POWER: MODEL OF NORTHERN ANCHOVY DRIFT 



MARCH 





Figure 2.— Resultant mean current vectors for the normal January and the March seasonal current data used in this study. See text 

 for cautions concerning figure interpretation. Length of arrow indicating north direction corresponds to a current velocity of 10 cm/s. 



producing offshore directed Ekman transport. Two 

 current fields were obtained by increasing the cross- 

 shore component of the mean March Ekman 

 velocities by the factors 1.5 and 3.0, and then com- 

 bining the April seasonal geostrophic and aug- 

 mented March Ekman velocities. Wind stress, and 

 hence transport, is proportional to the square of 

 wind speed. This means that roughly a 22% increase 

 in a downshore wind speed increases the corres- 

 ponding offshore directed Ekman transport by the 

 factor 1.5. A threefold increase in offshore Ekman 

 transport results from about a 77% increase in the 

 downshore wind speed. Bakun and Nelson (1976) 

 presented extensive statistical analyses of an 

 "up welling index" (defined as the offshore directed 

 component of Ekman transport) for the location lat. 

 33°N, long. 119°W (this point is very close to loca- 

 tion A used in the simulations; see below). Over an 

 annual cycle the mean upwelling index for this loca- 

 tion changes by at least a factor of two, with a rapid 

 increase in both mean and standard deviation dur- 



ing the spring. The March mean index at this point 

 was about 50 t/s per 100 m of coastline with a stan- 

 dard deviation of roughly 80, hence upwelling at this 

 particular time and location can be highly variable. 

 Further, Bakun and Nelson (1976) found that en- 

 hanced or diminished upwelling persists on a 

 seasonal time scale, so incorporation in the model 

 of prolonged increased Ekman transport was not 

 unrealistic. 



Diffusion was incorporated into the model solely 

 to parameterize subgrid scale mixing; including 

 larger scale and more ephemeral mixing processes 

 would obscure the broad seasonal trends the model 

 was intended to illustrate. The eddy diffusivity 

 parameter was computed using scale-dependent dif- 

 fusion formulae of Okubo (1976) and a regression 

 analysis of diffusion data presented by Okubo (1971). 

 The finite-difference representation of diffusion re- 

 quired the use of a pseudo-Fickian diffusivity coeffi- 

 cient, so the mean scale-dependent diffusivity for 

 the 37 km grid spacing (K x = K y = 101 m 2 /s) was 



589 



