This study consists in estimating long term weather characteristics and predicting extreme loads for a prospective offshore wind site located in the Puerto Rico archipelago. This location has been described in a recent study as the best possible area for wind energy generation in Puerto Rico. Regretfully, available field weather data for the target site is limited, which poses a challenge for insufficient for most statistical forecasting techniques. In order to overcome data scarcity in target location, two different approaches are used: (1) Measure-Correlate-Predict (MCP) methods, where long term data in nearby weather buoys are used to estimate long term wind characteristics in target site, and (2) Statistical extrapolation techniques, where distributions for the extreme mudline bending moment are established using parametric models as functions of wind speed and wave height in the target site to predict extreme loads, where fifty-years return loads are estimated. Finally, we discuss advantages and limitations using these techniques in target site based in data sets currently available.