and speed of the computed currents, a stream function ( ip ) analysis is 

 made using methods similar to those employed by Bedient and Vederman 

 (1964) to represent atmospheric flow in the tropics. 



The stream function field which corresponds to the ocean current 

 chart in Figure 6 (06 GMT 1 February 1965) is shown in Figure 7. Current 

 vectors have been plotted at selected grid points to show the degree of 

 fit. The fact that the derived stream function is nondivergent while there 

 is divergence in the initial velocity field explains some of the cross 

 contour flow. In general, however, this appears to be small in most 

 places, and the stream field provides a good representation of the current 

 pattern. 



The surface current forecasts are verified with sea surface temperature 

 analyses and heat exchange and advection computations , Detailed pro- 

 cedures are described by Hubert (1964 and 1965). Besides the direct use 

 of this product in various naval operations, it is used for computation 

 of divergence/convergence for thermocline depth and subsurface thermal 

 structure forecasts and for forecasting the advective portion of sea surface 

 temperature change. 



Figure 8 is a further breakdown of the scheme discussed earlier. As 

 can be seen, all of the computational programs described in the preceding 

 section enter into the determination of thermal structure with depth. The 

 analyses will be built downward from the surface (where the most reports 

 are available) . 



The previous day's analysis will first be modified by mechanical and 

 convective mixing, where applicable, and large-scale convergence and 

 divergence in the surface layers . Climatological (hydroclime) restraints 

 will be used to keep computed changes within reasonable limits. Finally, 

 BT observations will be introduced by means of a median seeking technique 

 such as used in the SST analysis program. 



Most of the components needed to assemble the complete subsurface 

 analysis and prediction package have been programmed for numerical solu- 

 tion. The first major portion completed is a hemispheric analysis and 

 prediction of mixed layer depth, which is shown in Figure 9. A partly 

 subjective, partly computerized thermal structure and sound speed profile 

 forecasting is now being done daily for twelve points in the Pacific. 

 Examples of these forecasts at three points, together with verifying obser- 

 vation and climatology, are shown in Figure 10. 



It should be noted that heraispheric oceanographic analyses and 

 predictions have certain limitations. They do, e.g., not give enough 

 details of complex conditions within about 100 miles of major current 

 boundaries, nor near coasts and islands. These charts also do not 

 present the truly short-term fluctuations (12 hours or less) caused, e.g., 

 by tides . Special explanatory notes on these short-term and small-scale 

 processes have been issued by FNWF to the users of the analyses and 



95 



