Evaluation of Two Numerical Weather Prediction Models for Hub-Height Wind Resource Parameters during Sudden High-Wind Events in North China

This study evaluates two approaches for estimating hub-height wind speeds within China’s operational NWP systems: the CMA_WSP model using similarity-theory and the WRF-FITCH framework with an embedded wind-farm parameterization. Focusing on a July 10–13, 2022 wind-speed fluctuation event at four Hebei wind farms, we compared forecasts against 15-min tower measurements at 65, 80 and 90 m heights over a 72-h window. Verification employed RMSE, correlation (R), accuracy rate (AR) and qualification rate (QR) metrics. Results show that WRF-FITCH reduces RMSE by approximately 30% (2.2 m/s vs. 3.2 m/s for CMA_WSP), raises R from 0.45 to 0.68, and improves AR and QR by 8.6 and 20.4 percentage points, respectively. Ramp-event analysis reveals that WRF-FITCH exhibits smaller ramp-rate biases. Spatial comparisons of wind-power density highlight an east–high, west–low gradient driven by terrain and model coordinates, with WRF_FITCH outperforming CMA_WSP by up to 120 W/m2 in the central plain. These differences stem from WRF_FITCH’s online coupling of turbine thrust and wake turbulence versus CMA_WSP’s surface-extrapolation approach. Our findings underscore the value of fully integrated boundary-layer schemes for improving hub-height wind and power forecasts under rapidly changing weather conditions.