This proposal addresses ocean modeling from an inverse point of view. Ocean data of varying types and quality have been collected for many years. In many analyses, a subjective approach is taken to describe the distribution of salinity and other tracers. This study will continue efforts to develop quantitative methods to constrain the data analyses. In particular, models will be made to conserve mass, momentum, and chemical constituents according to prescribed rules which describe the data and the errors associated with it. Specific attention will be focused on the North Atlantic, but other areas with relevant data bases will also be included. This work is an important element in the World Ocean Circulation Experiment (WOCE).