ABSTRACT

The motivation behind the regime-switching approach is to deal with nonstationarity in wind dynamics—in wind speed, in wind direction, or in spatial correlation. One particular class of wind forecasting methods worth special attention is the regime-switching approach. In essence, a regime-switching method is a collection of distinct, and most often linear, models. The regime-switching mechanism can be used with a temporal only process, considering only nonstationarity in time, or with a spatio-temporal process, considering nonstationarity in both space and time. In a regime-switching method, as well as in the methods introduced in the sequel, one crucial question is how to decide the number of wind regimes and the boundaries dividing these regimes. The regime-switching autoregressive model and the regime-switching spacetime method can be perceived as a “reactive” approach. Plainly speaking, a reactive model observes a regime change or a manifestation of it, and then adapts itself accordingly to accommodate it.