Dynamical and Statistical Modeling of the Multiscale

Abstract. Earth’s magnetosphere exhibits both global and multiscale behavior and the extensive database from spacecraft and ground-based measurements has been used to develop dynamical models. The dominant feature of the magnetosphere is its large scale or global behavior, as is evident during substorms. The dynamical features can be modeled using the techniques of phase space reconstruction and this has led to near real-time forecasts of space weather. The multiscale nature limits the predictability of such dynamical models and this is quantified in terms of conditional probabilities. The combination of the dynamical and probabilistic descriptions provides a comprehensive approach to the modeling of Earth’s magnetosphere. The spatial structure of the magnetosphere is modeled using distributed data from ground magnetometer stations and can be used to make local forecasts