The increasing share of electricity generation from wind power is accompanying by installations of wind farms at new and suitable locations. Analyzing the local wind conditions often requires time- and cost-intensive measurements. Therefore, methods are developed to pre-assess the quality of new wind farm locations. The aim of this thesis is to use MERRA reanalysis data in a wind power simulation model in order to assess the validity of the MERRA data by comparison with real production data. The wind power production of 4 wind farms (1 in Austria and 3 in New Zealand) are compared with the simulation results. At each wind farm, the wind speeds (of the point in the MERRA dataset closest to the wind farm) are extrapolated to the hub height of the turbines by using the empirical derived “power law”. The simulation of the production uses the power curves of the installed turbines (of each wind farm), which represent the relation between wind speed and electricity generation. Several temporal resolutions from hourly to annual are analyzed and compared between the 4 wind farms. Results show that the simulation model does over- and underestimate the total production depending on the location. Correlation coefficients for hourly production are between 0.67 and 0.75 and increase between 0.74 and 0.85 for daily production. In general, low production events are overestimated and high production events are underestimated by the simulation model, although there are exceptions for a production close to rated or zero power. This is a consequence of the low spatial resolution and the utilized smoothed elevation model in MERRA. This could be overcome by the use of an empirical derived optimization model. Hence, the MERRA data has limited suitability for the simulation of single wind farm locations, but it could be useful for the simulation of larger spatial extents.