The dataset consists of 115 CTD casts in the region north of Flemish Cap. Some casts cover the full water column, while others only cover the upper 1000 db. The CTD casts were obtained with a SeaBird SBE911+ system, measuring temperature (2 sensors), conductivity (2 sensors), pressure, beam transmission, height above the bottom, oxygen (2 sensors), and chlorophyll fluorescence.

All sensors were sampled at 24 Hz. The data were processed using the SeaBird data processing software suite, SBEDataProcessing-Win32, and with software in MATLAB. A low pass filter, with time constant of 0.15 s, was applied to the pressure record. To account for the transit time between the temperature and conductivity sensors, the conductivity measurements were aligned with the temperature measurements using empirically determined time delays.

The primary conductivity was delayed by 0.035 s relative to pressure (this is in addition to the advance of 0.073 s which is performed by the SeaBird deckbox during data acquisition, thus resulting in a net advance of 0.038 s). The secondary conductivity was advanced by 0.048 s (except for cast ctd001, which had the automatic deckbox advance value of 0.073 s applied. Thus the secondary conductivity from this cast was delayed by 0.025 s, giving a net advance of 0.048 s). The two oxygen voltages were advanced by 4 s relative to pressure. A correction for conductivity cell thermal mass effects was applied to both conductivity channels using the parameters recommended by SeaBird (alpha=0.03, 1/beta=7.0). The temperatures, conductivities, and oxygen voltages were then median filtered using a 7-scan window. A loop edit step was then applied, whereby portions of the cast in which the pressure was not changing sufficiently fast (0.2 dbar/s) were removed. This was followed by computation of salinity, potential temperature, potential density, sound velocity, geopotential anomaly, and oxygen concentration. Unfortunately, it was found that the SeaBird data processing module Derive used the primary temperature and salinity in computing both primary and secondary oxygen. Because there were several casts during which the primary temperature sensor intermittently failed, this resulted in loss of oxygen data. To get around this problem, the oxygen calculation was performed separately in MATLAB using the SeaBird algorithm and with primary/secondary oxygen computed using primary/secondary T and S respectively.

Finally, the data from the downcast were averaged into 1 dbar bins. Further details of the CTD data processing can be found in the header portion of the individual cast files. The final data files contain raw sensor values (1 dbar bin averages) plus a number of derived variables (e.g., potential temperature, salinity, sigma-theta, oxygen).

A full list of the output variables is contained in the header portion of the cast files. The casts were visually examined to determine the quality of the data from the 2 separate sensor suites (primary and secondary). A header line was placed in each file indicating the preferred sensor pair (PRIMARY or SECONDARY) if one was bad or whether both were of equal quality (BOTH GOOD).