Nonlinear and chaotic systems are a specific category of physical systems on the border between the structures of classical Newtonian physics and random processes. Atmosphere, klimatický systém i jejich dílčí komponenty přitom vykazují mnoho rysů deterministicky chaotického chování: Limitovanou deterministickou předpověditelnost (v případě počasí omezenou na několik dnů), citlivost na počáteční podmínky a rychlou (v průměru exponenciální) divergenci podobných stavů. These properties are key factors limiting the ability of forecasting and climate simulations realistically affect the behavior of the climate system and its components.
The research conducted at the Department of Meteorology characteristics of chaotic systems are studied both theoretically, and through low- and high dimensional numerical simulations of the atmosphere and statistical methods. Attention is paid to the identification of possible attractors in the phase space of the studied systems, odhadům charakteristických invariantů popisujících geometrii fázových trajektorií (např. prostřednictvím různých variant fraktální dimenze) i kvantifikaci předpověditelnosti (charakterizované Lyapunovovy exponenty či informační entropií). Acquired notes are used to analyze internal structures and internal links within both real and simulated physical systems.
At the intersection between statistical analysis and the study of chaotic systems are moving techniques nonlinear time series analysis. Appropriate methods, including advanced statistical algorithms based on artificial neural networks and reconstruction phase space, They are used for predictive purposes and the study of spatio-temporal relationships between variables in the climate system. Outputs nonlinear time series analysis are used for postprocessing outputs forecasting and climate models and also to verify the ability of these simulations reproduce the behavior of real physical systems.
