channel passes in the breakwater. Even with only a week of data, Site 8 shows 

 a trend to the northwest, perhaps because of sheltering and eddy effects from 

 the nearby island (Plate 23). 



93. The skewness observed in the rose plots are the result of statisti- 

 cal representation of cumulative events and do not imply a flow occurring at 

 any one time. To see the instantaneous currents, time series representations 

 are required, as shown in the next section. 



Flow field details 



94. To observe details of the harbor flow patterns and to illustrate 

 the correlations between sites at simultaneous points in time, currents must 

 be observed at a smaller scale. Two nested windows were selected for detailed 

 observation: 7 and 8 August and 0000 to 0400 hr on 8 August. This coincided 

 with the profiling of Range 5 and one of the Lagrangian, or drogue, tracking 

 experiments. At these time scales, it is also convenient to combine the velo- 

 city and direction information into a single vector time series plot. Both 

 types of plots were generated, and each has advantages (Appendixes F, G, 



and H) . 



95. To substantiate correlations between currents occurring simultan- 

 eously at separate locations, the current vectors should be examined to ensure 

 that they are reasonable and actually in phase. Since the currents are pri- 

 marily driven by the potential energy of the elevation difference existing at 

 different places in the harbor at any instant, the residuals between selected 

 tide gages should indicate expected current vectors. 



96. When seen at this scale, the residuals display higher frequency 

 oscillations as well as obvious diurnal harmonics. These could be due to ran- 

 dom errors in the pressure sensor signal, phase shifts in the clocks of dif- 

 ferent instruments, and long-period (30 to 60 min) wave energy in some unde- 

 termined combination. A filtering scheme could remove selected components, 

 but care should be taken in assuming that the actual residuals correspond to a 

 preconceived smooth curve or evidence of higher order oscillations could be 

 obscured. 



97. The tidal elevation differentials can be seen in residuals TGI - 

 TG3 (Figure 12). Slopes approach ~2 x 10" 5 , and comparisons with the 2-day 

 tidal plot (Figure 11) illustrate that rising, or inflowing, tides coincide 

 with positive residuals, falling tides with negative, and residuals near zero 

 occur near high and low water. 



34 



