Extended version of detrended cross-correlation analysis

We describe an extended version of the detrended cross-correlation analysis which is useful for studying time-varying dynamics of complex systems producing inhomogeneous datasets. The method computes two independent measures that quantity the detrended covariance and the impact of nonstationarity, respectively. We apply this approach to characterize entrainment phenomena in the chaotic dynamics of two coupled Lorenz models with an additional trend in the datasets under study.