it to any more conventional physical measurements. It is also not clear how 

 one might separate vertical and horizontal components of mixing using the 

 blocks. 



An additional indirect estimate of mixing in the microcosms and in the field 

 was obtained by measuring the diffusion coefficient for oxygen across the 

 air-water interface. The measurements were made using a small floating plastic 

 dome from which virtually all of the oxygen was displaced by nitrogen gas. The 

 partial pressure of oxygen in the dome and in the water was then monitored 

 over time and the tlux of oxygen from the water into the dome calculated. 

 Since the flux is a product of the gradient in partial pressure and the diffusion 

 coefficient, it was possible to obtain the coefficient from such a data set. 

 Measurements with domes containing turbulent or still air have confirmed that 

 the diffusion coefficient is largely a function of turbulence in the liquid phase, 

 and that the effect of wind is felt through its influence on water mixing. 



RESULTS 



Turbulence Levels Obtained 



The results of the various turbulence measurements lead us to be 

 particularly sympathetic to Okubo's lament that "diffusion is confusion." In 

 spite of, or perhaps because of the fact that a large number of floating pair 

 observations were made in Narragansett Bay and in the microcosms, there was 

 a very large amount of scatter in these data. As a result, the calculation of 

 neighbor diffusivity (F) and energy flux (e) was subject to a large uncertainty 

 and there is some question about how meaningful the numbers may be. While 

 the values of F tended to decrease approximately according to the 4/3 law 

 (Richardson 1926; Stommel 1948), it also appeared that e had a tendency to 

 fall off with size. The latter result is disturbing since the theoretical framework 

 for the computation suggests that e should be constant at steadystate in the 

 inertial range between the size at which energy is put into the system and the 

 viscous zone in which it is dissipated. 



Functional regressions (Ricker 1973) relating F to scale are given below. In 

 general, both the neighbor diffusivity and the energy flux indicated that the 

 turbulence levels in the microcosms with one paddle were appreciably higher 

 than found in the bay or in the mocrocosms with one half paddle and no 

 paddle (Table 25-1). The F Values did not show a great difference between the 

 half paddle microcosms and the bay and the very large variance associated with 

 the calculation of e in the half paddle tanks made it difficult to see any clear 

 differences in terms of energy flux. 



392 



