An Ocean Test of the A-0-A Approach to In-situ Calibrations of Pressure Sensors forLong-Term Seafloor Observations

Figure 11. September 7, 2017 (local time) Chiapas, Mexico earthquake.

Earthquake and Tsunami Records

Figure 7. Calibration measurements (left axis) and internal housingtemperature (right axis) plotted against time. The offset in temperature nearthe start of the deployment reflects a change in the operation of the valveactuator from power cycling to permanently on.

Figure 9. External pressure just before each calibration (left axes) and thedifferences in external pressures for the two sensors and internal pressuresfor the two sensors (right axes). The remarkable similarity between thelatter two curves is a result of the near constant span.

Figure 5. Day of pressure data that includes an A-0-A calibration plotted atfull scale to show the calibration (left axis) and magnified to show tides (rightaxis) The inset figure shows 10 minutes of data in which infragravity wavesand microseisms are clearly visible.

A-0-A Pressure Calibrations

IS24E-2615

William S. D. Wilcock1, Dana Manalang2, Michael Harrington2,Geoff Cram2, James Tilley2, Justin Burnett2, Derek Martin2, andJerome M. Paros31School of Oceanography, University of Washington, Seattle WA 981952Applied Physics Laboratory, University of Washington, Seattle WA 981953Paroscientific, Inc. & Quartz Seismic Sensors, Inc., Redmond, WA 98052

Figure 12. Tsunami for the January 23, 2018, Gulf of Alaska earthquake. Inthe lower panel, the de-tided pressure record is smoothed with a 3-minuterunning mean.

Figure 8. Calibration measurements after removal of a linear temperaturecorrection and then fit by curves of the form ∆p = C + Bt + A exp(-t/t0) wheret is time and C, B, A and t0 are constants chosen to optimize the fit. TheRMS misfit for each pressure sensor is ~0.1 hPa (1 mm of water).

Figure 10. Difference in the spans for the two pressure sensors afterapplying a linear correction for temperature. This plot shows that over the8-month deployment the spans diverge by only ~0.05 hPa (0.5 mm ofwater).

Figure 6. Example of a 5-minute calibration. The offsets between thepressure sensors and the accurate barometer are measured by averagingdata from 180 s to 240 s into the test.

AcknowledgementsParoscientific, Inc. provided guidance and engineering support for the effort to add remote control and operation capabilities to theGSSM.The MARS team at MBARI have provided engineering support throughout the design, testing, deployment and ongoing operationsof the GSSM.

Figure 1. Geodetic and Seismic Sensor Module (GSSM) comprising thefollowing suite of Digiquartz® sensors: two pressure sensors, a barometer,and a 3-component accelerometer. A valve, actuated from shore,periodically switches the pressure sensors between ambient seawaterpressure and internal housing pressure of ~1 atmosphere in order to correctfor long-term drift through comparison with the barometer. This process istermed “A-0-A”.

Figure 2. Remotely operated vehicle (ROV),Ventana prepares to deploy the GSSM on June13, 2017.

Figure 4. Two images of the ROV Ventanadeploying the cabled GSSM at a depth of 880m.

Deployment

Figure 3. Cabled GSSM in place on the seafloor near the MARS cabledobservatory node. Pressure and accelerations are transmitted in real timeto shore.

Key Figures

Abstract. Time-series observations of seafloor pressure are important for a varietyof oceanographic and geophysical research applications and for disaster warningsystems for earthquakes and tsunamis. Long-term measurements must distinguishbetween real changes in pressure and sensor drift that cannot be easily determinedfrom pre- and post- deployment calibrations. A means is needed to calibrate pressuremeasurements in situ to resolve signals from processes such as sea-level rise orvertical motion of the seafloor due to secular tectonic strain. One approach is to use adead-weight tester, an apparatus that produces an accurate reference pressure, tocalibrate a pressure sensor deployed on the seafloor by periodically switching betweenthe external pressure and the deadweight tester (Sasagawa et al., Earth Space Sci., 3,381–385, 2016). The “A-0-A” method replaces the reference pressure of the deadweight tester with the internal pressure of the instrument housing. We report on the firstnon-proprietary ocean test of this approach on the MARS cabled observatory at adepth of 900 m in Monterey Bay. We use the Paroscientific Seismic + Oceanic Sensors(SOS) module which comprises a three-component broadband accelerometer for co-located seismic measurements, two pressure sensors that for this deploymentmeasure an ocean pressure, “A”, up to 2000 psia (14 MPa), and a barometer tomeasure the internal housing reference pressure, “0”. A valve periodically switchesbetween external and internal pressures for 5-minute calibrations. The seafloor teststarted in mid-June and the results of >80 calibrations collected over the first 8 monthsof operation are extremely encouraging. After correcting for variations in the internaltemperature of the housing, the offset of the pressure sensors from the barometerreading as a function of time, can be fit with a smooth curve for each sensor with a RMSmisfit of ~0.1 hPa (1 mm of water). A comparison of the external pressures just prior toeach calibration with the pressures recorded during the calibration, shows that afterapplying a linear temperature correction, the apparent difference in the span of the twopressure sensors has drifted ~0.05 hPa (0.5 mm of water).