ESTIMATION OF WIND SPEED DATA WITH SETAR MODEL

Abstract

The threshold model allows expression with different Autoregressive Moving Average (ARMA) models sorted according to the threshold value of the observations. In this study, nineteen years of observed wind speed data have been modeled with the Self Exciting Threshold Autoregressive (SETAR) model. Two different Autoregressive (AR(3)) models have been obtained for the situation where the wind speed was below and above 2.5 m / s of the previous observation in the time series. In addition, in the SETAR (1,3,3) model, it has been determined that the residual terms have the effect of GARCH (1,1) and a range has been estimated for model predictions.