Previsão de vazão na bacia hidrográfica do rio Manuel Alves da Natividade utilizando o modelo de séries temporais SARIMA

Abstract

Streamflow knowledge and modeling is fundamental for water resources management. Hydrologic models help on water supply planning and projects regarding water budget, hydropower generation, irrigation systems, sustainable use and biodiversity conservation. The study area is Manuel Alves da Natividade (BHMAN) watershed, located in the Tocantins-Araguaia river basin. The BHMAN is one of the main tributaries of the Tocantins river. In this context, the objective was to analyze, model and make predictions of the surface runoff based on the time series model SARIMA, with monthly step. The methodology adopted was that of Box and Jenkins which consists of: identifying the model; estimate their parameters and apply the adjusted model for forecasting. The streamflow series was verified for its trend, seasonality and stationarity. The identification of the component orders of the models was made through the graphical analysis of the correlograms and periodograms. The selection of the best model was performed based on the selection criteria and quality measures. The model chosen was the SARIMA(1,0,1)(1,1,4)12, which presented Nash-Sutcliffe coefficients of 0.61 and 0.81 for the calibration and validation steps, respectively, which are considered good according to the classification applied for conceptual hydrological models. The model proved good performance in predicting hydrography recession, being indicated mainly as a tool for water resources management, irrigation planning and water supply.