temperature in the North Pacific and the American Geographical 

 Society is preparing similar charts for the Atlantic in the area of 

 the Gulf Stream. The technique used to summarize the data is 

 the same as that used in the earlier studies. Both studies sum- 

 marize data for a particular year by monthly time intervals. Tlie 

 Bureau of Commercial Fisheries uses 2-degree-square areas and 

 the Geographical Society uses 3 0-minute -square areas. 



This study examines the potential of multiple-regression 

 analysis as an approach to analyzing sea-surface temperature 

 observations. Although multiple-regression techniques were 

 developed by the statisticians many decades ago, they have rarely 

 been used by oceanographers , because of the complexity inherent 

 in developing realistic models and the magnitude of the arithmetic 

 task required to evaluate the necessary constants. With the rapid 

 progress in developing high-speed digital computers, the arith- 

 metic computational difficulties have been overcome to the extent 

 that it is now practical to consider relatively complex models. 



THE REGRESSION MODEL 



In the use of regression analysis it is necessary to know, or 

 to assume, (1) the major independent variables, or main effects; 

 and (2) a functional relationship between these variables, or a 

 regression model. In any given situation the desired functional 

 relationship is generally determined from analytical or theoretical 

 considerations or from a study of scatter diagrams prepared from 

 the data being analyzed. In this study the latter approach is used. 



12 



