This paper discusses the accuracy and characteristics of offshore wind speeds simulated by the Weather Research and Forecasting (WRF) model. Firstly, the accuracy of the simulated wind speed is examined using in-situ measurements from the Shirahama offshore research platform for the whole year of 2005. It is found that the WRF surface wind speed has an annual bias of 15.3%, which is much larger than the National Centers for Environmental Prediction Final Analysis data (NCEP FNL) used as input. Secondly, the possibility of accuracy improvement of the simulated wind speed is discussed in terms of the decomposition of RMSE and the estimation error of wind energy density. As a result, it is shown that reducing the large positive bias is a key to increase the accuracy of the WRF wind speed. Without the large bias, the error in the estimation of annual mean wind energy density drastically decreases from 47% to 4%. Finally, the accuracy of wind speeds above the surface layer is examined using wind profiler measurements from Mihama. It is found that the WRF wind speed has large positive biases not only near the sea surface but also in the lower PBL. Thus, the large positive biases are speculated to be mainly due to inaccuracy of the PBL scheme in WRF.
2011 by the Meteorological Society of Japan