A subjective study of these scatter diagrams suggests a 

 polynomial of the following form: 



T = a„ + n^L - a.L" - «„L 



(2) 



where L is the latitude, T' is the estimated sea-surface tempera- 

 ture, and the subscripted a's are the regression coefficients to 

 be estimated. 



Equation (2) was fitted to the above 24 sets of data. The 

 following statistics for the data shown in figure 2 were obtained: 







Number of 





100R-, 



c, 



Year 



Longitude 



Observations 



R 



percent 



degrees F 



MARCH 











1956 



126°W 



44 



0.95 



89.6 



1.5 





132° 



42 



0.96 



91.8 



1.8 





138° 



50 



0,97 



93.7 



2.1 



1957 



126°W 



38 



0.95 



90.0 



2.1 





132° 



105 



0.94 



89.0 



2.3 





138° 



65 



0.97 



94.4 



1.9 



26 



