Publikace na Matematicko-fyzikální fakulta |
2016
Abstrakt
The paper focuses on divergence of nearby trajectories for different data. Reconstruction of phase space via delayed coordinates is applied for the estimate of divergence by the Rosenstein method, which is then compared to its alternatives in the form of two analytical estimates.
Results are obtained for two simulated series (originating from the Lorenz system or representing a random signal) and for a single midlatitude temperature series (45N 0E) at the 500 hPa level of ERA reanalysis. It is shown that the analytical estimates fit for random and temperature data and Lorenz data in state called irrelevance.
However, they fit poorly for Lorenz data with correct time delay.