REGRESSION ANALYSIS OF SEA-SURFACE-TEMPERATURE PATTERNS FOR THE NORTH PACIFIC OCEAN.
Research and development rept. Feb 66-Jul 67,
NAVAL UNDERSEA WARFARE CENTER SAN DIEGO CALIF
Pagination or Media Count:
The classical method of summarizing large numbers of oceanographic data is to produce contour charts, by manual methods, indicative of the distribution of a physical characteristic. The report presents a statistical method of analysis in which a polynomial function of three variables latitude, longitude, and day-of-year is employed to represent the distribution of sea-surface temperature. For the mathematical model developed in this report, sea-surface temperatures from the North Pacific Ocean were selected as the source of data. A technique of stepwise regression analysis was used to determine the relationship of the variables involved. Computer-prepared contour charts resulting from this study showed a very close relationship to charts prepared by manual contouring. For instance, although there was more smoothing of the curves, major thermal features were readily discernible, and the latitudinal location of isotherms agreed with those produced by classical methods. This study demonstrates that a statistical model offers a potential tool to the oceanographer by providing a fast method of analyzing large numbers of data objectively. Author
- Physical and Dynamic Oceanography
- Statistics and Probability