CONCLUSIONS 



1. A regression model considering latitude, longitude, and 

 day-of-year as the independent variables together with empirically 

 determined interaction terms, was found capable of estimating 

 the seasonal variation of sea-surface temperatures off the west 

 coast of the United States, in water depths greater than 100 

 fathoms, to one standard deviation of something less than 1°F. 



2. From both a statistical and a physical viewpoint rela- 

 tively simple regression models have a considerable potential as 

 estimators of seasonal and spatial variation in sea-surface tem- 

 perature. 



3. The analysis suggests that more information than pre- 

 viously suspected can be obtained from a given number of observa- 

 tions provided realistic regression models can be developed. 



This has important implications with regard to sampling. A 

 sampling interval based on the model can be used in place of the 

 fixed time interval employed in the classical manner with an area 

 grid. The oceanographic problem becomes one of searching for 

 adequate models. It is indicated such models can be derived for 

 many ocean areas from the present archive of oceanic tempera- 

 ture data. 



4. On the assumption that the regression model is reason- 

 ably valid, the regression technique has the potential of being 



an effective, and objective, method for identifying and editing 

 raw temperature data for erroneous observations. 



5. When used to remove the seasonal and spatial variation 

 in a set of sea-surface temperature data, a regression model, 

 such as discussed in this study, may be used to detect and isolate 

 temperature anomalies. 



6. Finally, this study suggests regression techniques may 

 be used as a new approach to summarizing archived sea-sui-face 

 temperature data that is more objective and amenable to computer 

 usage than presently used methods. 



REVERSE SroE BLANK 79 



