In this study, deterministic predictability of the vertical mean state (i.e., barotropic component) of the atmosphere is examined by conducting perfect-twin model experiments. First, we evaluated the barotropic-baroclinic interactions from the observations. Given the interactions as the external forcing of the barotropic model, the present model accurately follows the trajectory of the real atmosphere for three months. We find that the predictability limit is inversely proportional to the logarithm of the initial error energy level. Specifically, the predictability increases 10 days when the initial error energy is reduced to 1/10. The result implies in theory that the predictability is unbounded when the initial error approached to zero. Since the model predicts the vertical mean quantities, the local observational error may be reduced in proportion to the square root of the total number of the data samples. The result suggests that the vertical mean state may be predictable beyond two weeks if we can have a perfect model with sufficient number of observations in the vertical. Next, the perfect-twin model experiments are conducted with the parameterized baroclinic instability for the external forcing. We demonstrated that the present model indicates a predictability somewhere around 35 days according to the anomaly correlation within a model atmosphere. Hence, the result suggests that the predicting vertical mean of the atmosphere can be one of the viable approach to the medium-range weather prediction, provided that the barotropic-baroclinic interactions are accurately parameterized.
Close window or click on previous window to return to the Conference Program.
12th Conference on Atmospheric and Oceanic Fluid Dynamics