Computing functional estimators of spatiotemporal long-range dependence parameters in the spectral-wavelet domain

This paper addresses the problem of parameter estimation of spatiotemporal long-range dependence models from functional spectral data. Four wavelet-based functional estimation algorithms are proposed to approximate the multidimensional strong-dependence parameter, characterizing the covariance tail behavior of the spatiotemporal non-self-similar model class studied in Frías et al. (2006b, 2009). Wavelet regression is performed in all of them. Functional spectral data are averaged in the first and fourth algorithms, while, in the second and third ones, averaging is performed on the wavelet regression estimates. Smoothing over the wavelet translation parameter is performed, within each resolution level, only in Algorithms 3 and 4. A simulation study is carried out to illustrate the performance of the four functional estimation algorithms proposed under different scenarios.