In this study mesoscale meteorological model and CFD model was used to downscale the numerical weather prediction (NWP) data and verified at Hachijo Jima wind power plant. To reduce the computational cost of mesoscale model, coefficient matrix method was proposed. Following results were obtained. The root mean square error (RMSE) was reduced to 3.1 m/s and 2.8m/s by mesoscale and CFD model respectively from 5.7m/s of original NWP. The coefficient matrix method reduced the computational time to a few seconds by one PC from two hours by parallel computers with 8CPUs without increasing the prediction error. KEYWORDS: Physical model, Wind speed Prediction, Wind energy forecasting, Mesoscale model, CFD 1 INTRODUCTION For electric power supply system, good agreement between demand and supply is essential. But in case of wind energy it is difficult to know the power output fluctuation in advance as it fluctuates with wind speed. This uncertainty causes problem in demand and supply planning for electric power and in real time electricity supply operation everyday. In countries with high wind energy penetration like Denmark and Germany, the day ahead forecast and the daily forecast of wind energy output are carried out based on the numerical weather prediction data and online measurement of the wind energy output))However, operational models use statistical methods somehow, which means the power prediction is based on the past experience on the relationship between the prediction and measurement. This approach is efficient when past measurement data is available. On the other hand, the needs for the wind power prediction exist at the beginning of the operation of a new wind farm or even before the construction of the wind farm. In such cases, physical approach is needed, in which wind speed is predicted by physical model without using any past measurement data.In this study, a physical model based wind speed prediction system was developed for online power prediction and verified by the anemometer located at the nacelle of the wind turbine at the Tokyo Electric Power Company Hachijo Jima wind power station.