5 
1. Introduction 
To describe the mean of a state variable and its changes in a fluid such as the ocean or the 
atmosphere climatologies are used. The single variable should be representative, its 
precision stated and its changes statistically significant. The climatology in three or even 
four dimensions will describe parameter fields that should be physically meaningful. To 
describe a mean ocean circulation from those parameters will be misleading since there is 
never such a circulation: it is a statistically justified mean, only. 
Therefore constructing a climatology has to take into account the underlying relevant 
physical processes and their scales in space and time. This in turn has implications on the 
sampling, or data coverage. To adequately cover a global ocean in view of the needs of a 
climatology will never be possible, or affordable. So the existing data base on the one side 
determines the temporal and spatial limits of a climatology, the formulation of the 
interpolation scheme its representativeness on the other. 
Any ocean climatology will be a statistical artifact of an unadequately sampled changing fluid. 
But it is serving an important purpose: it is a reference data-set of the system and its 
variability. In view of the state of our knowledge about the slowly varying ocean today, such a 
reference serves well for supplying a bench-mark for a given and stated period to which one 
compares future data and their interpretation. 
Discussing climate change needs to define in the above sense the “mean” and the low- 
frequency variability of the ocean. Only then can we measure changes and try to attribute 
these to either the natural variability or anthropogenic change. 
The first climatology of the World Ocean by S. Levitus (1982) has become a common 
standard for the oceanographic community. Its profile and gridded data sets have been since 
widely used both by observationalists and modellers. The Ocean Climate Laboratory (OCL) 
of the US-NODC has since produced three improved versions of the 1982 climatology which 
appeared in 1994 (World Ocean Atlas 1994 (WOA94) (Levitus et al., 1994a-c) ), in 1998 
(World Ocean Database 1998 (WOD98) (Levitus et al. 1998)), and in 2002 (World Ocean 
Atlas 2001 (WOA01) (Conkright et al., 2002)). 
Despite the overall success of the NOAA climatologies a number of deficiencies have been 
noted and need addressing. Thus, one of the main problems with the NODC/OCL 
climatologies is the production of artificial water masses. As Lozier et al (1994) have shown, 
averaging of oceanographic properties on isobaric surfaces results in the production of water 
masses which are not confirmed by the temperature-salinity diagrams of observed data. It 
has been suggested that averaging should be done on isopycnal surfaces which mimics the 
process of isopycnal mixing in the real ocean and does not produce artificial water masses. 
Gouretski and Jancke (1999) found a large degree of scattering of the deep temperature- 
salinity diagrams in the South Pacific, based on the WOA94 climatology in disagreement with 
high-quality data from the region. They also found that the High Salinity Shelf Water in the 
Pacific Sector was completely missing in the gridded climatology. Curry (2000) reported that 
both WOA94 and WOD98 climatologies omit information from the deepest samples resulting 
in the disappearance of features such as the deep western boundary currents. 
A number of ocean climatologies for particular oceans and for the global ocean as a whole 
have been produced in the former Special Analysis Centre (WHP SAC) of the World Ocean 
Circulation Experiment (WOCE) Hydrographic Programme WHP Gouretski and Jancke, 
1995, 1996, 1998). This centre was closed in 1999 but some of its activities have been 
continued by a group in the German Federal Maritime and Hydrographic Agency (Bundesamt 
fur Seeschifffahrt und Hydrographie, (BSH)).