*3.2. Data Validation*

The need for data providing information on long-term wave conditions over extensive areas makes it necessary to use alternative sources of information to that provided by measuring instruments, which are generally scarce, if any, and cover short periods of time. In this sense, the joint use of wind and wave numerical prediction models to obtain longterm databases on a spatial grid is currently a very common and useful tool. However, due to various reasons, wind and wave prediction models have some limitations in their ability to reproduce real conditions, mainly in areas with complex topography, such as islands (e.g., [19,20]). Consequently, whenever possible, it is necessary to validate the results against other sources of information, preferably from specific experimental measurement devices, such as meteoceanic buoys. Data validation has been achieved in this study by comparing data derived from models at two nodes, located at positions coinciding with that of BGN and BTS wave buoys (see Figure 2b), and data measured by these buoys, considering only the periods for which measured data were available.

Due to the interest in extreme wave storms and considering the difficulty of models to correctly reproduce extreme wave conditions [21,22], characteristic parameters associated with each selected wave storm were validated by linear regression and graphically represented for visually assessing the degree of agreement between simulations and experimental recordings. Moreover, joint validation of circular (wind and wave direction) and linear variables (wind speed, wave height and period) has been done by examining wind and wave roses for measured and simulated data.
