Using several data sources for offshore wind resource assessment.

The knowledge of wind resource is essential to the sitting of a wind farm. This is crucial especially in coastal areas where installation of windmills is more costly than onshore. More accurate estimation of the resource is necessary to preserve the confidence of investors and reduce the cost of production. Nowadays, the resource is evaluated by interpolation of discrete measurements but offshore measurements are rare and very costly (750,000 euros for an offshore mast) [1]. In addition, coastal wind distribution is characterised by large spatial variations, which are not represented by local measurements. Several studies demonstrated the potentials of using satellite images to map the wind field at sea surface. However, [3] and [4] show the limitations of the use of only one source of data. These limitations are caused by either low spatial resolution of data or localisation far away from coast or too low numbers of measurements for a given site. One solution is to merge al! available data to get a more accurate estimation of wind resource [4]. Satellite data have different spatial resolutions, localisations and temporal sampling. These differences impose constraints to the fusion method. In order to work out a data fusion method and evaluate it, it is necessary to study these constraints and their impacts. This paper gives an overview of various available offshore wind data. They are obtained by an appropriate processing of observations of the sea surface performed by spaceborne radars. We study the characteristics of the various data sets, namely, spatial and temporal coverage, spatial and temporal sampling, data support and accuracy. Finally, we discuss the design of a fusion algorithm