Impact of the recorded variable on recurrence quantification analysis of flows

Recurrence quantification analysis (RQA) is useful in analyzing dynamical systems from a time series s(t). This paper investigates the robustness of RQA in detecting different dynamical regimes with respect to the recorded variable s(t). RQA was applied to time series x(t), y(t) and z(t) of a drifting Rössler system, which are known to have different observability properties. It was found that some characteristics estimated via RQA are heavily influenced by the choice of s(t) in the case of flows but not in the case of maps.