SYSTEMATIC VARIATION OF SEA-SURFACE 

 TEMPERATURE OYER PERIODS OF SEVERAL YEARS 



In connection with the question of whether or not sea-surface temperatures 

 vary significantly over periods of several years, four specific time periods are ex- 

 amined here in some detail: 



1. 1947 to 1956 which has been referred to as displaying "A uniform 

 monotony of conditions in at least the eastern North Pacific that is scarcely sug- 

 gested by any similar series of years in this century." 5 



2. 1936 to 1956 where there are indications of a cooling trend. 



3. 1957 and 1958 recognized as "the changing years." 5 



4. 1930 to 1935, another period, unique to this century, which appears to 

 contain long-term oscillations. 



In the previous discussion on trends by Van Vliet and Anderson, 3 a 

 variety of statistical considerations led to the general conclusion that no trend 

 existed in the records for any of the locations examined and that quantities such 

 as the annual average temperature (/3 ) , annual amplitude (a t ), annual phase WJ, 

 and percent variance explained by regression (100R 2 ) all behaved as independent 

 random variables are expected to behave. In addition it was pointed out that this 

 conclusion does not deny the existence of real year-to-year differences in the 

 ocean, but rather emphasizes that these differences are not unexpected from the 

 viewpoint of statistics and thus are not considered unusual or improbable events. 



First the period 1947 to 1956 is examined in the light of the statistical 

 parameters developed in this analysis. Figure 6 contains a plot of the average 

 annual temperatures (/3 ) for four eastern North Pacific locations. During this 

 decade, for 8 of the 10 years at all four stations, the /3 's were below the median 

 value and for 2 years were at the median or slightly above suggesting that on the 

 average the decade was cooler than normal. Figure 7 contains parameters that 

 determine the shape of the seasonal variation (a,, a 2 , i9i, 6 2 ). A study of these 

 four parameters does not suggest anything unusual about the shape of the seasonal 

 surface temperature variation during this decade. Figure 8 contains the percent 

 variance explained by regression and the standard deviation, both parameters 

 concerned with the variability. Again a study of these factors does no suggest 

 anything unusual in the amount or degree of variability. Thus, this analysis 

 suggests that the surface temperature variation during this decade was unusual as 

 compared with other decades observed in this century in that the average temper- 

 ature was slightly below the median value. 



Although too much after-the-fact analysis of data is contrary to statisti- 

 cal philosophy, sometimes it is of interest to do such analysis. Specifically, for 



18 



