SAUR: SEA LEVEL DIFFERENCES 



smoother eight successive times. The 50% level 



to 

 tr. 



UJ +10, 



2 

 LiJ 



I «u 



YEAR 



Figure 7. — Monthly sea levels (trend removed and ad- 

 justed to normal atmospheric pressure) . Hilo and Avila, 

 1947-59 and 1961-67. A. Unsmoothed anomalies (circles) 

 and smoothed anomalies (solid curve) for difference, 

 Hilo-minus-Avila. B. Smoothed anomalies for Hilo 

 (solid curve) and Avila (dashed curve). Shaded areas 

 show periods when anomaly of difference is negative. 



R(n) 



1 — sin^ 



77 



n 



where n is the number of data intervals in the 

 Fourier component being smoothed. In our case 

 n is also the wave period in months. Periods of 

 2 months are eliminated completely in one ap- 

 plication of the smoother. Repeated applica- 

 tions of the smoother eliminate or reduce pro- 

 gressively longer periods. The response is a 

 positive value between zero and one for all finite 

 wave periods, so no 180° phase shift nor amplifi- 

 cation occurs. The filter can be applied to the 

 individual station anomalies followed by compu- 

 tation of the diflference, or applied directly to the 

 time series of the difference anomalies with the 

 same end result. 



The smoothed anomalies of monthly sea level 

 for each station and of sea level differences in 

 Figures 6 and 7 were obtained by applying the 



for the response function lies between periods of 

 5 and 6 months, and the response is 85% at 8 

 months. 



CORRELATIONS BETWEEN 



HONOLULU-MINUS-SAN FRANCISCO AND 



HILO-MINUS-AVILA ANOMALIES 



There are no independent data against which 

 to check the anomalies of sea level differences 

 as an index of nonseasonal variations in current. 

 Since the two sets of differences span very nearly 

 the same current region, the consistency between 

 them is examined in this section to see if they 

 agree reasonably well. 



The correlations by month for both un- 

 smoothed and smoothed anomalies are listed in 

 Table 3. A general improvement of the corre- 

 lations, particularly in winter, occurs as a re- 

 sult of smoothing to suppress the short-period 

 variability. The effect of different scales of 

 wind systems can be recognized by the seasonal 

 pattern of correlations, especially in the 

 smoothed anomalies. In the months from April 

 through November when the subtropical high 

 dominates the atmospheric pressure pattern over 

 the North Pacific Ocean, all the correlations of 

 smoothed anomalies are significant at the 1% 

 level. In the months from November through 

 March when meso-scale cyclonic and anticyclonic 



Table 3. — Correlations of monthly sea level differences, 

 by month Hilo-minus-Avila versus Honolulu-minus-San 

 Francisco; n = 20 years. 



Month 



Unsmoothed 

 anomalies 



Smoothed 

 anomalies 



* Correlation significant at 5% level (r ^ 0.42). 

 ** Correlation significant at 1% level (r ^ 0.54). 



629 



