Previsão de vazão em uma bacia do semiárido usando previsões climáticas numéricas de precipitação.

Abstract

Numerical precipitation climatic forecasts, resulting from the nesting of a regional model in a general circulation model of the atmosphere, are used to predict outflows in the Piancó river basin. The region is located in the semi-arid region of Paraíba, in the north-east of the Northeast, presenting great climatic variability and proven seasonal predictability. The precipitation forecast was generated from observed SST values, in a set of ten simulations of different initial conditions. due to the identification of systematic errors, a correction method was applied to the predicted values, based on the averages and deviations of the observed climatologies and the model. Daily precipitation is used as input to a daily rainfall-flow model. Then, the expected precipitations are accumulated monthly and corrected, being used as input for a concentrated rain-flow model, on a monthly scale. The results show that the forecast together provides better adjustments to the observed precipitation. The aggregation of predicted daily rainfall on a monthly scale did not result in better correlations between forecast and observation. Considering the seasonal scale, the flows were correctly predicted in 65.5% of the analyzed years, when categorized (low, medium or high).