Water level dynamics in the Amazon floodplain

Nonlinear analysis of Amazonian River systems suggests that water level variability is due to Brownian-like motions BH. Dynamical properties of the water level is probably a response of rainfall and runoff smoothing by the stored water in the floodplain; additional inputs of water results in attenuated perturbations of the water level. This implies that noise is a self-affine fractal with scale invariance, and in accordance with surrogate and correlation dimension analysis, which refused the hypothesis of low dimensional chaos for water level dynamics. Opposing to the chaotic approach, we present a fractal model formed by a periodic signal and a BH process that mimics original water level variability. Dynamical exponents are remarkably similar between real and modeled data when BH influence is about 1% the amplitude of the periodic signal. In comparison to natural systems, a hydroelectric reservoir produces different exponents, due to the control of the water level, as demonstrated for the Tucuruı́ dam in the Tocantins River. The Amazon floodplain at Óbidos and Curuai exhibits a fortnight cycle most probably due to bore tide influence in the river mouth.