A mesoscale to microscale model chain that relates large scale wind climatologies with site conditions, for the prediction of wind resources and siting parameters, is under development within the New European Wind Atlas (NEWA) project. A statistical-dynamic downscaling method is proposed here based on the blending of long- term statistics of mesoscale forcings, extracted from the WRF mesoscale model, with site conditions simulated with unsteady k-ε Raynolds-Averaged Navier Stokes (URANS). The dynamic coupling of WRF and CFDWind has been developed using the GABLS3 diurnal cycle benchmark in flat terrain. The coupling is done offline by first running the WRF model to obtain time series of mesoscale tendencies (pressure gradient and advection terms) at a horizontal resolution of 3 km. Then, these tendencies are averaged temporally using a 1-hr rolling mean and spatially over a 3x3 grid at the site of interest to filter out small-scale forcings that will be explicitly modelled at microscale. These tendencies are reduced to a cycle of input forcings for the microscale model to simulate prevailing wind conditions. The methodology is tested at the Cabauw 100-m mast in flat terrain.