Accurate wind prediction over complex coastal terrain is essential for mitigating the stochasticity and uncertainty of wind energy in power systems. This study focuses on day-ahead hourly wind speed forecasting for the coastal region around meteorological station no. 45032 in Hong Kong and develops a multiscale prediction methodology that combines a mesoscale Weather Research and Forecasting (WRF) model with a microscale computational fluid dynamics (CFD) model. Based on high-resolution WRF forecasts, the coupling framework dynamically assigns CFD open boundary conditions across multiple grid nodes and directions and incorporates momentum source terms and a relaxation zone in the CFD model to address inconsistencies arising from dynamical downscaling and turbulence parameterization. Results show that high-resolution WRF reduces the day-ahead hourly mean absolute percentage error (MAPE) from 37.2% to 22.8%, with the coupled WRF–CFD model further lowering the MAPE to 14.8%. Furthermore, the coupled WRF–CFD approach accurately captures terrain-induced wind variations, including the speed-up effect over ridges and the sheltering effect in valleys. The proposed forecasting framework achieves both high predictive accuracy and good timeliness, providing a practical and generalizable solution for refined wind resource assessment and wind farm planning in complex terrain.
Xiao et al. (Thu,) studied this question.