Generally historical climate datasets are not sufficient to cover
the entire lifecycle of offshore wind farms. Although the data may
cover the past 20–25 years, it is rare that the climate data will
present exactly the same track in the following years. On the other
hand, it is important to generate datasets that preserve characteristics
of the original dataset. The generation of different climate
datasets minimises the uncertainty of the simulation results. If a
single dataset is employed in the simulations, the risk of experiencing
rougher climate conditions may be ignored. Similarly,
experiencing lower wind speed values in the future may create
risk on the power production values. Therefore, wind speed, wave
height, and wave period historical time series are modelled; and
the developed climate model is employed to generate wind speed,
wave height, and wave period time series data at the beginning of
each simulation. In this context, the modelling approach adopted
in this work is a correlated Multivariate Auto-Regressive (MAR)
approach. The general form for an AR model, normalised by the
mean of the data is described in Eq. (2).