Portal de Periódicos da CAPES

Modelagem e Previsão de Vazão Afluente Média Mensal no Rio Tocantins, Usina Hidrelétrica Tucuruí-Pará, Amazônia, Brasil

2016; UNIVERSIDADE FEDERAL DO AMAPÁ; Volume: 6; Issue: 2 Linguagem: Inglês

10.18561/2179-5746/biotaamazonia.v6n2p9-16

2179-5746

P.G.P. Santos, Terezinha Ferreira de Oliveira,

Geography and Environmental Studies

The hydroelectric power stations located in the Amazon are of extreme importance for a series of issues (environmental, social, economic) due to their impacts. This article describes and models the temporal series of monthly mean streamflow using stochastic models from SARIMA (seasonal autoregressive integrated moving average). The data was collected from august 2002 to august 2012 in Tucurui Hydroelectric Power Station, Para, Eastern Amazon, Brazil. Several SARIMA models were tested to describe the data. Akaike information criterion and goodness fit tests were used to choose the most parsimonious model. The chosen model to the streamflow series was SARIMA (0,0,2) (1,2,2) 12 , which adjusted to the observed series with Nash-Sutchiffe (CNS) coefficient 0.9150. This method, built with the observed data only, had great modeling and forecasting success. Keywords: Hidrology; stochastic models; seasonality; SARIMA.