This study investigates an advanced 4D-REKF (four-dimensional relaxation Kalman filter) data assimilation and high-resolution model developed at National Center for Atmospheric Research (NCAR) for the wind/power analysis and forecasting at wind farms. The 4D-REKF data assimilation system is a hybrid data assimilation system, which combines ensemble Kalman filter (EnKF) and NCAR RTFDDA (real-time four dimensional data assimilation system). Kalman gain can be calculated based on the members of ensemble runs as well as the ensembles of different weather regimes of model historical data when ensemble runs are not available. The research focused on the impact of model grid resolution, from mesoscale grid (DX= 1 – 10km) to LES-scale grid (DX=300m), and how to effectively and efficiently use the wind farm and all available surrounding weather measurements with the advanced data assimilation technique to improve the 0 – 12 h wind forecasts. Case studies are conducted by running the WRF model at high resolution (3km/0.9km) and at ultra high resolution (e.g. large eddy simulation scale) at a wind farm in the Northern Colorado and an off-shore wind farm in the southwest Korea peninsular to demonstrate the capabilities of the advanced data assimilation and the high-and ultra-high resolution WRF model in improving the wind farm wind prediction on the short time scales, which is essential in wind power grid integration.