Previsão de vazão no semi-árido nordestino, utilizando modelos atmosféricos: um estudo de caso

Abstract

The numerical dynamical climatological precipitation prediction for Northeast Region of Brazil is already a reality and its results have shown to help decision makers on their matters. Those predictions, which are based on sea surface temperature (SST) evolution, may be used as base for flow predictions. This work makes use of FUNCEMEs numerical climatological rain prediction and Central Atlantic/Pacific SST, to propose and test a set of six possibilities (models) to allow the prediction of flow at a small hydrological basin from Brazils Northeast Semi-Arid region. The basin name is Fazenda Cajazeiras and is located at the central northwest region of Ceará State – Brazil. The model set is composed by a set of combinations between single and multiple regressions from SST series and basin flow series, interchanged or not, with the ECHAM and RSM atmospheric models. The CN-3S hydrological model is also used as part of some of those possibilities. Two ways of improving the atmospheric simulated data are presented and tested. One of them is adopted. The model that has shown being best suited to predict Fazenda Cajazeiras flows is that one which uses the rain data set generated from the linear multiple regression form lagged SST, transformed into flow by the CN-3S model. The two models which used ECHAM and RSM results going through the CN-3S to get flow also have a high potential for flow prediction. A case study for the prediction of 2002 and 2003 flow is done and the prediction for 2005 flow is also presented.