Noise Suppression in a Primitive Equation Prediction Model

Abstract

A large proportion of the noise in the National Meteorological Center 8-layer global model is due to a high-frequency external and a low-frequency internal gravity-inertia wave mode. Divergence diffusion effectively controls these two noise components, but it has undesirable side effects when used in this global model. However, divergence diffusion can be used to essentially damp either the external or the internal gravity-inertia wave mode individually. Damping the former controls the noise as well as a frequently used time filter, but has an unacceptable side effect in the vicinity of mountains. Damping the latter provides better noise control than the time filter and has no undesirable side effects.