Vertical Sections Information

The hydrographic and chemical properties measured along each line are shown in the vertical sections, plotted as a function of depth. For each line sections are shown for up to fourteen parameters: Potential temperature, salinity, neutral density, potential density, oxygen, nitrate, phosphate, silicate, CFC-11, total CO2, alkalinity,

helium, tritium and radiocarbon (see definitions and acronyms).

Although samples were collected on all cruise lines as detailed in Table 1, not all the data are represented in this atlas, either because the data were not made available to the WHPO by the deadline of September 2002 or the sampling density was not high enough to construct the sections.

Sections of potential temperature, salinity, neutral density and potential density are constructed from CTD data, not discrete bottle samples. Neutral density was calculated from the raw data following the method of Jackett and McDougall (1997), and potential density from the 1980 Equation of State (UNESCO, 1981). Potential density sections of
sigma-0 are shown above 1000 m, of sigma-2 from 1000-3000 m and of sigma-4 below 3000 m.

The sampling strategy for WOCE cruises generally provided closer station spacing over ocean ridges and continental slope regimes, where the expected scales of variability are smaller than in the oceanic regime. Vertical sections were constructed using optimal mapping (Gandin, 1965; Bretherton et al., 1976). This algorithm simply solves an equivalent least square problem applied to a practical subset of nearby measurements, i.e. a minimum variance solution. A uniform vertical grid spacing of 20 m was adopted for mapping the CTD data, whereas bottle data were mapped onto a vertical grid whose spacing increases progressively from 10 m near the sea surface to a maximum of 100 m at depths greater than 1000 m.

Horizontally, three additional equally-spaced grid points were positioned between each pair of stations. Elliptical correlation areas vary as a function of the local grid spacing. At each grid point a horizontal:vertical correlation length ratio of 7:2 (7:4) times the local grid spacing was used while mapping CTD (bottle) data. All gridded property fields were initially machine-contoured, but the resulting patterns were manually edited after careful inspection of property values measured at each sample position.

The vertical sections are constructed as a function of cumulative distance along the line, starting at the westernmost or southernmost station. Each section consists of an upper panel showing the sea surface to 1000 m and a lower panel showing the full depth range. The vertical scale in the lower (upper) panel is such that one inch corresponds to a water depth interval of 1000 m (400 m). For the meridional sections a vertical distortion (distance:depth) of 400:1 is used in the full water-column plots and of 1000:1 in the expanded plots of the upper 1000 m. The long zonal section uses vertical distortions of 600:1 and 1500:1 for its lower and upper panels. Station locations are indicated with tick-marks at the top of the upper panel. Interpolated latitude/longitude along the section is shown with tick-marks at the top of the lower panel. The bottom depth at station locations is taken from ship records, and the altimeter-derived bathymetric data (Smith and Sandwell, 1997) was projected between stations to construct the bottom topography used in the sections.

Contour intervals have been selected to emphasise the important features within each set of measurements. Colors have been chosen as far as possible to agree with those used in the GEOSECS atlases, with the exceptions of the CFCs, tritium,
helium, and radiocarbon. A particular color scheme was chosen for the Southern Ocean vertical sections.

Four shades of two colors have been used for all properties, varying from 100% of the base color at one extreme of the property to 25% at an intermediate level. Color or shade changes illustrate the major water masses of the Southern Ocean, and do not necessarily correspond to the same isoline in the other volumes of the WOCE atlases. Although efforts were made to keep the contour interval constant within a particular color shade, this was not always possible. Neighbouring contours are clearly labelled where this occurs. Contour intervals may also change from one shade to another.

Property-property plots

Scatter plots of two variables are frequently used to discriminate between different water masses. There are many possible combinations of property-property plots for the parameters shown in the atlas. The printed atlas shows only properties versus potential temperature. These are among the more commonly used relationships, but researchers can construct additional property-property plots from the online Southern Ocean Atlas (http:/ The plots include data from all stations along a given section. The color separation for the property-property plots is a function of depth.

The property-property plots use eight colors to indicate different depth intervals (inset, color scheme). In the full scale plots, dots corresponding to the shallower depth intervals overlie those from the deeper layers. The insets show data deeper than 1000 m, with the opposite color stacking order.