NAEFS (North American Ensemble Forecasting System) has been maintained and continuously enhanced since its first operational implementation in 2005 at NCEP. A statistical down-scaling method was applied to the NAEFS in December 2007 to generate 5-km resolution (the same as National Digital Guidance Database - NDGD) products for CONUS from 1.0 degree bias corrected NCEP and CMC's ensemble forecasts. There are four variables (2-meter temperature, surface pressure, 10-meter u and v) for this CONUS application in current operation. Now this down-scaling method has been implemented to Alaska region in 2010. In addition to the current CONUS four variables, there are 4 new important and high demand surface variables: wind speed, wind direction, maximum temperature and minimum temperature have been added. The downscaled products will include probabilistic forecasts (10%, 50%, and 90%), ensemble mean, spread and mode, four times per day out to 16 days. This presentation will mainly focus on the downscaling of wind and wind speed. A probabilistic distribution of wind direction will be discussed as well.