Um modelo estocástico de previsão da precipitação média em uma bacia hidrográfica.

Abstract

The development of a stochastic model for the prediction of the mean basin precipitation on the day following a certain know n precipitation is attempted on the present study utilizing a Markov process. The events have been considered in pairs - the present state and the subsequent state - and distributed in a symmetrical transition matrix in such a way that, all the different classes (or class intervals) have an adequate representation of the events considered. By fitting a probability distribution for each of the lines of the matrix, a stochastic forecasting model is obtained for the prediction of the precipitation in the immediate next period for a chosen level of the probability of exceedance and a known level of the precipitation at present. This model was developed for the Jaguaribe River basin, and three of its sub basins. The Distribution Gamma with two parameters fitted quite well in all cases, thus demonstrating the regional homogeneity. With the intent to demonstrate an application of the results obtained with the rainfall prediction model in the generation of basin discharges, a non linear second order rainfall-runoff was also developed utilizing two observed short period unit hydrographs for two sub basins of the river Jaguaribe. The results show that the precipitation runoff model is consistent and the generated flows satisfy the prediction needs.