A simple and efficient approach to the initialization of weather prediction models

An iterative method is suggested, by which the gravity modes of a primitive shallow water equation model are damped very efficiently while the meteorological modes are preserved fairly well. The model equations are integrated in a forward/backward cycle. It is shown that information about the normal modes of the model can be used to increase the damping rate even if these modes are not known explicitly. The meteorological modes can be conserved by keeping the relevant terms constant during each iteration cycle. Experiments with a simple one-dimensional model show that this model reaches a balanced state after a couple of iterations.