The variability of wind power production poses the greatest challenge in the integration of large-scale wind power in power systems. An analysis of the characteristics of the Croatian wind power resource has been carried out, based on observed wind speed data from recording sites for the period of one year. Also, a methodology to simulate realistic synthetic wind speed data at different locations is developed. The simulation methodology requires only basic statistical properties which describe characteristics of wind speed data, primary: Weibull scale and shape parameters, correlation and autocorrelation coefficients. The proposed methodology gives possibility of modeling the stochastic dependence between wind power time series in conjunction with individual persistence of wind speed data at each location. A comparison between measured wind time series from recording sites and simulated wind speed data series with similar parameters is conducted.