Deep-ocean Bottom Pressure and Temperature Sensors Report: Methods and Data

This report documents ocean bottom pressure data collected from September 1983 to May 1985 in eleven deployments of pressure sensors under the Gulf Stream northeast of Cape Hatteras in depths of 3300 to 3900 m, as part of the Gulf Stream Dynamics Experiment. In past experiments, pressure sensors suitable for ocean depths have typically exhibited systematic drifts in calibration that seriously contaminate any observed periodicities longer than a few days. We used Digiquartz sensors (manufactured by Paroscientific, Inc.), because these sensors offered potentially much lower drift than other commercially available sensors. In these sensors, either a bellows or a Bourdon tube applies stress to an osciilating quartz-crystal beam, causing its oscillation frequency to vary. Several factors influence the amount of drift: bellows vs. Bourdon-tube construction, the applied pressure, the duration of deployment, and, for some sensors, high-pressure preconditioning in the lab. For the sensors deployed in the Gulf Stream, the total drift during deployments lasting from 3 to 12 months ranged from undetectable (≦0.01 dbar) to 0.20 to 0.50 dbar. About half of the total drift typically occurred within the first 6 days of deployment. We estimate the residual error in the final pressure records, after the dedrifting calculations, to be typically 0.02 dbar r.m.s. (or 0.06 dbar r.m.s.) if the first 6 days of the record are excluded (or included, respectively). This low drift-error opens many possibilities for studies that require knowledge of the low-frequency dynamic pressure signal in the deep ocean. Part I on Methods contains a short review of bottom pressure measurement in the deep ocean, a description of the sensors that we used, a discussion of their performance and drift relative to type of construction and prior pressurization history ( preconditioning ), and estimates of the accuracy of the dedrifted pressure records. In Part II of this report, the full data processing is described, including calibaration parameters, corrections for the influence of temperature variations on the pressure sensor, and parameterization to remove sendor drift errors by least-squares regression onto an exponentially decaying time-dependence. Time series are plotted which illustrate several steps in the processing: the edited half-hourly pressure records, the detided pressures with drift-model curves superimposed, and the low-pass filtered, dedrifted pressure records (i.e., after subtracting the estimated drift curve)