Elimination of long-term variation from chaotic light curves

We performed a comparative dynamical investigation of chaotic test data using the global flow reconstruction method. We demonstrate that large-amplitude, long-term variations may have a disturbing effect in the analysis. The Empirical Mode Decomposition method (EMD) and the Fourier filtering were tested to remove the additional variations. Test results show that the elimination of these variations significantly increased the robustness of the reconstructions.