Wind direction forecasting plays an important role in wind power prediction and air pollution management. Weather quantities such as temperature, precipitation, and wind speed are linear variables in which traditional model output statistics and bias correction methods are applied. However, wind direction is an angular variable; therefore, such traditional methods are ineffective for its evaluation. This paper proposes an effective bias correction technique for wind direction forecasting of turbine height from numerical weather prediction models, which is based on a circular-circular regression approach. The technique is applied to a 24-h forecast of 65-m wind directions observed at Yangmeishan wind farm, Yunnan Province, China, which consistently yields improvements in forecast performance parameters such as smaller absolute mean error and stronger similarity in wind rose diagram pattern. Wind direction forecasting plays an important role in wind power prediction and air pollution management. However, wind direction is an angular role in wind power prediction and air pollution management. variable; therefore, such traditional methods are ineffective for its evaluation. This technique is applied to wind direction forecasting of turbine height from numerical weather prediction models. a 24-h forecast of 65-m wind directions observed at Yangmeishan wind farm, Yunnan Province, China, which consistently yield improvements in forecast performance parameters such as smaller absolute mean error and stronger similarity in wind rose diagram pattern.