Downscaling the wind energy resource on a complex terrain with a forest canopy using an atmospheric CFD code

Experience feedback on wind resource assessment for complex terrain shows that current operational methods lack precision, especially on complex terrain. The wind energy industry needs new tools like atmospheric CFD, remote wind sensors, data assimilation to better evaluate wind resource. Our study aims both to compare an atmospheric CFD code to a data set obtained in a very complex terrain and to investigate different methodological options for its application to wind resource purposes. In this paper we present the first step of this work. A field campaign of wind and turbulence measurements has been initiated in 2007 for about 1 year on a complex and forested terrain. Simulations are performed with the code Mercure-Saturne developed by CEREA. First simulations using coupling with meso-scale model Aladin have shown a globally good agreement with mast measurements. However, large differences between Aladin output and site measurements influence the quality of this agreement as the micro-scale model cannot correct the forecast errors of the meso-scale model. With currently available computing power, the need of long time period simulations lead to select some situations which have to be representative of the local wind characteristics. Thus, two clustering methods are investigated, one supervised by the user, and one unsupervised. Those two methods are now going to be compared on a one year Aladin data set. For each one, the annual wind speed distribution and wind direction rose will be calculated for the 4 instrumented locations and compared to the measurements and to WAsP simulations. Further investigations will concern especially improvement of the method used to take into account the vertical stability and correction of meso-scale output using site measurements.