Previsão climática sazonal para a região hidrográfica dos sertões de Crateús utilizando regressão linear múltipla

Abstract

Water scarcity is one of the major problems to be faced in the Sertão of the Brazilian Northeast. This region is characterized by low rainfall rates and a concentration of precipitation in only a few months of the year. Besides this, another factor that interferes negatively in the availability of water in the region is the lack of planning and adequate allocation of water resources. Precipitation forecasting has been shown to be an interesting alternative to instruct the public authorities on how to make decisions that mitigate the problems related to water availability. Thus, this work aims to generate precipitation forecast models for the Crateús Hinterlands Hydrographic Region, using Traditional, Ridge and Lasso regression. The model predictors were the Sea Surface Temperature (SST) indices, since the precipitation of the Northeast as a whole is greatly influenced by the oceans. The Interative Input Selection (IIS) algorithm was used to rank these indices, and from this ranking all models were generated and their accuracy was analyzed. Furthermore, to perform all regressions and the ranking of the TSM indices, the R-Studio program was used to generate the necessary routines. From the results generated, it was possible to evaluate the accuracy of differen t precipitation forecast models, identifying those that presented the highest correlations.