Previsão de velocidade do vento no nordeste brasileiro através do modelo ARIMAX

Abstract

The present paper uses the ARIMAX, which represents a kind of modeling of the Box-Jenkins meth-odology, to provide forecasts of monthly averages of wind-time time series in northeastern regions Brazil, at a level of 10 m in height with the use of free software R. As exogenous variables we opted for the use of atmos-pheric pressure and ambient temperature, which have physical characteristics that can be associated with the wind speed. The results of the error statistics show the efficacy of the model, with a percentage error close to 9%, comparing the observed and adjusted series. In ad- 1De sua denominação em inglês Intergovernmental Pa-nel on Climate Change. Maiores detalhes em sua página eletrônica < http://www.ipcc.ch/ >. 2 Consultado em (11/01/2017) na página eletrônica do IPCC em < https://www.ipcc.ch/ >. dition, it was possible to identify that the predicted se-ries can follow the profile of the observed series, through similarities of minimum and maximum values of the wind speed. This result may highlight the fact that the ARIMAX model proposed for the study regions can capture the existence of seasonality. This work could be useful as a possible tool to assist in decision making in the various areas of the wind sector, becoming another attempt of guarantees to reinforce the choice of a region in the exploration of its wind potential.