2026-10-12T19:53:31Z https://www.scidb.cn/oai

10.57760/sciencedb.35840 2026-05-07T09:58:09Z

2026-05-07 Dataset and Pre-trained Models for A global hourly surface temperature and wind dataset derived from deep learning temporal downscaling doi:10.57760/sciencedb.35840 en This dataset provides 1-hourly global surface meteorological information (T2M, U10, V10, and MSL) generated by the WeatherAttRes-ConvLSTM (WAR-C) framework. The WAR-C model utilizes a teacher-student architecture, leveraging 6-hour forecasts from the FuXi model as prior knowledge and learning from high-resolution ERA5 reanalysis data.models.zip.pthT2M: 2-meter temperatureU10: 10-meter U-wind componentV10: 10-meter V-wind componentMSL: Mean sea level pressurebest_convlstm.pthcheckpoint.pth Global surface meteorological dataset; Temporal downscaling; Deep learning; WeatherAttRes-ConvLSTM (WAR-C); ERA5; Hourly weather forecast. Xinyao Wang Junhong Guo Chenchen Liu PUBLIC https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank