We investigate the effect of rotation on the structure of turbulence i
n neutral and convective mixed layers and assess the applicability of
the Reynolds stress turbulence closure models to studies of deep conve
ction. Unlike in the case of stable stratification, where rotational c
ontribution is limited (Galperin et al., 1989), in neutral and convect
ive flows this contribution is significant. The rotational terms endow
eddy viscosity and eddy diffusivity with tensorial properties and alg
ebraic complexity. In the present study we thoroughly investigate the
mathematical features of this tensorial eddy viscosity-eddy diffusivit
y formulation. To ensure physical realizability of the model, we found
it necessary to impose upper bounds on the dimensionless velocity and
temperature gradients in the expressions for eddy coefficients. It wa
s also found absolutely crucial to include in the model a limitation o
n the magnitude of the turbulence macroscale due to background rotatio
n. This limitation is similar to the one imposed by stable stratificat
ion. The self-similar solutions derived from the model are compared wi
th the experimental data by Fernando et al. (1991) and are shown to be
in good agreement. Implications of the results for oceanographic mode
ling and for simulation of deep convection are discussed.