Automated system and method for vertical gradient correction

Description (excerpt)

CROSS-REFERENCE TO RELATED APPLICATIONS The present application is related by subject matter and inventors to U.S. Pat. No. 8,032,314 entitled MLD-Modified Synthetic Ocean Profiles, filed Sep. 29, 2008, issued on Oct. 4, 2011, which is incorporated by reference herein in its entirety. BACKGROUND Ocean temperature and salinity surface observations can be extrapolated to depths in the ocean by known methods, and the errors in performing such an extrapolation can be reduced, also according to known methods. However, vertical gradients tend to remain overly smooth. What is currently done is that the average temperature and salinity over various horizontal distances at each depth of interest are determined for normal conditions in a particular location. Operational ocean forecasting systems rely on a relative abundance of ocean surface observations for data assimilation into numerical models. Commercial fishermen utilize ocean forecasts to identify productive fishing grounds associated with ocean fronts and other dynamic ocean features. All numerical ocean forecasts rely on data assimilation. A common component of these systems aggregates surface observations and projects surface information downward to estimate the 3-dimensional temperature and salinity structure of the global ocean. Traditionally, systems minimize analysis temperature and salinity differences from observations using a least squares minimization technique. The cost function J in equation (1), an exemplary cost function of Fujii and Kamachi, A reconstruction of observed profiles in the sea east of Japan using vertical coupled temperature-salinity EOF modes, J. Oceanography, 59, 2003, 173-186, includes terms that minimize the temperature and salinity differences of the analysis from the first guess and observations. J = 1 2  ( x - x fg ) T  B - 1  ( x - x fg ) I + 1 2  ( H   x - x 0 ) T  R - 1