Redes Neurais Artificiais Aplicadas à Previsão Antecipada de Precipitações na Região Central de Manaus

Abstract

This works aims at showing the results of artificial neural networks for forecasting rainfall 30 min ahead in the central area of Manaus from data of meteorological stations located around the city. A total of 523 neural networks were proposed and trained for the problem under consideration where two of them showed better acuracy and precision. A second training with a balanced proportion of the existing classes resulted in an increase in the precision of the tests. The results obtained can be used by several agencies for strategic decision making aiming at minimizing the negative effects of rainfall in this city.

Publication
In: IV Escola Regional de Informática Norte I. Manaus, Amazonas.
Date