SOUSA, G. M.; SOUSA, GLAUCIO DE MENESES; http://lattes.cnpq.br/6522667213939252; SOUSA, Glaucio de Meneses.
Resumo:
Mathematical modeling is one of the tools widely used to perform the water balance of hydrographic basins. In semi-arid regions, this methodology is essentially important to account for water resources and equalize their supply and demand, given the frequent occurrence of scarcity. This work proposed the application of the Multiple Linear Regression (MRLM) rainfall-runoff model to generate average monthly flows from monthly rainfall totals in the sub-basin of the Engenheiro Ávidos reservoir. To select the MRLM input and output variables, the “forward stepwise” method was used. The results obtained in this research showed that the MRLM did not adjust to the data from the sub-basin of the Engenheiro Ávidos reservoir, given that, for all combinations of variable inputs in the MRLM, low correlation rates always resulted between the average monthly flows observed and estimated. In time, it is suggested that in future research models such as Artificial Neural Networks or Regression by Principal Components Analysis are used.