There are great difficulties in properly evaluating the power present in the waves in the Caribbean due to of the scarcity of marine instrumentation and the reduced length of the existing records. This research aims to design a new methodology for estimating wave power potential in places lacking instrumentation by using reanalysis winds and wave generation models to generate hourly synthetic wave information (maps and wave series), which is later compared and corrected with nearby buoy measurements. Nested runs of the models permit to downscale the results and to create detailed wave power maps, used to identify the best sites for a wave farm. These maps are built from propagation of characteristic sea-states of the synthetic series, and are chosen using joint probability, power percentiles and clustering algorithms. New series are generated in the chosen places and they are processed to determine the wave power potential in different time scales.