Title:On the accuracy with which the lower boundary conditions can be determined in numerical models of the atmosphere

Abstract: Since the prediction of climate is mainly considered as a prediction of second kind, it is indispensable to assess the accuracy with which these boundary conditions can be determined so that we can find a reasonable answer, whether climate is predictable with a sufficient degree of accuracy or not. Therefore, our contribution is mainly focused on the accuracy with which the fluxes of sensible heat and water vapor, required for predicting the lower boundary conditions in numerical models of the atmosphere using the coupled set of energy and water flux balance equations for the Earth's surface, can be determined. The parameterization schemes for the interfacial sublayer in the immediate vicinity of the Earth's surface and for the fully turbulent layer above presented and discussed here document that an appreciable degree of uncertainty exists. It is shown that the great uncertainty inherent in the universal functions of the Monin-Obukhov similarity laws on which the parameterization schemes for the fully turbulent atmospheric surface layer are based is reflected by the considerable scatter in the results of sophisticated field campaigns. This uncertainty affects also the results for the gradient-Richardson number, the turbulent Prandtl number, Pr_t, and the turbulent Schmidt number, Sc_{t,q}, (and the turbulent Lewis-Semenov number, LS_{t,q}) for water vapor used in such parameterization schemes. It is argued that the inherent uncertainty prevents that climate is predictable with a sufficient degree of accuracy.