Supporting Data for "Aircraft Icing Intensity Prediction Integrating Cloud-Resolving Model Simulations and an Interpretable Heterogeneous Ensemble of Machine Learning Models

The processed datasets used in the article include the WRF model output (wrfout) files from the 3-km inner-domain simulations over Southwest China (100°–111°E, 22°–33°N) , with 49 vertical levels and hourly output. These wrfout files contain the three-dimensional meteorological and cloud microphysical fields used to extract the nine environmental predictors for aircraft-icing intensity classification, including temperature, relative humidity, dew-point temperature, water-vapor and cloud-water mixing ratios, horizontal wind speed, 700-hPa temperature, and freezing-level height. ERA5 hourly reanalysis data at 0.25° resolution were used as initial and lateral boundary conditions for the WRF simulations. The original Pilot Icing Reports (PIREPs) are not included in this dataset due to data-sharing restrictions.Due to the large volume of wrfout data and slow upload speeds, the remaining wrfout files will be uploaded in a V2 version.