| With more wind energy being integrated into our grid systems, forecasting wind energy has become a necessity for all market participants. Recognizing the market demands, a physical approach to site-specific hub-height wind speed forecasting system has been developed. This system is driven by the outputs from the Canadian Global Environmental Multiscale (GEM) model. A simple interpolation approach benchmarks the forecasting accuracy inherited from GEM. Local, site specific winds are affected on a local scale by a variety of factors including representation of the land surface and local boundary-layer process over heterogeneous terrain which have been a continuing challenge in NWP models like GEM with typical horizontal resolution of order 15-km. In order to resolve these small scale effects, a wind energy industry standard model, WAsP, is coupled with GEM to improve the forecast. Coupling the WAsP model with GEM improves the overall forecasts, but remains unsatisfactory for forecasting winds with abrupt surface condition changes. Subsequently in this study, a new coupler that uses a 2-D RANS model of boundary-layer flow over surface condition changes with improved physics has been developed to further improve the forecasts when winds coming from a water surface to land experience abrupt changes in surface conditions. It has been demonstrated that using vertically averaged wind speeds to represent geostrophic winds for input into the micro-scale models could reduce forecast errors. The hub-height wind speed forecasts could be further improved using a linear MOS approach. The forecasting system has been evaluated, using a wind energy standard evaluation matrix, against data from an 80-m mast located near the north shore of Lake Erie. Coupling with GEM-LAM and a power conversion model using a theoretical power curve have also been investigated. For hub-height wind speeds GEM appears to perform better with a 15-Ian grid than the high resolution GEM-2.5Ian version at the validation site. |
