SECKEL and YONG: SEA-SURFACE TEMPERATURES AND SALINITIES 



TABLE 1. — Mean monthly sea-surface temperature (°C), 

 Christmas Island: National Marine Fisheries Service station 

 (NMFS) and Hawaii Institute of Geophysics station (HIG). 



Date 



NMFS 



HIG 



1973 



1974 



May 



June 



July 



August 



September 



October 



November 



December 



January 



February 



March 



April 



May 



June 



Average 



24.1 



24.6 



in comparison with that at Koko Head indicates 

 another data problem. The scatter probably is 

 caused by sampling of water in the shallow beach 

 area that is more sensitive to changes in the local 

 heating-cooling processes than the deep water of 

 an offshore site. 



Finally, there are no systematically observed 

 sea-surface temperatures near Christmas Island 

 against which the shore measurements can be 

 calibrated. However, the monthly pamphlet Fish- 

 ing Information 3 contains a temperature chart for 

 the equatorial Pacific. Contours near Christmas 

 Island are based on insufficient observations to 

 reproduce the temperature distribution reliably. 

 Therefore, the values from these charts, plotted 

 on the annual graphs of Appendix D, show large 

 variations in the difference between the Fishing 

 Information temperatures and Christmas Island 

 observations. On average the Fishing Information 

 values are 1.3°C higher than the midmonth ex- 

 pected values with differences ranging from -1.2° 

 to 4.1°C. 



The discrepancy between the temperature sets, 

 in part, may be due to a tendency toward a warm 

 bias of merchant vessel temperature reports. 

 More probable, however, Christmas Island tem- 

 peratures, being measured in the morning, reflect 

 the effect of night cooling in shallow water that 

 would be in excess of the temperature decline 

 taking place in deeper, offshore water. 



Despite the apparent discrepancies between the 

 beach and offshore temperatures, the data from 

 the shore sampling sites reflect climatic scale 



Wishing Information. March 1970 through December 1973. 

 U.S. Dep. Commer., NOAA, Natl. Mar. Fish. Serv., Southwest 

 Fish. Cent., La Jolla, Calif. 



anomalies. For example, both the Christmas 

 Island record (Seckel and Yong 1971) and the 

 Canton Island record (Bjerknes 1969) show the 

 equatorial cold anomaly of 1955-56, the warm 

 anomaly of 1957-58, and the anomalous biannual 

 temperature variations of 1963-67. 



ANALYSES OF 

 LONG-TERM DATA RECORDS 



In this section we present harmonic analysis 

 results of the entire data series and separately 

 display the long-term variability, the short-term 

 variability, the average annual cycle, and the 

 variability that characterizes individual years. 



The entire data series is expressed by the 

 function 



Sl = A + 2^ C n cos oj (nt 



OCn 



(2) 



n=l 



where A gives the average value of the series, 

 k is the highest harmonic of the analysis, and 

 other symbols have the same meaning as given 

 for Equation (1). 



Input values for these analyses were calculated 

 at 15-day intervals from the annual analyses 

 presented in this and our previous paper (Seckel 

 and Yong 1971). Analysis of the 1956-73 Koko 

 Head data was carried to the 72d harmonic and 

 of the 1954-73 Christmas Island data to the 80th 

 harmonic so that the shortest period resolved in 

 each series is 3 mo. Analyses were carried out 

 on the residuals from a linear fit as in the analyses 

 of the annual data sets. The harmonic and linear 

 coefficients for the long-term series are listed in 

 Tables 1, 2, and 3 of Appendix F. 



The fitted curves resulting from these analyses 

 together with the input values are shown in 

 panels A of Figures 1, 2, and 3. Dominant in the 

 Koko Head temperature is the annual variation 

 without pronounced longer term trends other 

 than the rise of maximum and minimum tempera- 

 tures from 1966 to 1968. In contrast to the Koko 

 Head temperature curve, the salinity curve shows 

 longer term variations that are larger than the 

 seasonal variations. Also, during some years such 

 as in 1957, annual variation is not apparent. The 

 Christmas Island temperatures convey a similar 

 picture; interannual changes are larger than the 

 annual changes. Again, the latter may be ob- 

 scured or absent as during the years 1963-66 and 

 in 1973 when biannual changes dominated. 



769 



