FISHERY BULLETIN: VOL. 75, NO. 4 



cr 



LLl 



a. 



5 



u 



1956 1957 1958 1959 I960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 



FIGURE 1. — Koko Head temperature, 1956-73: A. Fitted curve with a 3-mo resolution (n = 1-72). B. Long-term variation (n = 1-17). 

 C. Short-term variation (n = 18-72). D. Residual variation (n = 19-35, 37-53, 55-71). 



The amplitudes (C„) of the long-term analyses 

 (Figure 4) confirm these qualitative impressions. 

 In the Koko Head temperature, the amplitude 

 of the annual sinusoid (18th harmonic) is dom- 

 inant and almost six times as large as the largest 

 amplitude of the long periods. In the Koko Head 

 salinity and Christmas Island temperature, on 

 the other hand, long periods have the largest 

 amplitudes. For the Koko Head salinity the 

 amplitude of the fourth harmonic is larger than 

 that of the annual sinusoid and for the Christmas 

 Island temperature the amplitude of the first 

 harmonic is almost twice that of the annual 

 sinusoid. 



Long-Term Changes 



When long-term changes are of interest, the 

 annual and shorter term variability can be 

 filtered by a variety of methods including the 

 commonly used 12-mo moving average method. 

 After harmonic analysis has been used as a curve- 

 fitting technique, however, it is simple to evaluate 

 only the terms in the harmonic function up to 

 but not including the annual sinusoid in order 

 to display long-term changes. Thus, in Equation 

 (2), the Koko Head temperatures and salinities 

 were evaluated for n = 1 to 17 and the Christmas 

 Island temperatures for n =1 to 19. The resulting 



770 



