Abstract: In this paper, the causes of the inconsistencies between the initial data and the prognostic model, are discussed. One of the preliminary conclusions reached is that the inconsistencies are in many cases, primarily caused by the disagreement between the assumptions (or the treatments) made in the model and the observations taken as initial values. A scheme for initialization, constrained to total energy conservation and non-divergence of the whole air column, is proposed. Finally, a scheme for choice of certain parameters, subject to the global integral constrain is suggested.