Wednesday, 9 August 2000
Bulk methods for estimating air-sea fluxes from mean meteorological parameters are widely used when direct flux measurements are not available, such as for the estimation of near-surface EM-EO propagation characteristics (evaporation duct, refractivity, scintillation, etc.), for example. Bulk methods are also used to parameterize air-sea coupling and boundary conditions in mesoscale prediction models. Knowledge of the sources and nature of the errors associated with bulk models is essential to assess the validity of predictions based on the bulk flux estimates. In this study these sources of error are identified and are discussed in terms of 'sensitivity coefficients' which describe how dependent the bulk flux estimate is to variations in a model input parameter. The sensitivity of bulk air-sea flux estimates (u*, T* and q*) to each model input parameter is determined and illustrated for different stability and Bowen ratio conditions. In certain conditions, especially with low wind speed and positive air-sea temperature differences, bulk models often fail to converge to a solution, and the model results are highly sensitive to the input parameters. In such conditions the dimensionless profile functions used in the model for stable conditions can dramatically alter the final model results and the sensitivity and convergence characteristics of the model. In this study the convergence and sensitivity characteristics of different commonly used dimensionless profile functions are demonstrated. Specifically, the wind speed and air-sea temperature difference regimes in which the models converge and fail to converge are determined.
![](images/flag.png)
![](images/blue_ribbon.png)