above the CMG-1T threshold. At frequencies >10 mHz, PMD seismometer noise is probably close to the expected ground noise. Below 30 mHz, the performance deteriorates. Figure F10 shows a comparison of PMD seismometer and CMG-1T noise spectra.
The reason for including a second broad-band seismometer with slightly inferior performance to the primary unit (CMG-1T) is the requirement for a long service life. Although it is impossible to predict the lifetime and failure rate of any complicated mechanism given similar quality and quality control, it is reasonable to expect that the PMD seismometer device with fewer components and moving parts will have a longer life.
The tiltmeter is the model 510 developed by Applied Geomechanics. The instrument employs two orthogonal electrolytic tilt sensors to provide the complete tilt vector change as a function of time. We will operate the instrument at maximum sensitivity to provide tilt resolution <10 nanorad with a maximum of 14 microrad. The principle of operation is illustrated in Figure F11. The sensor consists of a conductive fluid and an air bubble inside a curved tube. Excitation electrodes supply alternating current excitation and a pickup electrode provides the output signal. As the sensor tilts, the bubble moves relative to the electrodes and the sensor behaves as a variable resistor. The design is such that the output signal varies linearly with the applied tilt. The sensors and associated electronics are mounted in a vertical cylinder in a housing pressure tested at 7000 psi, well above the operating pressure for our sites.
Installing the tiltmeter so that the cylinder is vertical is infeasible, particularly in this below-seafloor installation, and we need a provision to keep the sensors on scale in case of large tilt changes during the instrument's operation. Thus each sensor is mounted on a platform that can be leveled by means of an electric motor. In our installation procedure, the tiltmeter may be subjected to large accelerations, particularly when the instrument string strikes the reentry cone. These tiltmeters have been specially equipped with rugged versions of the leveling units so that they will not be damaged mechanically during installation.
Included in the electronic unit mounted at the ocean bottom is circuitry to provide the signals necessary to drive the leveling motors. This circuitry is controlled by a microprocessor that uses the digitized signal output from a 16-bit A/D to determine which way to drive the leveling motors. The leveling operation will take place at initial system power-up and, subsequently, whenever the tilt sensors move away from the zero position by >40% of the range. The recorded data will come from the 16-bit A/D and the signals will be sampled at 4 sps.
Figure F12 shows ~10 days of data from a similar tiltmeter we installed previously at Nokogiriyama (southeast of Tokyo). Tidal signals are clearly recorded.
The orientation of each component of the different sensors is to be estimated from seismometer records of known origin.