A spatial and temporal stochastic cascade analysis of meteorological models and reanalyses.

摘要

This thesis investigates the hypothesis that the stochastic structure of deterministic models of the atmosphere is captured by multiplicative cascade processes. Using data from reanalyses (ERA 40) and two meteorological models (GFS, GEM), we investigate the spatial and temporal cascade structures of the temperature, humidity, and horizontal wind at various altitudes, latitudes, and forecast times. Over the range spanning from the model dissipation scales (..100 km) to at least 5000 km and for statistical moments up to order 2, the cascade predictions are satisfied to typically better than +/-1%. In time, we find corresponding cascade structures with outer scales of roughly 15 days. By constructing space-time diagrammes, we find they are roughly linear up to 5-10 days with transformation velocities of about 1000 km/day, as predicted based on the solar energy flux. This transition time scale, which corresponds to planetary size structures, objectively defines the weather/climate transition.