Deep Learning for Daily 10-m Wind Downscaling

This study employs deep learning method U-Net to downscale 10-m Wind in North China, utilizing reforecast data from the National Centers for Environmental Prediction Global Ensemble Forecasting System and the fifth generation of global climate reanalysis data from the European Centre for Medium-Range Weather Forecasts. We evaluate and analyze the methods through comprehensive evaluation metrics. Downscaling of the 10-m wind field in North China is performed using the U-Net method compared with linear regression. The results indicate that the accuracy of wind speed prediction with U-Net surpasses that of linear regression, while for wind direction, U-Net performs similarly to linear regression. Overall, this paper explores deep learning technique for downscaling, which can provide valuable insights for operational departments and research efforts.