A regression method for estimating salinity in the Ocean

Specification of salinity is an important problem in initialization of the global ocean circulation models. Unlike the temperature, the salinity data in the World Ocean are irregular and nonuniform; thus, methods for estimating the salinity using pleutiful temperature data are urgently needed. A new regression method for estimating the salinity in the ocean is suggested in this paper. Unlike similar currently known approaches, the method suggested applies a set of polynomials and their powers invariant for the entire ocean, while the latter is divided into a number of study subregions for the estimates. The best-fit regression curves and the minimal errors of the salinity estimates are found for each of the regions. The method uses the World Oceanographic Database (WOD-2001 NODC NOAA) to determine the regression coefficients and the confidence intervals (RCCI). A special RCCI database was organized on the basis of these data. The RCCI database makes possible determination of the salinity at any point in the ocean if the temperature data are provided (measurements, XBT profiles, etc.). A realization of the method suggested is demonstrated by the example of the Atlantic Ocean.