Gradient Spectral Analysis for Short Time Series of Hydro-Ecological Seasonal Variation in the Pantanal Wetland, Brazil
The Pantanal wetland is one of the largest wetlands in the world (ca. 140.000 km2). Most of the time series collected from this natural system result in partial data set, specially water quality, thus compromising the performance of usual statistical analysis. The main goal of this research is to apply a new computational methodology for short time series analysis, showing non-linear behaviour in the time, amplitude and frequency domains to understand the hydro-ecological functioning of this river-floodplain system. The applied methodology, called Gradient Spectral Analysis (GSA), combines two mathematical techniques, the so-called Gradient Pattern Analysis (GPA) and the Wavelet Multiresolution Analysis (WMA). The GSA classifies different non-linear regimes taking into account short time series samples generated from dynamical processes previously associated with chaotic and stochastic models. We classified short times series (240 events) related to long-term hydro-ecological research of the Paraguay River, the main river of the Pantanal floodplain. From preliminary results, we correlate the seasonal variation of the river level records with the variation of 25 parameters of water quality, sampled monthly during 20 years (1989-2009). There is a good indication that this analysis will be robust enough for the prediction of behaviour variables and further application on climate change forecasting and correlation with primary and secondary aquatic production.
