In Romania, wind power has found a favorable environment, in both legislation as well as in the natural conditions, to evolve and meet the targets imposed by the European Union. As a consequence, in just five years, starting from the 1 st of January 2010, the installed wind power had an increase of almost 23410%, reaching today 2952.854 MW. This large value and the uncontrollable and stochastic character of wind increases the difficulty of power system operation. In order to help the operators, wind power forecasting systems are needed so that a close-to-reality planning of the power system can be made. One critical component of these systems is the upscaling algorithm used to extrapolate the wind power plant (WPP) level forecasts for a large area. To this end, some representative WPPs are needed. This paper is presenting the first steps taken in finding the representative WPPs in order to create an upscaling algorithm for the Romanian power system. The entire analysis is made on real production data recorded through the EMS/SCADA system that runs at the Romanian transmission system operator (TSO).