Numerical experiments with alternative boundary layer formulations using bomex data

The basic numerical air-sea boundary-layer model described in Pandolfo (1969a, b) was varied to produce a set of models with differing atmospheric boundary-layer formulas, four of which are discussed here. Model I is the basic model itself, with stability and sea-state dependent eddy viscosity, conductivity and diffusivity which may, in certain ranges ofRi, be unequal. This model is applied on a relatively fine grid. Model II, applied on the same grid, uses formulas which yield equal eddy conductivity, diffusivity, and viscosity. The calculated eddy coefficients depend only on the height and wind shear. Model III uses the same exchange coefficient formulas as Model II. However, the surface-layer eddy flux in Model III is calculated by assuming that logarithmic profiles of the transported variables are present in this layer. Model IV is the same as Model III in these respects, but employs a relatively coarse vertical grid. This model, therefore, includes boundary layer formulas most like those conventionally used in large scale atmospheric models (e.g. Miyakoda, 1969).The four models were integrated numerically with identical inputs of initial, boundary, and auxiliary data prepared from observations made over the eastern half of the BOMEX observational area during June 21–25, 1969.Models I and IV are, in general, in better agreement with each other than either is with Model II. This is true for the model-generated upper and lower boundary fluxes of mean momentum and latent heat; and for the internal boundary layer production of mean kinetic energy by the cross-isobaric flow component. Model I agrees, on balance, about as well with Model IV as does Model III. The solutions for Models I, III, and IV are also, in general, more consistent with observed data, viz. 5-day average temperature profiles in the layer from the surface to 1000 meters, and 5-day averages of sea surface temperature and of surface-layer atmospheric humidity. Solutions for Model I are in better overall agreement with the observed data, and with the average observed surface-layer wind.The results show that, under the limitations implicit in these preliminary experiments, accurate simulations of observed data are possible with boundary-layer formulas of the type used in Model IV, and even more accurate simulation with the modest refinements represented by Model I. Piecemeal imposition of such refinements could, however, lead to models, like Model II, with significantly different energetic properties and less simulative accuracy. Specifically, the results support the speculation (Miyakodaet al., 1969) that the shallowness of the simulated Trades noted in some large-scale models is due to deficiencies in the boundary-layer eddy stress formulations used.