Temporal and Spatial Downscaling of Wind Forecast of New Energy Stations Based on an Optimal Frequency Bias Algorithm

Accurate wind speed prediction through numerical weather prediction (NWP) can reduce the operating cost of wind farms. To satisfy the demand of wind forecasts of high temporal and spatial resolution for power prediction at wind power stations in Shanxi province, surface wind field forecasts at the new energy stations from the global forecast system (GFS) model are temporally and spatially downscaled from September 1, 2021, to August 31, 2022, in this paper. Results showed that the performance of the downscaled forecast product based on the optimal frequency bias (OFB) algorithm is better than that of the original model forecast for different initial times, lead times, ground levels, and wind speed grades. The threat score (TS) of surface wind speed forecast fluctuates periodically with forecast lead time. In general, 14 PM corresponds to the peak of the TS time series, while 02 AM corresponds to the trough. The TS improvement of the downscaled products initialized at 00 UTC is greater than that initialized at 12 UTC. Furthermore, the improvement is more obvious for wind speed levels that are closer to the ground.