https://www.sciencedirect.com/science/article/pii/S0960148118314782
The WRF forecast model was used to downscale GFS driving data with domains (shown below) going from 9km to 3km to 1km (two-way nested with 34 levels) and an inner domain of 333m resolution (one-way nested with 67 levels) around a windfarm in Galicia (NW Spain). The MYNN2.5 PBL scheme was used and a year of simulations were produced (with and without the Fitch wind turbine wake parametrization scheme).
Forecasts were compared to both nacelle wind speed measurements and wind power output for 31 different turbines in this particular wind farm. The validation indicated that the output from the 333m domain did offer enhanced prediction compared to the coarser domain but the Fitch scheme did not offer any improvement over the forecasts run without the scheme.
Plots of the difference between the two sets of simulations (such as above) highlight the wind speed losses produced due to turbine wakes by the Fitch scheme indicating that it is indeed having an impact at curbing the wind speeds, even though it is not producing a better forecast.